{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.8909327640762443, "global_step": 257000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 1.9974766972995618e-05, "loss": 6.2381, "step": 500 }, { "epoch": 0.02, "learning_rate": 1.994953394599123e-05, "loss": 4.9832, "step": 1000 }, { "epoch": 0.02, "learning_rate": 1.9924300918986846e-05, "loss": 4.5021, "step": 1500 }, { "epoch": 0.03, "learning_rate": 1.989906789198246e-05, "loss": 4.0662, "step": 2000 }, { "epoch": 0.04, "learning_rate": 1.9873834864978074e-05, "loss": 3.7652, "step": 2500 }, { "epoch": 0.05, "learning_rate": 1.984860183797369e-05, "loss": 3.5294, "step": 3000 }, { "epoch": 0.05, "learning_rate": 1.9823368810969303e-05, "loss": 3.3246, "step": 3500 }, { "epoch": 0.06, "learning_rate": 1.979813578396492e-05, "loss": 3.1381, "step": 4000 }, { "epoch": 0.07, "learning_rate": 1.977290275696053e-05, "loss": 3.0398, "step": 4500 }, { "epoch": 0.08, "learning_rate": 1.9747669729956147e-05, "loss": 2.9391, "step": 5000 }, { "epoch": 0.08, "learning_rate": 1.9722436702951763e-05, "loss": 2.8603, "step": 5500 }, { "epoch": 0.09, "learning_rate": 1.9697203675947376e-05, "loss": 2.7789, "step": 6000 }, { "epoch": 0.1, "learning_rate": 1.9671970648942992e-05, "loss": 2.6988, "step": 6500 }, { "epoch": 0.11, "learning_rate": 1.9646737621938604e-05, "loss": 2.6212, "step": 7000 }, { "epoch": 0.11, "learning_rate": 1.962150459493422e-05, "loss": 2.6162, "step": 7500 }, { "epoch": 0.12, "learning_rate": 1.9596271567929833e-05, "loss": 2.5455, "step": 8000 }, { "epoch": 0.13, "learning_rate": 1.957103854092545e-05, "loss": 2.5131, "step": 8500 }, { "epoch": 0.14, "learning_rate": 1.9545805513921065e-05, "loss": 2.4976, "step": 9000 }, { "epoch": 0.14, "learning_rate": 1.9520572486916677e-05, "loss": 2.4328, "step": 9500 }, { "epoch": 0.15, "learning_rate": 1.9495339459912293e-05, "loss": 2.3948, "step": 10000 }, { "epoch": 0.16, "learning_rate": 1.9470106432907906e-05, "loss": 2.3677, "step": 10500 }, { "epoch": 0.17, "learning_rate": 1.944487340590352e-05, "loss": 2.3427, "step": 11000 }, { "epoch": 0.17, "learning_rate": 1.9419640378899137e-05, "loss": 2.3043, "step": 11500 }, { "epoch": 0.18, "learning_rate": 1.939440735189475e-05, "loss": 2.2655, "step": 12000 }, { "epoch": 0.19, "learning_rate": 1.9369174324890363e-05, "loss": 2.279, "step": 12500 }, { "epoch": 0.2, "learning_rate": 1.934394129788598e-05, "loss": 2.233, "step": 13000 }, { "epoch": 0.2, "learning_rate": 1.931870827088159e-05, "loss": 2.2257, "step": 13500 }, { "epoch": 0.21, "learning_rate": 1.9293475243877207e-05, "loss": 2.187, "step": 14000 }, { "epoch": 0.22, "learning_rate": 1.9268242216872823e-05, "loss": 2.1743, "step": 14500 }, { "epoch": 0.23, "learning_rate": 1.9243009189868435e-05, "loss": 2.1596, "step": 15000 }, { "epoch": 0.23, "learning_rate": 1.921777616286405e-05, "loss": 2.1595, "step": 15500 }, { "epoch": 0.24, "learning_rate": 1.9192543135859664e-05, "loss": 2.1252, "step": 16000 }, { "epoch": 0.25, "learning_rate": 1.916731010885528e-05, "loss": 2.1394, "step": 16500 }, { "epoch": 0.26, "learning_rate": 1.9142077081850892e-05, "loss": 2.1028, "step": 17000 }, { "epoch": 0.26, "learning_rate": 1.9116844054846508e-05, "loss": 2.0825, "step": 17500 }, { "epoch": 0.27, "learning_rate": 1.9091611027842124e-05, "loss": 2.0636, "step": 18000 }, { "epoch": 0.28, "learning_rate": 1.9066378000837737e-05, "loss": 2.034, "step": 18500 }, { "epoch": 0.29, "learning_rate": 1.9041144973833353e-05, "loss": 2.0458, "step": 19000 }, { "epoch": 0.3, "learning_rate": 1.9015911946828965e-05, "loss": 2.0003, "step": 19500 }, { "epoch": 0.3, "learning_rate": 1.899067891982458e-05, "loss": 2.0075, "step": 20000 }, { "epoch": 0.31, "learning_rate": 1.8965445892820197e-05, "loss": 1.9797, "step": 20500 }, { "epoch": 0.32, "learning_rate": 1.894021286581581e-05, "loss": 1.992, "step": 21000 }, { "epoch": 0.33, "learning_rate": 1.8914979838811425e-05, "loss": 2.0055, "step": 21500 }, { "epoch": 0.33, "learning_rate": 1.8889746811807038e-05, "loss": 1.9899, "step": 22000 }, { "epoch": 0.34, "learning_rate": 1.8864513784802654e-05, "loss": 1.9286, "step": 22500 }, { "epoch": 0.35, "learning_rate": 1.883928075779827e-05, "loss": 1.9526, "step": 23000 }, { "epoch": 0.36, "learning_rate": 1.8814047730793882e-05, "loss": 1.9664, "step": 23500 }, { "epoch": 0.36, "learning_rate": 1.87888147037895e-05, "loss": 1.9361, "step": 24000 }, { "epoch": 0.37, "learning_rate": 1.876358167678511e-05, "loss": 1.9081, "step": 24500 }, { "epoch": 0.38, "learning_rate": 1.8738348649780727e-05, "loss": 1.946, "step": 25000 }, { "epoch": 0.39, "learning_rate": 1.8713115622776343e-05, "loss": 1.9021, "step": 25500 }, { "epoch": 0.39, "learning_rate": 1.8687882595771955e-05, "loss": 1.9108, "step": 26000 }, { "epoch": 0.4, "learning_rate": 1.866264956876757e-05, "loss": 1.8936, "step": 26500 }, { "epoch": 0.41, "learning_rate": 1.8637416541763184e-05, "loss": 1.8365, "step": 27000 }, { "epoch": 0.42, "learning_rate": 1.86121835147588e-05, "loss": 1.8602, "step": 27500 }, { "epoch": 0.42, "learning_rate": 1.8586950487754412e-05, "loss": 1.8453, "step": 28000 }, { "epoch": 0.43, "learning_rate": 1.8561717460750028e-05, "loss": 1.8968, "step": 28500 }, { "epoch": 0.44, "learning_rate": 1.8536484433745644e-05, "loss": 1.8502, "step": 29000 }, { "epoch": 0.45, "learning_rate": 1.8511251406741257e-05, "loss": 1.8216, "step": 29500 }, { "epoch": 0.45, "learning_rate": 1.8486018379736872e-05, "loss": 1.8147, "step": 30000 }, { "epoch": 0.46, "learning_rate": 1.8460785352732485e-05, "loss": 1.8327, "step": 30500 }, { "epoch": 0.47, "learning_rate": 1.84355523257281e-05, "loss": 1.8269, "step": 31000 }, { "epoch": 0.48, "learning_rate": 1.8410319298723717e-05, "loss": 1.7861, "step": 31500 }, { "epoch": 0.48, "learning_rate": 1.838508627171933e-05, "loss": 1.7778, "step": 32000 }, { "epoch": 0.49, "learning_rate": 1.8359853244714945e-05, "loss": 1.7793, "step": 32500 }, { "epoch": 0.5, "learning_rate": 1.8334620217710558e-05, "loss": 1.8129, "step": 33000 }, { "epoch": 0.51, "learning_rate": 1.8309387190706174e-05, "loss": 1.8, "step": 33500 }, { "epoch": 0.51, "learning_rate": 1.828415416370179e-05, "loss": 1.7884, "step": 34000 }, { "epoch": 0.52, "learning_rate": 1.8258921136697402e-05, "loss": 1.7957, "step": 34500 }, { "epoch": 0.53, "learning_rate": 1.8233688109693018e-05, "loss": 1.7527, "step": 35000 }, { "epoch": 0.54, "learning_rate": 1.820845508268863e-05, "loss": 1.7802, "step": 35500 }, { "epoch": 0.55, "learning_rate": 1.8183222055684247e-05, "loss": 1.7547, "step": 36000 }, { "epoch": 0.55, "learning_rate": 1.8157989028679863e-05, "loss": 1.757, "step": 36500 }, { "epoch": 0.56, "learning_rate": 1.8132756001675475e-05, "loss": 1.7334, "step": 37000 }, { "epoch": 0.57, "learning_rate": 1.8107522974671088e-05, "loss": 1.7337, "step": 37500 }, { "epoch": 0.58, "learning_rate": 1.8082289947666704e-05, "loss": 1.7287, "step": 38000 }, { "epoch": 0.58, "learning_rate": 1.8057056920662316e-05, "loss": 1.7256, "step": 38500 }, { "epoch": 0.59, "learning_rate": 1.8031823893657932e-05, "loss": 1.7266, "step": 39000 }, { "epoch": 0.6, "learning_rate": 1.8006590866653545e-05, "loss": 1.7145, "step": 39500 }, { "epoch": 0.61, "learning_rate": 1.798135783964916e-05, "loss": 1.7111, "step": 40000 }, { "epoch": 0.61, "learning_rate": 1.7956124812644776e-05, "loss": 1.7262, "step": 40500 }, { "epoch": 0.62, "learning_rate": 1.793089178564039e-05, "loss": 1.6894, "step": 41000 }, { "epoch": 0.63, "learning_rate": 1.7905658758636005e-05, "loss": 1.7264, "step": 41500 }, { "epoch": 0.64, "learning_rate": 1.7880425731631617e-05, "loss": 1.6879, "step": 42000 }, { "epoch": 0.64, "learning_rate": 1.7855192704627233e-05, "loss": 1.7057, "step": 42500 }, { "epoch": 0.65, "learning_rate": 1.782995967762285e-05, "loss": 1.678, "step": 43000 }, { "epoch": 0.66, "learning_rate": 1.7804726650618462e-05, "loss": 1.6776, "step": 43500 }, { "epoch": 0.67, "learning_rate": 1.7779493623614078e-05, "loss": 1.6706, "step": 44000 }, { "epoch": 0.67, "learning_rate": 1.775426059660969e-05, "loss": 1.6841, "step": 44500 }, { "epoch": 0.68, "learning_rate": 1.7729027569605306e-05, "loss": 1.6577, "step": 45000 }, { "epoch": 0.69, "learning_rate": 1.7703794542600922e-05, "loss": 1.6711, "step": 45500 }, { "epoch": 0.7, "learning_rate": 1.7678561515596535e-05, "loss": 1.667, "step": 46000 }, { "epoch": 0.7, "learning_rate": 1.765332848859215e-05, "loss": 1.6545, "step": 46500 }, { "epoch": 0.71, "learning_rate": 1.7628095461587763e-05, "loss": 1.6689, "step": 47000 }, { "epoch": 0.72, "learning_rate": 1.760286243458338e-05, "loss": 1.6267, "step": 47500 }, { "epoch": 0.73, "learning_rate": 1.757762940757899e-05, "loss": 1.6582, "step": 48000 }, { "epoch": 0.73, "learning_rate": 1.7552396380574608e-05, "loss": 1.6262, "step": 48500 }, { "epoch": 0.74, "learning_rate": 1.7527163353570223e-05, "loss": 1.6412, "step": 49000 }, { "epoch": 0.75, "learning_rate": 1.7501930326565836e-05, "loss": 1.6321, "step": 49500 }, { "epoch": 0.76, "learning_rate": 1.7476697299561452e-05, "loss": 1.6204, "step": 50000 }, { "epoch": 0.76, "learning_rate": 1.7451464272557064e-05, "loss": 1.636, "step": 50500 }, { "epoch": 0.77, "learning_rate": 1.742623124555268e-05, "loss": 1.6452, "step": 51000 }, { "epoch": 0.78, "learning_rate": 1.7400998218548296e-05, "loss": 1.6317, "step": 51500 }, { "epoch": 0.79, "learning_rate": 1.737576519154391e-05, "loss": 1.5938, "step": 52000 }, { "epoch": 0.79, "learning_rate": 1.7350532164539525e-05, "loss": 1.6206, "step": 52500 }, { "epoch": 0.8, "learning_rate": 1.7325299137535137e-05, "loss": 1.6148, "step": 53000 }, { "epoch": 0.81, "learning_rate": 1.7300066110530753e-05, "loss": 1.6055, "step": 53500 }, { "epoch": 0.82, "learning_rate": 1.727483308352637e-05, "loss": 1.5982, "step": 54000 }, { "epoch": 0.83, "learning_rate": 1.7249600056521982e-05, "loss": 1.5766, "step": 54500 }, { "epoch": 0.83, "learning_rate": 1.7224367029517598e-05, "loss": 1.6206, "step": 55000 }, { "epoch": 0.84, "learning_rate": 1.719913400251321e-05, "loss": 1.5962, "step": 55500 }, { "epoch": 0.85, "learning_rate": 1.7173900975508826e-05, "loss": 1.5649, "step": 56000 }, { "epoch": 0.86, "learning_rate": 1.7148667948504442e-05, "loss": 1.6079, "step": 56500 }, { "epoch": 0.86, "learning_rate": 1.7123434921500055e-05, "loss": 1.5598, "step": 57000 }, { "epoch": 0.87, "learning_rate": 1.709820189449567e-05, "loss": 1.5903, "step": 57500 }, { "epoch": 0.88, "learning_rate": 1.7072968867491283e-05, "loss": 1.5817, "step": 58000 }, { "epoch": 0.89, "learning_rate": 1.70477358404869e-05, "loss": 1.5581, "step": 58500 }, { "epoch": 0.89, "learning_rate": 1.7022502813482515e-05, "loss": 1.5733, "step": 59000 }, { "epoch": 0.9, "learning_rate": 1.6997269786478127e-05, "loss": 1.5958, "step": 59500 }, { "epoch": 0.91, "learning_rate": 1.6972036759473743e-05, "loss": 1.5571, "step": 60000 }, { "epoch": 0.92, "learning_rate": 1.6946803732469356e-05, "loss": 1.5557, "step": 60500 }, { "epoch": 0.92, "learning_rate": 1.6921570705464972e-05, "loss": 1.5613, "step": 61000 }, { "epoch": 0.93, "learning_rate": 1.6896337678460584e-05, "loss": 1.5778, "step": 61500 }, { "epoch": 0.94, "learning_rate": 1.68711046514562e-05, "loss": 1.5501, "step": 62000 }, { "epoch": 0.95, "learning_rate": 1.6845871624451813e-05, "loss": 1.5356, "step": 62500 }, { "epoch": 0.95, "learning_rate": 1.682063859744743e-05, "loss": 1.5369, "step": 63000 }, { "epoch": 0.96, "learning_rate": 1.679540557044304e-05, "loss": 1.546, "step": 63500 }, { "epoch": 0.97, "learning_rate": 1.6770172543438657e-05, "loss": 1.5583, "step": 64000 }, { "epoch": 0.98, "learning_rate": 1.674493951643427e-05, "loss": 1.5268, "step": 64500 }, { "epoch": 0.98, "learning_rate": 1.6719706489429886e-05, "loss": 1.5218, "step": 65000 }, { "epoch": 0.99, "learning_rate": 1.66944734624255e-05, "loss": 1.5108, "step": 65500 }, { "epoch": 1.0, "learning_rate": 1.6669240435421114e-05, "loss": 1.5397, "step": 66000 }, { "epoch": 1.01, "learning_rate": 1.664400740841673e-05, "loss": 1.523, "step": 66500 }, { "epoch": 1.01, "learning_rate": 1.6618774381412343e-05, "loss": 1.5161, "step": 67000 }, { "epoch": 1.02, "learning_rate": 1.659354135440796e-05, "loss": 1.5167, "step": 67500 }, { "epoch": 1.03, "learning_rate": 1.6568308327403574e-05, "loss": 1.5351, "step": 68000 }, { "epoch": 1.04, "learning_rate": 1.6543075300399187e-05, "loss": 1.5123, "step": 68500 }, { "epoch": 1.04, "learning_rate": 1.6517842273394803e-05, "loss": 1.5138, "step": 69000 }, { "epoch": 1.05, "learning_rate": 1.6492609246390415e-05, "loss": 1.5198, "step": 69500 }, { "epoch": 1.06, "learning_rate": 1.646737621938603e-05, "loss": 1.5005, "step": 70000 }, { "epoch": 1.07, "learning_rate": 1.6442143192381644e-05, "loss": 1.5152, "step": 70500 }, { "epoch": 1.07, "learning_rate": 1.641691016537726e-05, "loss": 1.5064, "step": 71000 }, { "epoch": 1.08, "learning_rate": 1.6391677138372876e-05, "loss": 1.5044, "step": 71500 }, { "epoch": 1.09, "learning_rate": 1.6366444111368488e-05, "loss": 1.5009, "step": 72000 }, { "epoch": 1.1, "learning_rate": 1.6341211084364104e-05, "loss": 1.5121, "step": 72500 }, { "epoch": 1.11, "learning_rate": 1.6315978057359717e-05, "loss": 1.5015, "step": 73000 }, { "epoch": 1.11, "learning_rate": 1.6290745030355333e-05, "loss": 1.4928, "step": 73500 }, { "epoch": 1.12, "learning_rate": 1.626551200335095e-05, "loss": 1.4982, "step": 74000 }, { "epoch": 1.13, "learning_rate": 1.624027897634656e-05, "loss": 1.484, "step": 74500 }, { "epoch": 1.14, "learning_rate": 1.6215045949342177e-05, "loss": 1.4812, "step": 75000 }, { "epoch": 1.14, "learning_rate": 1.618981292233779e-05, "loss": 1.5233, "step": 75500 }, { "epoch": 1.15, "learning_rate": 1.6164579895333406e-05, "loss": 1.4745, "step": 76000 }, { "epoch": 1.16, "learning_rate": 1.613934686832902e-05, "loss": 1.5162, "step": 76500 }, { "epoch": 1.17, "learning_rate": 1.6114113841324634e-05, "loss": 1.4978, "step": 77000 }, { "epoch": 1.17, "learning_rate": 1.608888081432025e-05, "loss": 1.4742, "step": 77500 }, { "epoch": 1.18, "learning_rate": 1.6063647787315862e-05, "loss": 1.478, "step": 78000 }, { "epoch": 1.19, "learning_rate": 1.603841476031148e-05, "loss": 1.479, "step": 78500 }, { "epoch": 1.2, "learning_rate": 1.6013181733307094e-05, "loss": 1.4811, "step": 79000 }, { "epoch": 1.2, "learning_rate": 1.5987948706302707e-05, "loss": 1.4882, "step": 79500 }, { "epoch": 1.21, "learning_rate": 1.5962715679298323e-05, "loss": 1.4776, "step": 80000 }, { "epoch": 1.22, "learning_rate": 1.5937482652293935e-05, "loss": 1.4561, "step": 80500 }, { "epoch": 1.23, "learning_rate": 1.591224962528955e-05, "loss": 1.4773, "step": 81000 }, { "epoch": 1.23, "learning_rate": 1.5887016598285164e-05, "loss": 1.4672, "step": 81500 }, { "epoch": 1.24, "learning_rate": 1.586178357128078e-05, "loss": 1.485, "step": 82000 }, { "epoch": 1.25, "learning_rate": 1.5836550544276396e-05, "loss": 1.4688, "step": 82500 }, { "epoch": 1.26, "learning_rate": 1.5811317517272008e-05, "loss": 1.4922, "step": 83000 }, { "epoch": 1.26, "learning_rate": 1.5786084490267624e-05, "loss": 1.451, "step": 83500 }, { "epoch": 1.27, "learning_rate": 1.5760851463263237e-05, "loss": 1.4635, "step": 84000 }, { "epoch": 1.28, "learning_rate": 1.5735618436258853e-05, "loss": 1.4393, "step": 84500 }, { "epoch": 1.29, "learning_rate": 1.571038540925447e-05, "loss": 1.4622, "step": 85000 }, { "epoch": 1.29, "learning_rate": 1.568515238225008e-05, "loss": 1.4618, "step": 85500 }, { "epoch": 1.3, "learning_rate": 1.5659919355245697e-05, "loss": 1.4717, "step": 86000 }, { "epoch": 1.31, "learning_rate": 1.563468632824131e-05, "loss": 1.4269, "step": 86500 }, { "epoch": 1.32, "learning_rate": 1.5609453301236922e-05, "loss": 1.4413, "step": 87000 }, { "epoch": 1.32, "learning_rate": 1.5584220274232538e-05, "loss": 1.4491, "step": 87500 }, { "epoch": 1.33, "learning_rate": 1.5558987247228154e-05, "loss": 1.4621, "step": 88000 }, { "epoch": 1.34, "learning_rate": 1.5533754220223766e-05, "loss": 1.4352, "step": 88500 }, { "epoch": 1.35, "learning_rate": 1.5508521193219382e-05, "loss": 1.4291, "step": 89000 }, { "epoch": 1.36, "learning_rate": 1.5483288166214995e-05, "loss": 1.4476, "step": 89500 }, { "epoch": 1.36, "learning_rate": 1.545805513921061e-05, "loss": 1.4675, "step": 90000 }, { "epoch": 1.37, "learning_rate": 1.5432822112206223e-05, "loss": 1.4022, "step": 90500 }, { "epoch": 1.38, "learning_rate": 1.540758908520184e-05, "loss": 1.418, "step": 91000 }, { "epoch": 1.39, "learning_rate": 1.5382356058197455e-05, "loss": 1.4323, "step": 91500 }, { "epoch": 1.39, "learning_rate": 1.5357123031193068e-05, "loss": 1.4493, "step": 92000 }, { "epoch": 1.4, "learning_rate": 1.5331890004188684e-05, "loss": 1.4278, "step": 92500 }, { "epoch": 1.41, "learning_rate": 1.5306656977184296e-05, "loss": 1.422, "step": 93000 }, { "epoch": 1.42, "learning_rate": 1.5281423950179912e-05, "loss": 1.4329, "step": 93500 }, { "epoch": 1.42, "learning_rate": 1.5256190923175526e-05, "loss": 1.428, "step": 94000 }, { "epoch": 1.43, "learning_rate": 1.523095789617114e-05, "loss": 1.4299, "step": 94500 }, { "epoch": 1.44, "learning_rate": 1.5205724869166756e-05, "loss": 1.4482, "step": 95000 }, { "epoch": 1.45, "learning_rate": 1.518049184216237e-05, "loss": 1.4555, "step": 95500 }, { "epoch": 1.45, "learning_rate": 1.5155258815157985e-05, "loss": 1.4202, "step": 96000 }, { "epoch": 1.46, "learning_rate": 1.51300257881536e-05, "loss": 1.4404, "step": 96500 }, { "epoch": 1.47, "learning_rate": 1.5104792761149213e-05, "loss": 1.4107, "step": 97000 }, { "epoch": 1.48, "learning_rate": 1.507955973414483e-05, "loss": 1.3974, "step": 97500 }, { "epoch": 1.48, "learning_rate": 1.5054326707140444e-05, "loss": 1.4127, "step": 98000 }, { "epoch": 1.49, "learning_rate": 1.5029093680136058e-05, "loss": 1.4157, "step": 98500 }, { "epoch": 1.5, "learning_rate": 1.5003860653131672e-05, "loss": 1.4374, "step": 99000 }, { "epoch": 1.51, "learning_rate": 1.4978627626127286e-05, "loss": 1.4116, "step": 99500 }, { "epoch": 1.51, "learning_rate": 1.49533945991229e-05, "loss": 1.3902, "step": 100000 }, { "epoch": 1.52, "learning_rate": 1.4928161572118516e-05, "loss": 1.4201, "step": 100500 }, { "epoch": 1.53, "learning_rate": 1.490292854511413e-05, "loss": 1.4126, "step": 101000 }, { "epoch": 1.54, "learning_rate": 1.4877695518109745e-05, "loss": 1.4185, "step": 101500 }, { "epoch": 1.54, "learning_rate": 1.4852462491105359e-05, "loss": 1.3914, "step": 102000 }, { "epoch": 1.55, "learning_rate": 1.4827229464100973e-05, "loss": 1.4107, "step": 102500 }, { "epoch": 1.56, "learning_rate": 1.480199643709659e-05, "loss": 1.3912, "step": 103000 }, { "epoch": 1.57, "learning_rate": 1.4776763410092204e-05, "loss": 1.4036, "step": 103500 }, { "epoch": 1.57, "learning_rate": 1.4751530383087818e-05, "loss": 1.4003, "step": 104000 }, { "epoch": 1.58, "learning_rate": 1.4726297356083432e-05, "loss": 1.376, "step": 104500 }, { "epoch": 1.59, "learning_rate": 1.4701064329079046e-05, "loss": 1.4063, "step": 105000 }, { "epoch": 1.6, "learning_rate": 1.467583130207466e-05, "loss": 1.4074, "step": 105500 }, { "epoch": 1.6, "learning_rate": 1.4650598275070276e-05, "loss": 1.3998, "step": 106000 }, { "epoch": 1.61, "learning_rate": 1.462536524806589e-05, "loss": 1.3836, "step": 106500 }, { "epoch": 1.62, "learning_rate": 1.4600132221061505e-05, "loss": 1.3912, "step": 107000 }, { "epoch": 1.63, "learning_rate": 1.4574899194057119e-05, "loss": 1.416, "step": 107500 }, { "epoch": 1.64, "learning_rate": 1.4549666167052733e-05, "loss": 1.3718, "step": 108000 }, { "epoch": 1.64, "learning_rate": 1.452443314004835e-05, "loss": 1.3984, "step": 108500 }, { "epoch": 1.65, "learning_rate": 1.4499200113043963e-05, "loss": 1.3896, "step": 109000 }, { "epoch": 1.66, "learning_rate": 1.4473967086039578e-05, "loss": 1.3883, "step": 109500 }, { "epoch": 1.67, "learning_rate": 1.4448734059035192e-05, "loss": 1.3608, "step": 110000 }, { "epoch": 1.67, "learning_rate": 1.4423501032030806e-05, "loss": 1.3636, "step": 110500 }, { "epoch": 1.68, "learning_rate": 1.439826800502642e-05, "loss": 1.3777, "step": 111000 }, { "epoch": 1.69, "learning_rate": 1.4373034978022036e-05, "loss": 1.3714, "step": 111500 }, { "epoch": 1.7, "learning_rate": 1.4347801951017649e-05, "loss": 1.3965, "step": 112000 }, { "epoch": 1.7, "learning_rate": 1.4322568924013263e-05, "loss": 1.3417, "step": 112500 }, { "epoch": 1.71, "learning_rate": 1.4297335897008877e-05, "loss": 1.3878, "step": 113000 }, { "epoch": 1.72, "learning_rate": 1.4272102870004492e-05, "loss": 1.3798, "step": 113500 }, { "epoch": 1.73, "learning_rate": 1.4246869843000106e-05, "loss": 1.3626, "step": 114000 }, { "epoch": 1.73, "learning_rate": 1.422163681599572e-05, "loss": 1.39, "step": 114500 }, { "epoch": 1.74, "learning_rate": 1.4196403788991336e-05, "loss": 1.3792, "step": 115000 }, { "epoch": 1.75, "learning_rate": 1.417117076198695e-05, "loss": 1.3883, "step": 115500 }, { "epoch": 1.76, "learning_rate": 1.4145937734982564e-05, "loss": 1.3978, "step": 116000 }, { "epoch": 1.76, "learning_rate": 1.4120704707978179e-05, "loss": 1.3772, "step": 116500 }, { "epoch": 1.77, "learning_rate": 1.4095471680973793e-05, "loss": 1.3643, "step": 117000 }, { "epoch": 1.78, "learning_rate": 1.4070238653969409e-05, "loss": 1.35, "step": 117500 }, { "epoch": 1.79, "learning_rate": 1.4045005626965023e-05, "loss": 1.3809, "step": 118000 }, { "epoch": 1.79, "learning_rate": 1.4019772599960637e-05, "loss": 1.3539, "step": 118500 }, { "epoch": 1.8, "learning_rate": 1.3994539572956251e-05, "loss": 1.3944, "step": 119000 }, { "epoch": 1.81, "learning_rate": 1.3969306545951866e-05, "loss": 1.3338, "step": 119500 }, { "epoch": 1.82, "learning_rate": 1.394407351894748e-05, "loss": 1.3415, "step": 120000 }, { "epoch": 1.82, "learning_rate": 1.3918840491943096e-05, "loss": 1.3685, "step": 120500 }, { "epoch": 1.83, "learning_rate": 1.389360746493871e-05, "loss": 1.3734, "step": 121000 }, { "epoch": 1.84, "learning_rate": 1.3868374437934324e-05, "loss": 1.3617, "step": 121500 }, { "epoch": 1.85, "learning_rate": 1.3843141410929939e-05, "loss": 1.3585, "step": 122000 }, { "epoch": 1.85, "learning_rate": 1.3817908383925553e-05, "loss": 1.3454, "step": 122500 }, { "epoch": 1.86, "learning_rate": 1.3792675356921169e-05, "loss": 1.3404, "step": 123000 }, { "epoch": 1.87, "learning_rate": 1.3767442329916783e-05, "loss": 1.34, "step": 123500 }, { "epoch": 1.88, "learning_rate": 1.3742209302912397e-05, "loss": 1.3444, "step": 124000 }, { "epoch": 1.88, "learning_rate": 1.3716976275908011e-05, "loss": 1.3538, "step": 124500 }, { "epoch": 1.89, "learning_rate": 1.3691743248903626e-05, "loss": 1.3349, "step": 125000 }, { "epoch": 1.9, "learning_rate": 1.366651022189924e-05, "loss": 1.3484, "step": 125500 }, { "epoch": 1.91, "learning_rate": 1.3641277194894856e-05, "loss": 1.3337, "step": 126000 }, { "epoch": 1.92, "learning_rate": 1.361604416789047e-05, "loss": 1.3376, "step": 126500 }, { "epoch": 1.92, "learning_rate": 1.3590811140886084e-05, "loss": 1.3361, "step": 127000 }, { "epoch": 1.93, "learning_rate": 1.3565578113881699e-05, "loss": 1.3491, "step": 127500 }, { "epoch": 1.94, "learning_rate": 1.3540345086877313e-05, "loss": 1.3185, "step": 128000 }, { "epoch": 1.95, "learning_rate": 1.3515112059872929e-05, "loss": 1.3663, "step": 128500 }, { "epoch": 1.95, "learning_rate": 1.3489879032868543e-05, "loss": 1.3703, "step": 129000 }, { "epoch": 1.96, "learning_rate": 1.3464646005864157e-05, "loss": 1.336, "step": 129500 }, { "epoch": 1.97, "learning_rate": 1.3439412978859771e-05, "loss": 1.3374, "step": 130000 }, { "epoch": 1.98, "learning_rate": 1.3414179951855386e-05, "loss": 1.3132, "step": 130500 }, { "epoch": 1.98, "learning_rate": 1.3388946924851002e-05, "loss": 1.3289, "step": 131000 }, { "epoch": 1.99, "learning_rate": 1.3363713897846616e-05, "loss": 1.359, "step": 131500 }, { "epoch": 2.0, "learning_rate": 1.333848087084223e-05, "loss": 1.3327, "step": 132000 }, { "epoch": 2.01, "learning_rate": 1.3313247843837844e-05, "loss": 1.3262, "step": 132500 }, { "epoch": 2.01, "learning_rate": 1.3288014816833458e-05, "loss": 1.3355, "step": 133000 }, { "epoch": 2.02, "learning_rate": 1.3262781789829073e-05, "loss": 1.2965, "step": 133500 }, { "epoch": 2.03, "learning_rate": 1.3237548762824689e-05, "loss": 1.3154, "step": 134000 }, { "epoch": 2.04, "learning_rate": 1.3212315735820303e-05, "loss": 1.3214, "step": 134500 }, { "epoch": 2.04, "learning_rate": 1.3187082708815917e-05, "loss": 1.315, "step": 135000 }, { "epoch": 2.05, "learning_rate": 1.3161849681811531e-05, "loss": 1.3065, "step": 135500 }, { "epoch": 2.06, "learning_rate": 1.3136616654807146e-05, "loss": 1.334, "step": 136000 }, { "epoch": 2.07, "learning_rate": 1.3111383627802761e-05, "loss": 1.3141, "step": 136500 }, { "epoch": 2.07, "learning_rate": 1.3086150600798372e-05, "loss": 1.3115, "step": 137000 }, { "epoch": 2.08, "learning_rate": 1.3060917573793988e-05, "loss": 1.324, "step": 137500 }, { "epoch": 2.09, "learning_rate": 1.3035684546789602e-05, "loss": 1.3269, "step": 138000 }, { "epoch": 2.1, "learning_rate": 1.3010451519785217e-05, "loss": 1.3237, "step": 138500 }, { "epoch": 2.1, "learning_rate": 1.2985218492780831e-05, "loss": 1.3176, "step": 139000 }, { "epoch": 2.11, "learning_rate": 1.2959985465776445e-05, "loss": 1.3242, "step": 139500 }, { "epoch": 2.12, "learning_rate": 1.293475243877206e-05, "loss": 1.2984, "step": 140000 }, { "epoch": 2.13, "learning_rate": 1.2909519411767675e-05, "loss": 1.299, "step": 140500 }, { "epoch": 2.13, "learning_rate": 1.288428638476329e-05, "loss": 1.3124, "step": 141000 }, { "epoch": 2.14, "learning_rate": 1.2859053357758904e-05, "loss": 1.3012, "step": 141500 }, { "epoch": 2.15, "learning_rate": 1.2833820330754518e-05, "loss": 1.3082, "step": 142000 }, { "epoch": 2.16, "learning_rate": 1.2808587303750132e-05, "loss": 1.309, "step": 142500 }, { "epoch": 2.16, "learning_rate": 1.2783354276745748e-05, "loss": 1.2984, "step": 143000 }, { "epoch": 2.17, "learning_rate": 1.2758121249741362e-05, "loss": 1.2796, "step": 143500 }, { "epoch": 2.18, "learning_rate": 1.2732888222736977e-05, "loss": 1.2999, "step": 144000 }, { "epoch": 2.19, "learning_rate": 1.2707655195732591e-05, "loss": 1.3163, "step": 144500 }, { "epoch": 2.2, "learning_rate": 1.2682422168728205e-05, "loss": 1.2858, "step": 145000 }, { "epoch": 2.2, "learning_rate": 1.2657189141723821e-05, "loss": 1.303, "step": 145500 }, { "epoch": 2.21, "learning_rate": 1.2631956114719435e-05, "loss": 1.3047, "step": 146000 }, { "epoch": 2.22, "learning_rate": 1.260672308771505e-05, "loss": 1.2985, "step": 146500 }, { "epoch": 2.23, "learning_rate": 1.2581490060710664e-05, "loss": 1.2954, "step": 147000 }, { "epoch": 2.23, "learning_rate": 1.2556257033706278e-05, "loss": 1.3003, "step": 147500 }, { "epoch": 2.24, "learning_rate": 1.2531024006701892e-05, "loss": 1.2755, "step": 148000 }, { "epoch": 2.25, "learning_rate": 1.2505790979697508e-05, "loss": 1.2912, "step": 148500 }, { "epoch": 2.26, "learning_rate": 1.2480557952693122e-05, "loss": 1.2818, "step": 149000 }, { "epoch": 2.26, "learning_rate": 1.2455324925688737e-05, "loss": 1.2873, "step": 149500 }, { "epoch": 2.27, "learning_rate": 1.243009189868435e-05, "loss": 1.2935, "step": 150000 }, { "epoch": 2.28, "learning_rate": 1.2404858871679965e-05, "loss": 1.2652, "step": 150500 }, { "epoch": 2.29, "learning_rate": 1.2379625844675581e-05, "loss": 1.3012, "step": 151000 }, { "epoch": 2.29, "learning_rate": 1.2354392817671195e-05, "loss": 1.2873, "step": 151500 }, { "epoch": 2.3, "learning_rate": 1.232915979066681e-05, "loss": 1.2863, "step": 152000 }, { "epoch": 2.31, "learning_rate": 1.2303926763662424e-05, "loss": 1.3169, "step": 152500 }, { "epoch": 2.32, "learning_rate": 1.2278693736658038e-05, "loss": 1.2902, "step": 153000 }, { "epoch": 2.32, "learning_rate": 1.2253460709653652e-05, "loss": 1.2854, "step": 153500 }, { "epoch": 2.33, "learning_rate": 1.2228227682649268e-05, "loss": 1.3066, "step": 154000 }, { "epoch": 2.34, "learning_rate": 1.2202994655644882e-05, "loss": 1.2713, "step": 154500 }, { "epoch": 2.35, "learning_rate": 1.2177761628640496e-05, "loss": 1.2783, "step": 155000 }, { "epoch": 2.35, "learning_rate": 1.215252860163611e-05, "loss": 1.3072, "step": 155500 }, { "epoch": 2.36, "learning_rate": 1.2127295574631725e-05, "loss": 1.2997, "step": 156000 }, { "epoch": 2.37, "learning_rate": 1.2102062547627341e-05, "loss": 1.2875, "step": 156500 }, { "epoch": 2.38, "learning_rate": 1.2076829520622955e-05, "loss": 1.2819, "step": 157000 }, { "epoch": 2.38, "learning_rate": 1.205159649361857e-05, "loss": 1.2678, "step": 157500 }, { "epoch": 2.39, "learning_rate": 1.2026363466614184e-05, "loss": 1.3049, "step": 158000 }, { "epoch": 2.4, "learning_rate": 1.2001130439609798e-05, "loss": 1.3011, "step": 158500 }, { "epoch": 2.41, "learning_rate": 1.1975897412605412e-05, "loss": 1.2912, "step": 159000 }, { "epoch": 2.41, "learning_rate": 1.1950664385601028e-05, "loss": 1.2833, "step": 159500 }, { "epoch": 2.42, "learning_rate": 1.1925431358596642e-05, "loss": 1.28, "step": 160000 }, { "epoch": 2.43, "learning_rate": 1.1900198331592256e-05, "loss": 1.2757, "step": 160500 }, { "epoch": 2.44, "learning_rate": 1.187496530458787e-05, "loss": 1.2693, "step": 161000 }, { "epoch": 2.45, "learning_rate": 1.1849732277583485e-05, "loss": 1.2828, "step": 161500 }, { "epoch": 2.45, "learning_rate": 1.1824499250579097e-05, "loss": 1.2503, "step": 162000 }, { "epoch": 2.46, "learning_rate": 1.1799266223574712e-05, "loss": 1.2652, "step": 162500 }, { "epoch": 2.47, "learning_rate": 1.1774033196570328e-05, "loss": 1.2594, "step": 163000 }, { "epoch": 2.48, "learning_rate": 1.1748800169565942e-05, "loss": 1.2702, "step": 163500 }, { "epoch": 2.48, "learning_rate": 1.1723567142561556e-05, "loss": 1.2894, "step": 164000 }, { "epoch": 2.49, "learning_rate": 1.169833411555717e-05, "loss": 1.2768, "step": 164500 }, { "epoch": 2.5, "learning_rate": 1.1673101088552785e-05, "loss": 1.2626, "step": 165000 }, { "epoch": 2.51, "learning_rate": 1.16478680615484e-05, "loss": 1.2833, "step": 165500 }, { "epoch": 2.51, "learning_rate": 1.1622635034544015e-05, "loss": 1.2663, "step": 166000 }, { "epoch": 2.52, "learning_rate": 1.1597402007539629e-05, "loss": 1.2604, "step": 166500 }, { "epoch": 2.53, "learning_rate": 1.1572168980535243e-05, "loss": 1.3075, "step": 167000 }, { "epoch": 2.54, "learning_rate": 1.1546935953530857e-05, "loss": 1.2865, "step": 167500 }, { "epoch": 2.54, "learning_rate": 1.1521702926526472e-05, "loss": 1.2404, "step": 168000 }, { "epoch": 2.55, "learning_rate": 1.1496469899522088e-05, "loss": 1.2784, "step": 168500 }, { "epoch": 2.56, "learning_rate": 1.1471236872517702e-05, "loss": 1.2711, "step": 169000 }, { "epoch": 2.57, "learning_rate": 1.1446003845513316e-05, "loss": 1.2725, "step": 169500 }, { "epoch": 2.57, "learning_rate": 1.142077081850893e-05, "loss": 1.287, "step": 170000 }, { "epoch": 2.58, "learning_rate": 1.1395537791504544e-05, "loss": 1.2588, "step": 170500 }, { "epoch": 2.59, "learning_rate": 1.137030476450016e-05, "loss": 1.2637, "step": 171000 }, { "epoch": 2.6, "learning_rate": 1.1345071737495775e-05, "loss": 1.2669, "step": 171500 }, { "epoch": 2.6, "learning_rate": 1.1319838710491389e-05, "loss": 1.2662, "step": 172000 }, { "epoch": 2.61, "learning_rate": 1.1294605683487003e-05, "loss": 1.276, "step": 172500 }, { "epoch": 2.62, "learning_rate": 1.1269372656482617e-05, "loss": 1.2637, "step": 173000 }, { "epoch": 2.63, "learning_rate": 1.1244139629478232e-05, "loss": 1.269, "step": 173500 }, { "epoch": 2.63, "learning_rate": 1.1218906602473847e-05, "loss": 1.2505, "step": 174000 }, { "epoch": 2.64, "learning_rate": 1.1193673575469462e-05, "loss": 1.2476, "step": 174500 }, { "epoch": 2.65, "learning_rate": 1.1168440548465076e-05, "loss": 1.2341, "step": 175000 }, { "epoch": 2.66, "learning_rate": 1.114320752146069e-05, "loss": 1.2746, "step": 175500 }, { "epoch": 2.66, "learning_rate": 1.1117974494456304e-05, "loss": 1.2675, "step": 176000 }, { "epoch": 2.67, "learning_rate": 1.109274146745192e-05, "loss": 1.2921, "step": 176500 }, { "epoch": 2.68, "learning_rate": 1.1067508440447535e-05, "loss": 1.2655, "step": 177000 }, { "epoch": 2.69, "learning_rate": 1.1042275413443149e-05, "loss": 1.2351, "step": 177500 }, { "epoch": 2.69, "learning_rate": 1.1017042386438763e-05, "loss": 1.2519, "step": 178000 }, { "epoch": 2.7, "learning_rate": 1.0991809359434377e-05, "loss": 1.2453, "step": 178500 }, { "epoch": 2.71, "learning_rate": 1.0966576332429991e-05, "loss": 1.2487, "step": 179000 }, { "epoch": 2.72, "learning_rate": 1.0941343305425607e-05, "loss": 1.2243, "step": 179500 }, { "epoch": 2.73, "learning_rate": 1.0916110278421222e-05, "loss": 1.2586, "step": 180000 }, { "epoch": 2.73, "learning_rate": 1.0890877251416836e-05, "loss": 1.2514, "step": 180500 }, { "epoch": 2.74, "learning_rate": 1.086564422441245e-05, "loss": 1.2489, "step": 181000 }, { "epoch": 2.75, "learning_rate": 1.0840411197408064e-05, "loss": 1.2478, "step": 181500 }, { "epoch": 2.76, "learning_rate": 1.081517817040368e-05, "loss": 1.2538, "step": 182000 }, { "epoch": 2.76, "learning_rate": 1.0789945143399294e-05, "loss": 1.2287, "step": 182500 }, { "epoch": 2.77, "learning_rate": 1.0764712116394909e-05, "loss": 1.2564, "step": 183000 }, { "epoch": 2.78, "learning_rate": 1.0739479089390523e-05, "loss": 1.2537, "step": 183500 }, { "epoch": 2.79, "learning_rate": 1.0714246062386137e-05, "loss": 1.2487, "step": 184000 }, { "epoch": 2.79, "learning_rate": 1.0689013035381753e-05, "loss": 1.246, "step": 184500 }, { "epoch": 2.8, "learning_rate": 1.0663780008377367e-05, "loss": 1.2306, "step": 185000 }, { "epoch": 2.81, "learning_rate": 1.0638546981372982e-05, "loss": 1.2301, "step": 185500 }, { "epoch": 2.82, "learning_rate": 1.0613313954368596e-05, "loss": 1.2467, "step": 186000 }, { "epoch": 2.82, "learning_rate": 1.0588080927364208e-05, "loss": 1.25, "step": 186500 }, { "epoch": 2.83, "learning_rate": 1.0562847900359823e-05, "loss": 1.2359, "step": 187000 }, { "epoch": 2.84, "learning_rate": 1.0537614873355437e-05, "loss": 1.2429, "step": 187500 }, { "epoch": 2.85, "learning_rate": 1.0512381846351051e-05, "loss": 1.2487, "step": 188000 }, { "epoch": 2.85, "learning_rate": 1.0487148819346667e-05, "loss": 1.233, "step": 188500 }, { "epoch": 2.86, "learning_rate": 1.0461915792342281e-05, "loss": 1.2311, "step": 189000 }, { "epoch": 2.87, "learning_rate": 1.0436682765337895e-05, "loss": 1.2138, "step": 189500 }, { "epoch": 2.88, "learning_rate": 1.041144973833351e-05, "loss": 1.2425, "step": 190000 }, { "epoch": 2.88, "learning_rate": 1.0386216711329124e-05, "loss": 1.2315, "step": 190500 }, { "epoch": 2.89, "learning_rate": 1.036098368432474e-05, "loss": 1.2616, "step": 191000 }, { "epoch": 2.9, "learning_rate": 1.0335750657320354e-05, "loss": 1.2043, "step": 191500 }, { "epoch": 2.91, "learning_rate": 1.0310517630315968e-05, "loss": 1.2243, "step": 192000 }, { "epoch": 2.91, "learning_rate": 1.0285284603311583e-05, "loss": 1.2432, "step": 192500 }, { "epoch": 2.92, "learning_rate": 1.0260051576307197e-05, "loss": 1.225, "step": 193000 }, { "epoch": 2.93, "learning_rate": 1.0234818549302811e-05, "loss": 1.2464, "step": 193500 }, { "epoch": 2.94, "learning_rate": 1.0209585522298427e-05, "loss": 1.2439, "step": 194000 }, { "epoch": 2.94, "learning_rate": 1.0184352495294041e-05, "loss": 1.2283, "step": 194500 }, { "epoch": 2.95, "learning_rate": 1.0159119468289655e-05, "loss": 1.2511, "step": 195000 }, { "epoch": 2.96, "learning_rate": 1.013388644128527e-05, "loss": 1.229, "step": 195500 }, { "epoch": 2.97, "learning_rate": 1.0108653414280884e-05, "loss": 1.2392, "step": 196000 }, { "epoch": 2.97, "learning_rate": 1.00834203872765e-05, "loss": 1.2513, "step": 196500 }, { "epoch": 2.98, "learning_rate": 1.0058187360272114e-05, "loss": 1.2183, "step": 197000 }, { "epoch": 2.99, "learning_rate": 1.0032954333267728e-05, "loss": 1.2347, "step": 197500 }, { "epoch": 3.0, "learning_rate": 1.0007721306263342e-05, "loss": 1.2627, "step": 198000 }, { "epoch": 3.01, "learning_rate": 9.982488279258957e-06, "loss": 1.2184, "step": 198500 }, { "epoch": 3.01, "learning_rate": 9.957255252254573e-06, "loss": 1.2124, "step": 199000 }, { "epoch": 3.02, "learning_rate": 9.932022225250187e-06, "loss": 1.2295, "step": 199500 }, { "epoch": 3.03, "learning_rate": 9.906789198245801e-06, "loss": 1.2152, "step": 200000 }, { "epoch": 3.04, "learning_rate": 9.881556171241415e-06, "loss": 1.2166, "step": 200500 }, { "epoch": 3.04, "learning_rate": 9.85632314423703e-06, "loss": 1.2043, "step": 201000 }, { "epoch": 3.05, "learning_rate": 9.831090117232644e-06, "loss": 1.2175, "step": 201500 }, { "epoch": 3.06, "learning_rate": 9.80585709022826e-06, "loss": 1.2167, "step": 202000 }, { "epoch": 3.07, "learning_rate": 9.780624063223874e-06, "loss": 1.2092, "step": 202500 }, { "epoch": 3.07, "learning_rate": 9.755391036219488e-06, "loss": 1.2046, "step": 203000 }, { "epoch": 3.08, "learning_rate": 9.730158009215102e-06, "loss": 1.2283, "step": 203500 }, { "epoch": 3.09, "learning_rate": 9.704924982210717e-06, "loss": 1.21, "step": 204000 }, { "epoch": 3.1, "learning_rate": 9.679691955206333e-06, "loss": 1.2099, "step": 204500 }, { "epoch": 3.1, "learning_rate": 9.654458928201947e-06, "loss": 1.2227, "step": 205000 }, { "epoch": 3.11, "learning_rate": 9.62922590119756e-06, "loss": 1.2168, "step": 205500 }, { "epoch": 3.12, "learning_rate": 9.603992874193174e-06, "loss": 1.1994, "step": 206000 }, { "epoch": 3.13, "learning_rate": 9.57875984718879e-06, "loss": 1.2261, "step": 206500 }, { "epoch": 3.13, "learning_rate": 9.553526820184404e-06, "loss": 1.2076, "step": 207000 }, { "epoch": 3.14, "learning_rate": 9.528293793180018e-06, "loss": 1.1942, "step": 207500 }, { "epoch": 3.15, "learning_rate": 9.503060766175632e-06, "loss": 1.1884, "step": 208000 }, { "epoch": 3.16, "learning_rate": 9.477827739171246e-06, "loss": 1.2089, "step": 208500 }, { "epoch": 3.16, "learning_rate": 9.452594712166862e-06, "loss": 1.2118, "step": 209000 }, { "epoch": 3.17, "learning_rate": 9.427361685162477e-06, "loss": 1.1925, "step": 209500 }, { "epoch": 3.18, "learning_rate": 9.40212865815809e-06, "loss": 1.2072, "step": 210000 }, { "epoch": 3.19, "learning_rate": 9.376895631153705e-06, "loss": 1.2392, "step": 210500 }, { "epoch": 3.19, "learning_rate": 9.35166260414932e-06, "loss": 1.1911, "step": 211000 }, { "epoch": 3.2, "learning_rate": 9.326429577144933e-06, "loss": 1.2112, "step": 211500 }, { "epoch": 3.21, "learning_rate": 9.30119655014055e-06, "loss": 1.2101, "step": 212000 }, { "epoch": 3.22, "learning_rate": 9.275963523136164e-06, "loss": 1.2202, "step": 212500 }, { "epoch": 3.22, "learning_rate": 9.250730496131778e-06, "loss": 1.1832, "step": 213000 }, { "epoch": 3.23, "learning_rate": 9.225497469127392e-06, "loss": 1.2253, "step": 213500 }, { "epoch": 3.24, "learning_rate": 9.200264442123006e-06, "loss": 1.1799, "step": 214000 }, { "epoch": 3.25, "learning_rate": 9.175031415118622e-06, "loss": 1.1755, "step": 214500 }, { "epoch": 3.26, "learning_rate": 9.149798388114236e-06, "loss": 1.1861, "step": 215000 }, { "epoch": 3.26, "learning_rate": 9.12456536110985e-06, "loss": 1.1737, "step": 215500 }, { "epoch": 3.27, "learning_rate": 9.099332334105465e-06, "loss": 1.1994, "step": 216000 }, { "epoch": 3.28, "learning_rate": 9.07409930710108e-06, "loss": 1.2178, "step": 216500 }, { "epoch": 3.29, "learning_rate": 9.048866280096693e-06, "loss": 1.1875, "step": 217000 }, { "epoch": 3.29, "learning_rate": 9.023633253092308e-06, "loss": 1.1975, "step": 217500 }, { "epoch": 3.3, "learning_rate": 8.998400226087922e-06, "loss": 1.2149, "step": 218000 }, { "epoch": 3.31, "learning_rate": 8.973167199083536e-06, "loss": 1.2025, "step": 218500 }, { "epoch": 3.32, "learning_rate": 8.947934172079152e-06, "loss": 1.1945, "step": 219000 }, { "epoch": 3.32, "learning_rate": 8.922701145074766e-06, "loss": 1.2, "step": 219500 }, { "epoch": 3.33, "learning_rate": 8.89746811807038e-06, "loss": 1.1883, "step": 220000 }, { "epoch": 3.34, "learning_rate": 8.872235091065995e-06, "loss": 1.2085, "step": 220500 }, { "epoch": 3.35, "learning_rate": 8.847002064061609e-06, "loss": 1.2045, "step": 221000 }, { "epoch": 3.35, "learning_rate": 8.821769037057223e-06, "loss": 1.1894, "step": 221500 }, { "epoch": 3.36, "learning_rate": 8.796536010052839e-06, "loss": 1.1832, "step": 222000 }, { "epoch": 3.37, "learning_rate": 8.771302983048453e-06, "loss": 1.1688, "step": 222500 }, { "epoch": 3.38, "learning_rate": 8.746069956044068e-06, "loss": 1.2142, "step": 223000 }, { "epoch": 3.38, "learning_rate": 8.720836929039682e-06, "loss": 1.1868, "step": 223500 }, { "epoch": 3.39, "learning_rate": 8.695603902035296e-06, "loss": 1.2084, "step": 224000 }, { "epoch": 3.4, "learning_rate": 8.670370875030912e-06, "loss": 1.1819, "step": 224500 }, { "epoch": 3.41, "learning_rate": 8.645137848026526e-06, "loss": 1.1974, "step": 225000 }, { "epoch": 3.41, "learning_rate": 8.61990482102214e-06, "loss": 1.1902, "step": 225500 }, { "epoch": 3.42, "learning_rate": 8.594671794017755e-06, "loss": 1.1873, "step": 226000 }, { "epoch": 3.43, "learning_rate": 8.569438767013369e-06, "loss": 1.1803, "step": 226500 }, { "epoch": 3.44, "learning_rate": 8.544205740008983e-06, "loss": 1.1896, "step": 227000 }, { "epoch": 3.44, "learning_rate": 8.518972713004599e-06, "loss": 1.1651, "step": 227500 }, { "epoch": 3.45, "learning_rate": 8.493739686000213e-06, "loss": 1.2066, "step": 228000 }, { "epoch": 3.46, "learning_rate": 8.468506658995828e-06, "loss": 1.1972, "step": 228500 }, { "epoch": 3.47, "learning_rate": 8.443273631991442e-06, "loss": 1.2008, "step": 229000 }, { "epoch": 3.47, "learning_rate": 8.418040604987056e-06, "loss": 1.1903, "step": 229500 }, { "epoch": 3.48, "learning_rate": 8.39280757798267e-06, "loss": 1.1912, "step": 230000 }, { "epoch": 3.49, "learning_rate": 8.367574550978284e-06, "loss": 1.2012, "step": 230500 }, { "epoch": 3.5, "learning_rate": 8.342341523973899e-06, "loss": 1.1836, "step": 231000 }, { "epoch": 3.5, "learning_rate": 8.317108496969513e-06, "loss": 1.1667, "step": 231500 }, { "epoch": 3.51, "learning_rate": 8.291875469965129e-06, "loss": 1.1823, "step": 232000 }, { "epoch": 3.52, "learning_rate": 8.266642442960743e-06, "loss": 1.1591, "step": 232500 }, { "epoch": 3.53, "learning_rate": 8.241409415956357e-06, "loss": 1.1547, "step": 233000 }, { "epoch": 3.54, "learning_rate": 8.216176388951972e-06, "loss": 1.1759, "step": 233500 }, { "epoch": 3.54, "learning_rate": 8.190943361947586e-06, "loss": 1.1732, "step": 234000 }, { "epoch": 3.55, "learning_rate": 8.165710334943202e-06, "loss": 1.2048, "step": 234500 }, { "epoch": 3.56, "learning_rate": 8.140477307938816e-06, "loss": 1.1718, "step": 235000 }, { "epoch": 3.57, "learning_rate": 8.11524428093443e-06, "loss": 1.1891, "step": 235500 }, { "epoch": 3.57, "learning_rate": 8.090011253930044e-06, "loss": 1.1712, "step": 236000 }, { "epoch": 3.58, "learning_rate": 8.064778226925659e-06, "loss": 1.1521, "step": 236500 }, { "epoch": 3.59, "learning_rate": 8.039545199921273e-06, "loss": 1.1788, "step": 237000 }, { "epoch": 3.6, "learning_rate": 8.014312172916889e-06, "loss": 1.1822, "step": 237500 }, { "epoch": 3.6, "learning_rate": 7.989079145912503e-06, "loss": 1.1529, "step": 238000 }, { "epoch": 3.61, "learning_rate": 7.963846118908117e-06, "loss": 1.1929, "step": 238500 }, { "epoch": 3.62, "learning_rate": 7.938613091903731e-06, "loss": 1.1856, "step": 239000 }, { "epoch": 3.63, "learning_rate": 7.913380064899346e-06, "loss": 1.1885, "step": 239500 }, { "epoch": 3.63, "learning_rate": 7.888147037894962e-06, "loss": 1.1942, "step": 240000 }, { "epoch": 3.64, "learning_rate": 7.862914010890576e-06, "loss": 1.1776, "step": 240500 }, { "epoch": 3.65, "learning_rate": 7.83768098388619e-06, "loss": 1.1635, "step": 241000 }, { "epoch": 3.66, "learning_rate": 7.812447956881804e-06, "loss": 1.1753, "step": 241500 }, { "epoch": 3.66, "learning_rate": 7.787214929877419e-06, "loss": 1.2012, "step": 242000 }, { "epoch": 3.67, "learning_rate": 7.761981902873033e-06, "loss": 1.1649, "step": 242500 }, { "epoch": 3.68, "learning_rate": 7.736748875868647e-06, "loss": 1.1804, "step": 243000 }, { "epoch": 3.69, "learning_rate": 7.711515848864261e-06, "loss": 1.1853, "step": 243500 }, { "epoch": 3.69, "learning_rate": 7.686282821859875e-06, "loss": 1.1932, "step": 244000 }, { "epoch": 3.7, "learning_rate": 7.661049794855491e-06, "loss": 1.1918, "step": 244500 }, { "epoch": 3.71, "learning_rate": 7.635816767851106e-06, "loss": 1.1762, "step": 245000 }, { "epoch": 3.72, "learning_rate": 7.61058374084672e-06, "loss": 1.1731, "step": 245500 }, { "epoch": 3.72, "learning_rate": 7.585350713842334e-06, "loss": 1.1776, "step": 246000 }, { "epoch": 3.73, "learning_rate": 7.560117686837949e-06, "loss": 1.1952, "step": 246500 }, { "epoch": 3.74, "learning_rate": 7.5348846598335634e-06, "loss": 1.1493, "step": 247000 }, { "epoch": 3.75, "learning_rate": 7.509651632829178e-06, "loss": 1.2093, "step": 247500 }, { "epoch": 3.75, "learning_rate": 7.484418605824793e-06, "loss": 1.1711, "step": 248000 }, { "epoch": 3.76, "learning_rate": 7.459185578820407e-06, "loss": 1.1861, "step": 248500 }, { "epoch": 3.77, "learning_rate": 7.433952551816021e-06, "loss": 1.1625, "step": 249000 }, { "epoch": 3.78, "learning_rate": 7.408719524811636e-06, "loss": 1.1597, "step": 249500 }, { "epoch": 3.78, "learning_rate": 7.3834864978072505e-06, "loss": 1.1639, "step": 250000 }, { "epoch": 3.79, "learning_rate": 7.358253470802866e-06, "loss": 1.1958, "step": 250500 }, { "epoch": 3.8, "learning_rate": 7.33302044379848e-06, "loss": 1.1739, "step": 251000 }, { "epoch": 3.81, "learning_rate": 7.307787416794094e-06, "loss": 1.1808, "step": 251500 }, { "epoch": 3.82, "learning_rate": 7.282554389789709e-06, "loss": 1.1711, "step": 252000 }, { "epoch": 3.82, "learning_rate": 7.257321362785323e-06, "loss": 1.1676, "step": 252500 }, { "epoch": 3.83, "learning_rate": 7.232088335780938e-06, "loss": 1.1635, "step": 253000 }, { "epoch": 3.84, "learning_rate": 7.206855308776553e-06, "loss": 1.1554, "step": 253500 }, { "epoch": 3.85, "learning_rate": 7.181622281772167e-06, "loss": 1.1709, "step": 254000 }, { "epoch": 3.85, "learning_rate": 7.156389254767782e-06, "loss": 1.188, "step": 254500 }, { "epoch": 3.86, "learning_rate": 7.131156227763395e-06, "loss": 1.1686, "step": 255000 }, { "epoch": 3.87, "learning_rate": 7.10592320075901e-06, "loss": 1.1723, "step": 255500 }, { "epoch": 3.88, "learning_rate": 7.080690173754624e-06, "loss": 1.1436, "step": 256000 }, { "epoch": 3.88, "learning_rate": 7.055457146750239e-06, "loss": 1.1694, "step": 256500 }, { "epoch": 3.89, "learning_rate": 7.030224119745853e-06, "loss": 1.1541, "step": 257000 } ], "max_steps": 396306, "num_train_epochs": 6, "total_flos": 1.4494787538348096e+17, "trial_name": null, "trial_params": null }