{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9999926053552017, "global_step": 67616, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999260530052059e-05, "loss": 2.919, "step": 100 }, { "epoch": 0.0, "learning_rate": 4.9985210601041174e-05, "loss": 2.7423, "step": 200 }, { "epoch": 0.0, "learning_rate": 4.9977815901561766e-05, "loss": 2.6872, "step": 300 }, { "epoch": 0.01, "learning_rate": 4.997042120208235e-05, "loss": 2.6637, "step": 400 }, { "epoch": 0.01, "learning_rate": 4.996302650260294e-05, "loss": 2.6088, "step": 500 }, { "epoch": 0.01, "learning_rate": 4.995563180312352e-05, "loss": 2.5799, "step": 600 }, { "epoch": 0.01, "learning_rate": 4.994823710364411e-05, "loss": 2.5585, "step": 700 }, { "epoch": 0.01, "learning_rate": 4.9940842404164694e-05, "loss": 2.5208, "step": 800 }, { "epoch": 0.01, "learning_rate": 4.993344770468528e-05, "loss": 2.5201, "step": 900 }, { "epoch": 0.01, "learning_rate": 4.992605300520587e-05, "loss": 2.4521, "step": 1000 }, { "epoch": 0.02, "learning_rate": 4.991865830572646e-05, "loss": 2.4482, "step": 1100 }, { "epoch": 0.02, "learning_rate": 4.9911263606247044e-05, "loss": 2.4095, "step": 1200 }, { "epoch": 0.02, "learning_rate": 4.990386890676763e-05, "loss": 2.4293, "step": 1300 }, { "epoch": 0.02, "learning_rate": 4.989647420728822e-05, "loss": 2.4321, "step": 1400 }, { "epoch": 0.02, "learning_rate": 4.988907950780881e-05, "loss": 2.3756, "step": 1500 }, { "epoch": 0.02, "learning_rate": 4.988168480832939e-05, "loss": 2.3572, "step": 1600 }, { "epoch": 0.03, "learning_rate": 4.987429010884998e-05, "loss": 2.3661, "step": 1700 }, { "epoch": 0.03, "learning_rate": 4.9866895409370564e-05, "loss": 2.3842, "step": 1800 }, { "epoch": 0.03, "learning_rate": 4.9859500709891157e-05, "loss": 2.3186, "step": 1900 }, { "epoch": 0.03, "learning_rate": 4.9852106010411735e-05, "loss": 2.3267, "step": 2000 }, { "epoch": 0.03, "learning_rate": 4.984471131093232e-05, "loss": 2.3331, "step": 2100 }, { "epoch": 0.03, "learning_rate": 4.9837316611452914e-05, "loss": 2.2617, "step": 2200 }, { "epoch": 0.03, "learning_rate": 4.98299219119735e-05, "loss": 2.3182, "step": 2300 }, { "epoch": 0.04, "learning_rate": 4.9822527212494085e-05, "loss": 2.3079, "step": 2400 }, { "epoch": 0.04, "learning_rate": 4.981513251301467e-05, "loss": 2.2781, "step": 2500 }, { "epoch": 0.04, "learning_rate": 4.980773781353526e-05, "loss": 2.2769, "step": 2600 }, { "epoch": 0.04, "learning_rate": 4.980034311405585e-05, "loss": 2.3029, "step": 2700 }, { "epoch": 0.04, "learning_rate": 4.9792948414576434e-05, "loss": 2.2748, "step": 2800 }, { "epoch": 0.04, "learning_rate": 4.978555371509702e-05, "loss": 2.2387, "step": 2900 }, { "epoch": 0.04, "learning_rate": 4.977815901561761e-05, "loss": 2.2111, "step": 3000 }, { "epoch": 0.05, "learning_rate": 4.97707643161382e-05, "loss": 2.2477, "step": 3100 }, { "epoch": 0.05, "learning_rate": 4.9763369616658783e-05, "loss": 2.229, "step": 3200 }, { "epoch": 0.05, "learning_rate": 4.975597491717937e-05, "loss": 2.256, "step": 3300 }, { "epoch": 0.05, "learning_rate": 4.9748580217699955e-05, "loss": 2.2034, "step": 3400 }, { "epoch": 0.05, "learning_rate": 4.974118551822054e-05, "loss": 2.241, "step": 3500 }, { "epoch": 0.05, "learning_rate": 4.9733790818741126e-05, "loss": 2.2128, "step": 3600 }, { "epoch": 0.05, "learning_rate": 4.972639611926171e-05, "loss": 2.2185, "step": 3700 }, { "epoch": 0.06, "learning_rate": 4.9719001419782304e-05, "loss": 2.2213, "step": 3800 }, { "epoch": 0.06, "learning_rate": 4.971160672030289e-05, "loss": 2.1884, "step": 3900 }, { "epoch": 0.06, "learning_rate": 4.9704212020823475e-05, "loss": 2.1957, "step": 4000 }, { "epoch": 0.06, "learning_rate": 4.969681732134406e-05, "loss": 2.204, "step": 4100 }, { "epoch": 0.06, "learning_rate": 4.968949656885945e-05, "loss": 2.1919, "step": 4200 }, { "epoch": 0.06, "learning_rate": 4.9682101869380034e-05, "loss": 2.1941, "step": 4300 }, { "epoch": 0.07, "learning_rate": 4.967470716990062e-05, "loss": 2.1981, "step": 4400 }, { "epoch": 0.07, "learning_rate": 4.9667312470421205e-05, "loss": 2.1546, "step": 4500 }, { "epoch": 0.07, "learning_rate": 4.965991777094179e-05, "loss": 2.1156, "step": 4600 }, { "epoch": 0.07, "learning_rate": 4.9652523071462376e-05, "loss": 2.142, "step": 4700 }, { "epoch": 0.07, "learning_rate": 4.964512837198296e-05, "loss": 2.1557, "step": 4800 }, { "epoch": 0.07, "learning_rate": 4.963773367250355e-05, "loss": 2.1273, "step": 4900 }, { "epoch": 0.07, "learning_rate": 4.963033897302414e-05, "loss": 2.1309, "step": 5000 }, { "epoch": 0.08, "learning_rate": 4.9622944273544726e-05, "loss": 2.1652, "step": 5100 }, { "epoch": 0.08, "learning_rate": 4.961554957406531e-05, "loss": 2.149, "step": 5200 }, { "epoch": 0.08, "learning_rate": 4.96081548745859e-05, "loss": 2.1456, "step": 5300 }, { "epoch": 0.08, "learning_rate": 4.960076017510649e-05, "loss": 2.1594, "step": 5400 }, { "epoch": 0.08, "learning_rate": 4.9593365475627075e-05, "loss": 2.1398, "step": 5500 }, { "epoch": 0.08, "learning_rate": 4.958597077614766e-05, "loss": 2.1148, "step": 5600 }, { "epoch": 0.08, "learning_rate": 4.9578576076668246e-05, "loss": 2.1583, "step": 5700 }, { "epoch": 0.09, "learning_rate": 4.957118137718884e-05, "loss": 2.1114, "step": 5800 }, { "epoch": 0.09, "learning_rate": 4.9563786677709424e-05, "loss": 2.1155, "step": 5900 }, { "epoch": 0.09, "learning_rate": 4.955639197823e-05, "loss": 2.1098, "step": 6000 }, { "epoch": 0.09, "learning_rate": 4.954899727875059e-05, "loss": 2.0925, "step": 6100 }, { "epoch": 0.09, "learning_rate": 4.9541676526265976e-05, "loss": 2.1266, "step": 6200 }, { "epoch": 0.09, "learning_rate": 4.953428182678656e-05, "loss": 2.1477, "step": 6300 }, { "epoch": 0.09, "learning_rate": 4.952688712730715e-05, "loss": 2.1214, "step": 6400 }, { "epoch": 0.1, "learning_rate": 4.951949242782773e-05, "loss": 2.093, "step": 6500 }, { "epoch": 0.1, "learning_rate": 4.9512097728348325e-05, "loss": 2.0955, "step": 6600 }, { "epoch": 0.1, "learning_rate": 4.950470302886891e-05, "loss": 2.0927, "step": 6700 }, { "epoch": 0.1, "learning_rate": 4.94973083293895e-05, "loss": 2.0823, "step": 6800 }, { "epoch": 0.1, "learning_rate": 4.948991362991008e-05, "loss": 2.0801, "step": 6900 }, { "epoch": 0.1, "learning_rate": 4.948251893043067e-05, "loss": 2.1013, "step": 7000 }, { "epoch": 0.11, "learning_rate": 4.947512423095126e-05, "loss": 2.0966, "step": 7100 }, { "epoch": 0.11, "learning_rate": 4.9467729531471846e-05, "loss": 2.0613, "step": 7200 }, { "epoch": 0.11, "learning_rate": 4.9460334831992425e-05, "loss": 2.063, "step": 7300 }, { "epoch": 0.11, "learning_rate": 4.945294013251302e-05, "loss": 2.0941, "step": 7400 }, { "epoch": 0.11, "learning_rate": 4.94455454330336e-05, "loss": 2.0996, "step": 7500 }, { "epoch": 0.11, "learning_rate": 4.943815073355419e-05, "loss": 2.0831, "step": 7600 }, { "epoch": 0.11, "learning_rate": 4.9430756034074774e-05, "loss": 2.0941, "step": 7700 }, { "epoch": 0.12, "learning_rate": 4.942336133459537e-05, "loss": 2.1278, "step": 7800 }, { "epoch": 0.12, "learning_rate": 4.941596663511595e-05, "loss": 2.0738, "step": 7900 }, { "epoch": 0.12, "learning_rate": 4.940857193563654e-05, "loss": 2.0666, "step": 8000 }, { "epoch": 0.12, "learning_rate": 4.9401177236157124e-05, "loss": 2.0527, "step": 8100 }, { "epoch": 0.12, "learning_rate": 4.9393782536677716e-05, "loss": 2.0532, "step": 8200 }, { "epoch": 0.12, "learning_rate": 4.93863878371983e-05, "loss": 2.0297, "step": 8300 }, { "epoch": 0.12, "learning_rate": 4.937899313771889e-05, "loss": 2.0484, "step": 8400 }, { "epoch": 0.13, "learning_rate": 4.937159843823947e-05, "loss": 2.0632, "step": 8500 }, { "epoch": 0.13, "learning_rate": 4.936420373876006e-05, "loss": 2.0324, "step": 8600 }, { "epoch": 0.13, "learning_rate": 4.935688298627544e-05, "loss": 2.0349, "step": 8700 }, { "epoch": 0.13, "learning_rate": 4.9349488286796025e-05, "loss": 2.0879, "step": 8800 }, { "epoch": 0.13, "learning_rate": 4.934209358731661e-05, "loss": 2.0388, "step": 8900 }, { "epoch": 0.13, "learning_rate": 4.93346988878372e-05, "loss": 2.0342, "step": 9000 }, { "epoch": 0.13, "learning_rate": 4.932730418835779e-05, "loss": 2.0462, "step": 9100 }, { "epoch": 0.14, "learning_rate": 4.9319909488878374e-05, "loss": 2.0469, "step": 9200 }, { "epoch": 0.14, "learning_rate": 4.931251478939896e-05, "loss": 2.0628, "step": 9300 }, { "epoch": 0.14, "learning_rate": 4.930512008991955e-05, "loss": 2.0183, "step": 9400 }, { "epoch": 0.14, "learning_rate": 4.929772539044014e-05, "loss": 2.0261, "step": 9500 }, { "epoch": 0.14, "learning_rate": 4.929033069096072e-05, "loss": 2.0298, "step": 9600 }, { "epoch": 0.14, "learning_rate": 4.928293599148131e-05, "loss": 2.0427, "step": 9700 }, { "epoch": 0.14, "learning_rate": 4.9275541292001895e-05, "loss": 2.0566, "step": 9800 }, { "epoch": 0.15, "learning_rate": 4.926814659252249e-05, "loss": 1.998, "step": 9900 }, { "epoch": 0.15, "learning_rate": 4.9260751893043066e-05, "loss": 2.0508, "step": 10000 }, { "epoch": 0.15, "learning_rate": 4.925335719356365e-05, "loss": 1.9931, "step": 10100 }, { "epoch": 0.15, "learning_rate": 4.9245962494084244e-05, "loss": 2.04, "step": 10200 }, { "epoch": 0.15, "learning_rate": 4.923856779460483e-05, "loss": 1.9981, "step": 10300 }, { "epoch": 0.15, "learning_rate": 4.9231173095125415e-05, "loss": 2.0008, "step": 10400 }, { "epoch": 0.16, "learning_rate": 4.9223778395646e-05, "loss": 1.9967, "step": 10500 }, { "epoch": 0.16, "learning_rate": 4.921638369616659e-05, "loss": 2.013, "step": 10600 }, { "epoch": 0.16, "learning_rate": 4.920898899668718e-05, "loss": 2.0167, "step": 10700 }, { "epoch": 0.16, "learning_rate": 4.9201594297207764e-05, "loss": 2.012, "step": 10800 }, { "epoch": 0.16, "learning_rate": 4.919419959772835e-05, "loss": 1.9876, "step": 10900 }, { "epoch": 0.16, "learning_rate": 4.918680489824894e-05, "loss": 2.0142, "step": 11000 }, { "epoch": 0.16, "learning_rate": 4.917941019876953e-05, "loss": 1.9965, "step": 11100 }, { "epoch": 0.17, "learning_rate": 4.917208944628491e-05, "loss": 1.9778, "step": 11200 }, { "epoch": 0.17, "learning_rate": 4.916469474680549e-05, "loss": 2.0163, "step": 11300 }, { "epoch": 0.17, "learning_rate": 4.915730004732608e-05, "loss": 2.0398, "step": 11400 }, { "epoch": 0.17, "learning_rate": 4.9149905347846666e-05, "loss": 2.0177, "step": 11500 }, { "epoch": 0.17, "learning_rate": 4.914251064836725e-05, "loss": 1.9847, "step": 11600 }, { "epoch": 0.17, "learning_rate": 4.913511594888784e-05, "loss": 1.9609, "step": 11700 }, { "epoch": 0.17, "learning_rate": 4.912772124940843e-05, "loss": 1.9641, "step": 11800 }, { "epoch": 0.18, "learning_rate": 4.9120326549929015e-05, "loss": 2.0005, "step": 11900 }, { "epoch": 0.18, "learning_rate": 4.91129318504496e-05, "loss": 1.9991, "step": 12000 }, { "epoch": 0.18, "learning_rate": 4.9105537150970186e-05, "loss": 1.9912, "step": 12100 }, { "epoch": 0.18, "learning_rate": 4.909814245149077e-05, "loss": 1.987, "step": 12200 }, { "epoch": 0.18, "learning_rate": 4.9090747752011364e-05, "loss": 1.9887, "step": 12300 }, { "epoch": 0.18, "learning_rate": 4.908335305253195e-05, "loss": 1.9956, "step": 12400 }, { "epoch": 0.18, "learning_rate": 4.9075958353052535e-05, "loss": 1.9633, "step": 12500 }, { "epoch": 0.19, "learning_rate": 4.906856365357312e-05, "loss": 1.9802, "step": 12600 }, { "epoch": 0.19, "learning_rate": 4.906116895409371e-05, "loss": 2.0055, "step": 12700 }, { "epoch": 0.19, "learning_rate": 4.905377425461429e-05, "loss": 1.9626, "step": 12800 }, { "epoch": 0.19, "learning_rate": 4.904637955513488e-05, "loss": 1.9506, "step": 12900 }, { "epoch": 0.19, "learning_rate": 4.903898485565547e-05, "loss": 1.963, "step": 13000 }, { "epoch": 0.19, "learning_rate": 4.9031590156176056e-05, "loss": 1.9647, "step": 13100 }, { "epoch": 0.2, "learning_rate": 4.9024269403691436e-05, "loss": 1.9834, "step": 13200 }, { "epoch": 0.2, "learning_rate": 4.901687470421202e-05, "loss": 1.9763, "step": 13300 }, { "epoch": 0.2, "learning_rate": 4.900948000473261e-05, "loss": 1.971, "step": 13400 }, { "epoch": 0.2, "learning_rate": 4.90020853052532e-05, "loss": 1.9911, "step": 13500 }, { "epoch": 0.2, "learning_rate": 4.8994690605773786e-05, "loss": 1.9565, "step": 13600 }, { "epoch": 0.2, "learning_rate": 4.898729590629437e-05, "loss": 1.9361, "step": 13700 }, { "epoch": 0.2, "learning_rate": 4.897990120681496e-05, "loss": 1.9895, "step": 13800 }, { "epoch": 0.21, "learning_rate": 4.897250650733554e-05, "loss": 1.9424, "step": 13900 }, { "epoch": 0.21, "learning_rate": 4.896511180785613e-05, "loss": 1.9614, "step": 14000 }, { "epoch": 0.21, "learning_rate": 4.8957717108376714e-05, "loss": 1.9662, "step": 14100 }, { "epoch": 0.21, "learning_rate": 4.8950322408897306e-05, "loss": 2.0129, "step": 14200 }, { "epoch": 0.21, "learning_rate": 4.894292770941789e-05, "loss": 1.9652, "step": 14300 }, { "epoch": 0.21, "learning_rate": 4.893553300993848e-05, "loss": 1.9475, "step": 14400 }, { "epoch": 0.21, "learning_rate": 4.892813831045906e-05, "loss": 1.9473, "step": 14500 }, { "epoch": 0.22, "learning_rate": 4.8920743610979656e-05, "loss": 1.944, "step": 14600 }, { "epoch": 0.22, "learning_rate": 4.891334891150024e-05, "loss": 1.9409, "step": 14700 }, { "epoch": 0.22, "learning_rate": 4.890595421202083e-05, "loss": 1.949, "step": 14800 }, { "epoch": 0.22, "learning_rate": 4.889855951254141e-05, "loss": 1.9782, "step": 14900 }, { "epoch": 0.22, "learning_rate": 4.8891164813062e-05, "loss": 1.9485, "step": 15000 }, { "epoch": 0.22, "learning_rate": 4.888377011358259e-05, "loss": 1.9484, "step": 15100 }, { "epoch": 0.22, "learning_rate": 4.887637541410317e-05, "loss": 1.9551, "step": 15200 }, { "epoch": 0.23, "learning_rate": 4.8868980714623755e-05, "loss": 1.9261, "step": 15300 }, { "epoch": 0.23, "learning_rate": 4.886158601514435e-05, "loss": 1.9625, "step": 15400 }, { "epoch": 0.23, "learning_rate": 4.885419131566493e-05, "loss": 1.9619, "step": 15500 }, { "epoch": 0.23, "learning_rate": 4.884679661618552e-05, "loss": 1.9496, "step": 15600 }, { "epoch": 0.23, "learning_rate": 4.8839401916706105e-05, "loss": 1.975, "step": 15700 }, { "epoch": 0.23, "learning_rate": 4.88320072172267e-05, "loss": 1.9324, "step": 15800 }, { "epoch": 0.24, "learning_rate": 4.882461251774728e-05, "loss": 1.9643, "step": 15900 }, { "epoch": 0.24, "learning_rate": 4.881721781826787e-05, "loss": 1.9366, "step": 16000 }, { "epoch": 0.24, "learning_rate": 4.8809823118788454e-05, "loss": 1.934, "step": 16100 }, { "epoch": 0.24, "learning_rate": 4.8802502366303834e-05, "loss": 1.9236, "step": 16200 }, { "epoch": 0.24, "learning_rate": 4.879510766682443e-05, "loss": 1.9284, "step": 16300 }, { "epoch": 0.24, "learning_rate": 4.878771296734501e-05, "loss": 1.9677, "step": 16400 }, { "epoch": 0.24, "learning_rate": 4.87803182678656e-05, "loss": 1.9315, "step": 16500 }, { "epoch": 0.25, "learning_rate": 4.8772923568386184e-05, "loss": 1.9345, "step": 16600 }, { "epoch": 0.25, "learning_rate": 4.876552886890677e-05, "loss": 1.9373, "step": 16700 }, { "epoch": 0.25, "learning_rate": 4.8758134169427355e-05, "loss": 1.932, "step": 16800 }, { "epoch": 0.25, "learning_rate": 4.875073946994794e-05, "loss": 1.9039, "step": 16900 }, { "epoch": 0.25, "learning_rate": 4.874334477046853e-05, "loss": 1.9559, "step": 17000 }, { "epoch": 0.25, "learning_rate": 4.873595007098912e-05, "loss": 1.9104, "step": 17100 }, { "epoch": 0.25, "learning_rate": 4.8728555371509704e-05, "loss": 1.9314, "step": 17200 }, { "epoch": 0.26, "learning_rate": 4.872116067203029e-05, "loss": 1.9595, "step": 17300 }, { "epoch": 0.26, "learning_rate": 4.8713765972550876e-05, "loss": 1.9264, "step": 17400 }, { "epoch": 0.26, "learning_rate": 4.870637127307147e-05, "loss": 1.8972, "step": 17500 }, { "epoch": 0.26, "learning_rate": 4.8698976573592054e-05, "loss": 1.9434, "step": 17600 }, { "epoch": 0.26, "learning_rate": 4.869158187411264e-05, "loss": 1.9297, "step": 17700 }, { "epoch": 0.26, "learning_rate": 4.8684187174633225e-05, "loss": 1.9031, "step": 17800 }, { "epoch": 0.26, "learning_rate": 4.867679247515381e-05, "loss": 1.9358, "step": 17900 }, { "epoch": 0.27, "learning_rate": 4.8669397775674396e-05, "loss": 1.8736, "step": 18000 }, { "epoch": 0.27, "learning_rate": 4.866200307619498e-05, "loss": 1.9253, "step": 18100 }, { "epoch": 0.27, "learning_rate": 4.8654608376715574e-05, "loss": 1.9201, "step": 18200 }, { "epoch": 0.27, "learning_rate": 4.864721367723616e-05, "loss": 1.9401, "step": 18300 }, { "epoch": 0.27, "learning_rate": 4.8639818977756745e-05, "loss": 1.9467, "step": 18400 }, { "epoch": 0.27, "learning_rate": 4.8632498225272126e-05, "loss": 1.9439, "step": 18500 }, { "epoch": 0.28, "learning_rate": 4.862510352579271e-05, "loss": 1.9153, "step": 18600 }, { "epoch": 0.28, "learning_rate": 4.8617708826313304e-05, "loss": 1.9149, "step": 18700 }, { "epoch": 0.28, "learning_rate": 4.861031412683389e-05, "loss": 1.9285, "step": 18800 }, { "epoch": 0.28, "learning_rate": 4.8602919427354475e-05, "loss": 1.9138, "step": 18900 }, { "epoch": 0.28, "learning_rate": 4.859552472787506e-05, "loss": 1.9528, "step": 19000 }, { "epoch": 0.28, "learning_rate": 4.858813002839565e-05, "loss": 1.8888, "step": 19100 }, { "epoch": 0.28, "learning_rate": 4.858073532891623e-05, "loss": 1.9066, "step": 19200 }, { "epoch": 0.29, "learning_rate": 4.857334062943682e-05, "loss": 1.932, "step": 19300 }, { "epoch": 0.29, "learning_rate": 4.856594592995741e-05, "loss": 1.9112, "step": 19400 }, { "epoch": 0.29, "learning_rate": 4.8558551230477996e-05, "loss": 1.9011, "step": 19500 }, { "epoch": 0.29, "learning_rate": 4.855115653099858e-05, "loss": 1.9064, "step": 19600 }, { "epoch": 0.29, "learning_rate": 4.854376183151917e-05, "loss": 1.9445, "step": 19700 }, { "epoch": 0.29, "learning_rate": 4.853636713203976e-05, "loss": 1.9436, "step": 19800 }, { "epoch": 0.29, "learning_rate": 4.8528972432560345e-05, "loss": 1.8964, "step": 19900 }, { "epoch": 0.3, "learning_rate": 4.852157773308093e-05, "loss": 1.8949, "step": 20000 }, { "epoch": 0.3, "learning_rate": 4.8514183033601516e-05, "loss": 1.9179, "step": 20100 }, { "epoch": 0.3, "learning_rate": 4.85067883341221e-05, "loss": 1.9006, "step": 20200 }, { "epoch": 0.3, "learning_rate": 4.8499393634642694e-05, "loss": 1.8683, "step": 20300 }, { "epoch": 0.3, "learning_rate": 4.849199893516328e-05, "loss": 1.8887, "step": 20400 }, { "epoch": 0.3, "learning_rate": 4.848460423568386e-05, "loss": 1.8973, "step": 20500 }, { "epoch": 0.3, "learning_rate": 4.847720953620445e-05, "loss": 1.8831, "step": 20600 }, { "epoch": 0.31, "learning_rate": 4.846981483672504e-05, "loss": 1.8713, "step": 20700 }, { "epoch": 0.31, "learning_rate": 4.846242013724562e-05, "loss": 1.9004, "step": 20800 }, { "epoch": 0.31, "learning_rate": 4.845502543776621e-05, "loss": 1.8996, "step": 20900 }, { "epoch": 0.31, "learning_rate": 4.84476307382868e-05, "loss": 1.9057, "step": 21000 }, { "epoch": 0.31, "learning_rate": 4.8440236038807386e-05, "loss": 1.9165, "step": 21100 }, { "epoch": 0.31, "learning_rate": 4.843284133932797e-05, "loss": 1.885, "step": 21200 }, { "epoch": 0.32, "learning_rate": 4.842544663984856e-05, "loss": 1.8674, "step": 21300 }, { "epoch": 0.32, "learning_rate": 4.841812588736394e-05, "loss": 1.9026, "step": 21400 }, { "epoch": 0.32, "learning_rate": 4.841073118788453e-05, "loss": 1.8985, "step": 21500 }, { "epoch": 0.32, "learning_rate": 4.8403336488405116e-05, "loss": 1.8923, "step": 21600 }, { "epoch": 0.32, "learning_rate": 4.83959417889257e-05, "loss": 1.8978, "step": 21700 }, { "epoch": 0.32, "learning_rate": 4.838854708944629e-05, "loss": 1.8871, "step": 21800 }, { "epoch": 0.32, "learning_rate": 4.838115238996687e-05, "loss": 1.8718, "step": 21900 }, { "epoch": 0.33, "learning_rate": 4.837375769048746e-05, "loss": 1.9119, "step": 22000 }, { "epoch": 0.33, "learning_rate": 4.8366362991008044e-05, "loss": 1.8625, "step": 22100 }, { "epoch": 0.33, "learning_rate": 4.835896829152864e-05, "loss": 1.8834, "step": 22200 }, { "epoch": 0.33, "learning_rate": 4.835157359204922e-05, "loss": 1.8734, "step": 22300 }, { "epoch": 0.33, "learning_rate": 4.834417889256981e-05, "loss": 1.8839, "step": 22400 }, { "epoch": 0.33, "learning_rate": 4.8336784193090394e-05, "loss": 1.8718, "step": 22500 }, { "epoch": 0.33, "learning_rate": 4.832938949361098e-05, "loss": 1.9251, "step": 22600 }, { "epoch": 0.34, "learning_rate": 4.832199479413157e-05, "loss": 1.8665, "step": 22700 }, { "epoch": 0.34, "learning_rate": 4.831460009465216e-05, "loss": 1.8725, "step": 22800 }, { "epoch": 0.34, "learning_rate": 4.830720539517274e-05, "loss": 1.881, "step": 22900 }, { "epoch": 0.34, "learning_rate": 4.829981069569333e-05, "loss": 1.875, "step": 23000 }, { "epoch": 0.34, "learning_rate": 4.829241599621392e-05, "loss": 1.8611, "step": 23100 }, { "epoch": 0.34, "learning_rate": 4.82850212967345e-05, "loss": 1.8724, "step": 23200 }, { "epoch": 0.34, "learning_rate": 4.8277626597255086e-05, "loss": 1.8807, "step": 23300 }, { "epoch": 0.35, "learning_rate": 4.827023189777568e-05, "loss": 1.8552, "step": 23400 }, { "epoch": 0.35, "learning_rate": 4.8262837198296264e-05, "loss": 1.853, "step": 23500 }, { "epoch": 0.35, "learning_rate": 4.825544249881685e-05, "loss": 1.8351, "step": 23600 }, { "epoch": 0.35, "learning_rate": 4.8248047799337435e-05, "loss": 1.875, "step": 23700 }, { "epoch": 0.35, "learning_rate": 4.824065309985803e-05, "loss": 1.8807, "step": 23800 }, { "epoch": 0.35, "learning_rate": 4.823325840037861e-05, "loss": 1.8734, "step": 23900 }, { "epoch": 0.35, "learning_rate": 4.82258637008992e-05, "loss": 1.8642, "step": 24000 }, { "epoch": 0.36, "learning_rate": 4.8218469001419784e-05, "loss": 1.8599, "step": 24100 }, { "epoch": 0.36, "learning_rate": 4.8211074301940377e-05, "loss": 1.8656, "step": 24200 }, { "epoch": 0.36, "learning_rate": 4.820367960246096e-05, "loss": 1.8629, "step": 24300 }, { "epoch": 0.36, "learning_rate": 4.819628490298155e-05, "loss": 1.8598, "step": 24400 }, { "epoch": 0.36, "learning_rate": 4.818889020350213e-05, "loss": 1.8552, "step": 24500 }, { "epoch": 0.36, "learning_rate": 4.818149550402272e-05, "loss": 1.8565, "step": 24600 }, { "epoch": 0.37, "learning_rate": 4.81741747515381e-05, "loss": 1.8701, "step": 24700 }, { "epoch": 0.37, "learning_rate": 4.8166780052058685e-05, "loss": 1.8765, "step": 24800 }, { "epoch": 0.37, "learning_rate": 4.815938535257927e-05, "loss": 1.8726, "step": 24900 }, { "epoch": 0.37, "learning_rate": 4.815199065309986e-05, "loss": 1.8706, "step": 25000 }, { "epoch": 0.37, "learning_rate": 4.814459595362045e-05, "loss": 1.873, "step": 25100 }, { "epoch": 0.37, "learning_rate": 4.8137201254141035e-05, "loss": 1.8408, "step": 25200 }, { "epoch": 0.37, "learning_rate": 4.812980655466162e-05, "loss": 1.8744, "step": 25300 }, { "epoch": 0.38, "learning_rate": 4.8122411855182206e-05, "loss": 1.885, "step": 25400 }, { "epoch": 0.38, "learning_rate": 4.81150171557028e-05, "loss": 1.8794, "step": 25500 }, { "epoch": 0.38, "learning_rate": 4.8107622456223384e-05, "loss": 1.842, "step": 25600 }, { "epoch": 0.38, "learning_rate": 4.810022775674397e-05, "loss": 1.8753, "step": 25700 }, { "epoch": 0.38, "learning_rate": 4.8092833057264555e-05, "loss": 1.8326, "step": 25800 }, { "epoch": 0.38, "learning_rate": 4.808543835778514e-05, "loss": 1.8327, "step": 25900 }, { "epoch": 0.38, "learning_rate": 4.8078043658305726e-05, "loss": 1.8257, "step": 26000 }, { "epoch": 0.39, "learning_rate": 4.807064895882631e-05, "loss": 1.8652, "step": 26100 }, { "epoch": 0.39, "learning_rate": 4.8063254259346904e-05, "loss": 1.8509, "step": 26200 }, { "epoch": 0.39, "learning_rate": 4.805585955986749e-05, "loss": 1.8735, "step": 26300 }, { "epoch": 0.39, "learning_rate": 4.8048464860388076e-05, "loss": 1.8531, "step": 26400 }, { "epoch": 0.39, "learning_rate": 4.804107016090866e-05, "loss": 1.8617, "step": 26500 }, { "epoch": 0.39, "learning_rate": 4.8033675461429254e-05, "loss": 1.8583, "step": 26600 }, { "epoch": 0.39, "learning_rate": 4.802628076194984e-05, "loss": 1.8491, "step": 26700 }, { "epoch": 0.4, "learning_rate": 4.8018886062470425e-05, "loss": 1.8717, "step": 26800 }, { "epoch": 0.4, "learning_rate": 4.8011565309985806e-05, "loss": 1.8676, "step": 26900 }, { "epoch": 0.4, "learning_rate": 4.800417061050639e-05, "loss": 1.8635, "step": 27000 }, { "epoch": 0.4, "learning_rate": 4.799677591102698e-05, "loss": 1.8291, "step": 27100 }, { "epoch": 0.4, "learning_rate": 4.798938121154756e-05, "loss": 1.859, "step": 27200 }, { "epoch": 0.4, "learning_rate": 4.798198651206815e-05, "loss": 1.8158, "step": 27300 }, { "epoch": 0.41, "learning_rate": 4.797459181258874e-05, "loss": 1.8652, "step": 27400 }, { "epoch": 0.41, "learning_rate": 4.7967197113109326e-05, "loss": 1.8637, "step": 27500 }, { "epoch": 0.41, "learning_rate": 4.795980241362991e-05, "loss": 1.8458, "step": 27600 }, { "epoch": 0.41, "learning_rate": 4.79524077141505e-05, "loss": 1.8086, "step": 27700 }, { "epoch": 0.41, "learning_rate": 4.794501301467108e-05, "loss": 1.8499, "step": 27800 }, { "epoch": 0.41, "learning_rate": 4.7937618315191675e-05, "loss": 1.879, "step": 27900 }, { "epoch": 0.41, "learning_rate": 4.793022361571226e-05, "loss": 1.8575, "step": 28000 }, { "epoch": 0.42, "learning_rate": 4.792282891623285e-05, "loss": 1.8326, "step": 28100 }, { "epoch": 0.42, "learning_rate": 4.791543421675343e-05, "loss": 1.875, "step": 28200 }, { "epoch": 0.42, "learning_rate": 4.7908039517274025e-05, "loss": 1.8682, "step": 28300 }, { "epoch": 0.42, "learning_rate": 4.790064481779461e-05, "loss": 1.8373, "step": 28400 }, { "epoch": 0.42, "learning_rate": 4.789325011831519e-05, "loss": 1.7958, "step": 28500 }, { "epoch": 0.42, "learning_rate": 4.788585541883578e-05, "loss": 1.8596, "step": 28600 }, { "epoch": 0.42, "learning_rate": 4.787846071935637e-05, "loss": 1.8747, "step": 28700 }, { "epoch": 0.43, "learning_rate": 4.787106601987695e-05, "loss": 1.8629, "step": 28800 }, { "epoch": 0.43, "learning_rate": 4.786367132039754e-05, "loss": 1.8307, "step": 28900 }, { "epoch": 0.43, "learning_rate": 4.785627662091813e-05, "loss": 1.8578, "step": 29000 }, { "epoch": 0.43, "learning_rate": 4.784888192143872e-05, "loss": 1.8484, "step": 29100 }, { "epoch": 0.43, "learning_rate": 4.78414872219593e-05, "loss": 1.8428, "step": 29200 }, { "epoch": 0.43, "learning_rate": 4.783409252247989e-05, "loss": 1.8385, "step": 29300 }, { "epoch": 0.43, "learning_rate": 4.7826697823000474e-05, "loss": 1.8431, "step": 29400 }, { "epoch": 0.44, "learning_rate": 4.7819303123521066e-05, "loss": 1.8292, "step": 29500 }, { "epoch": 0.44, "learning_rate": 4.781190842404165e-05, "loss": 1.8305, "step": 29600 }, { "epoch": 0.44, "learning_rate": 4.780451372456224e-05, "loss": 1.8412, "step": 29700 }, { "epoch": 0.44, "learning_rate": 4.779711902508282e-05, "loss": 1.8244, "step": 29800 }, { "epoch": 0.44, "learning_rate": 4.778972432560341e-05, "loss": 1.8344, "step": 29900 }, { "epoch": 0.44, "learning_rate": 4.7782329626123994e-05, "loss": 1.8502, "step": 30000 }, { "epoch": 0.45, "learning_rate": 4.777493492664458e-05, "loss": 1.8605, "step": 30100 }, { "epoch": 0.45, "learning_rate": 4.776754022716517e-05, "loss": 1.8364, "step": 30200 }, { "epoch": 0.45, "learning_rate": 4.776021947468055e-05, "loss": 1.8621, "step": 30300 }, { "epoch": 0.45, "learning_rate": 4.775282477520114e-05, "loss": 1.8395, "step": 30400 }, { "epoch": 0.45, "learning_rate": 4.7745430075721724e-05, "loss": 1.7994, "step": 30500 }, { "epoch": 0.45, "learning_rate": 4.773803537624231e-05, "loss": 1.8275, "step": 30600 }, { "epoch": 0.45, "learning_rate": 4.77306406767629e-05, "loss": 1.8363, "step": 30700 }, { "epoch": 0.46, "learning_rate": 4.772324597728349e-05, "loss": 1.8223, "step": 30800 }, { "epoch": 0.46, "learning_rate": 4.771585127780407e-05, "loss": 1.8288, "step": 30900 }, { "epoch": 0.46, "learning_rate": 4.770845657832466e-05, "loss": 1.8547, "step": 31000 }, { "epoch": 0.46, "learning_rate": 4.7701061878845245e-05, "loss": 1.8582, "step": 31100 }, { "epoch": 0.46, "learning_rate": 4.769366717936583e-05, "loss": 1.8086, "step": 31200 }, { "epoch": 0.46, "learning_rate": 4.7686272479886416e-05, "loss": 1.8162, "step": 31300 }, { "epoch": 0.46, "learning_rate": 4.767887778040701e-05, "loss": 1.8085, "step": 31400 }, { "epoch": 0.47, "learning_rate": 4.7671483080927594e-05, "loss": 1.8304, "step": 31500 }, { "epoch": 0.47, "learning_rate": 4.766408838144818e-05, "loss": 1.8654, "step": 31600 }, { "epoch": 0.47, "learning_rate": 4.7656693681968765e-05, "loss": 1.8079, "step": 31700 }, { "epoch": 0.47, "learning_rate": 4.764929898248936e-05, "loss": 1.8224, "step": 31800 }, { "epoch": 0.47, "learning_rate": 4.764190428300994e-05, "loss": 1.8346, "step": 31900 }, { "epoch": 0.47, "learning_rate": 4.763450958353053e-05, "loss": 1.8433, "step": 32000 }, { "epoch": 0.47, "learning_rate": 4.7627114884051115e-05, "loss": 1.8315, "step": 32100 }, { "epoch": 0.48, "learning_rate": 4.76197201845717e-05, "loss": 1.8166, "step": 32200 }, { "epoch": 0.48, "learning_rate": 4.761232548509229e-05, "loss": 1.8255, "step": 32300 }, { "epoch": 0.48, "learning_rate": 4.760493078561287e-05, "loss": 1.7861, "step": 32400 }, { "epoch": 0.48, "learning_rate": 4.759753608613346e-05, "loss": 1.8063, "step": 32500 }, { "epoch": 0.48, "learning_rate": 4.759014138665405e-05, "loss": 1.8162, "step": 32600 }, { "epoch": 0.48, "learning_rate": 4.758282063416943e-05, "loss": 1.8133, "step": 32700 }, { "epoch": 0.49, "learning_rate": 4.7575425934690016e-05, "loss": 1.8252, "step": 32800 }, { "epoch": 0.49, "learning_rate": 4.75680312352106e-05, "loss": 1.8241, "step": 32900 }, { "epoch": 0.49, "learning_rate": 4.756063653573119e-05, "loss": 1.8034, "step": 33000 }, { "epoch": 0.49, "learning_rate": 4.755324183625178e-05, "loss": 1.8063, "step": 33100 }, { "epoch": 0.49, "learning_rate": 4.7545847136772365e-05, "loss": 1.846, "step": 33200 }, { "epoch": 0.49, "learning_rate": 4.753845243729295e-05, "loss": 1.8431, "step": 33300 }, { "epoch": 0.49, "learning_rate": 4.7531057737813536e-05, "loss": 1.8001, "step": 33400 }, { "epoch": 0.5, "learning_rate": 4.752366303833413e-05, "loss": 1.8041, "step": 33500 }, { "epoch": 0.5, "learning_rate": 4.7516268338854714e-05, "loss": 1.8031, "step": 33600 }, { "epoch": 0.5, "learning_rate": 4.750887363937529e-05, "loss": 1.8613, "step": 33700 }, { "epoch": 0.5, "learning_rate": 4.7501478939895885e-05, "loss": 1.8254, "step": 33800 }, { "epoch": 0.5, "learning_rate": 4.749408424041647e-05, "loss": 1.8084, "step": 33900 }, { "epoch": 0.5, "learning_rate": 4.748668954093706e-05, "loss": 1.8285, "step": 34000 }, { "epoch": 0.5, "learning_rate": 4.747929484145764e-05, "loss": 1.8077, "step": 34100 }, { "epoch": 0.51, "learning_rate": 4.7471900141978235e-05, "loss": 1.797, "step": 34200 }, { "epoch": 0.51, "learning_rate": 4.746450544249882e-05, "loss": 1.8165, "step": 34300 }, { "epoch": 0.51, "learning_rate": 4.7457110743019406e-05, "loss": 1.8253, "step": 34400 }, { "epoch": 0.51, "learning_rate": 4.744971604353999e-05, "loss": 1.8192, "step": 34500 }, { "epoch": 0.51, "learning_rate": 4.744232134406058e-05, "loss": 1.817, "step": 34600 }, { "epoch": 0.51, "learning_rate": 4.743492664458117e-05, "loss": 1.8394, "step": 34700 }, { "epoch": 0.51, "learning_rate": 4.7427531945101755e-05, "loss": 1.8201, "step": 34800 }, { "epoch": 0.52, "learning_rate": 4.742013724562234e-05, "loss": 1.8149, "step": 34900 }, { "epoch": 0.52, "learning_rate": 4.741274254614293e-05, "loss": 1.8006, "step": 35000 }, { "epoch": 0.52, "learning_rate": 4.740534784666351e-05, "loss": 1.8284, "step": 35100 }, { "epoch": 0.52, "learning_rate": 4.73979531471841e-05, "loss": 1.834, "step": 35200 }, { "epoch": 0.52, "learning_rate": 4.7390558447704684e-05, "loss": 1.8289, "step": 35300 }, { "epoch": 0.52, "learning_rate": 4.7383163748225276e-05, "loss": 1.776, "step": 35400 }, { "epoch": 0.53, "learning_rate": 4.737576904874586e-05, "loss": 1.7996, "step": 35500 }, { "epoch": 0.53, "learning_rate": 4.736837434926645e-05, "loss": 1.8297, "step": 35600 }, { "epoch": 0.53, "learning_rate": 4.736105359678183e-05, "loss": 1.8323, "step": 35700 }, { "epoch": 0.53, "learning_rate": 4.735365889730241e-05, "loss": 1.7792, "step": 35800 }, { "epoch": 0.53, "learning_rate": 4.7346264197823006e-05, "loss": 1.8141, "step": 35900 }, { "epoch": 0.53, "learning_rate": 4.733886949834359e-05, "loss": 1.8022, "step": 36000 }, { "epoch": 0.53, "learning_rate": 4.733147479886418e-05, "loss": 1.8013, "step": 36100 }, { "epoch": 0.54, "learning_rate": 4.732408009938476e-05, "loss": 1.8144, "step": 36200 }, { "epoch": 0.54, "learning_rate": 4.7316685399905355e-05, "loss": 1.8317, "step": 36300 }, { "epoch": 0.54, "learning_rate": 4.7309290700425934e-05, "loss": 1.7999, "step": 36400 }, { "epoch": 0.54, "learning_rate": 4.730189600094652e-05, "loss": 1.838, "step": 36500 }, { "epoch": 0.54, "learning_rate": 4.729450130146711e-05, "loss": 1.8051, "step": 36600 }, { "epoch": 0.54, "learning_rate": 4.72871066019877e-05, "loss": 1.828, "step": 36700 }, { "epoch": 0.54, "learning_rate": 4.727971190250828e-05, "loss": 1.7934, "step": 36800 }, { "epoch": 0.55, "learning_rate": 4.727231720302887e-05, "loss": 1.8061, "step": 36900 }, { "epoch": 0.55, "learning_rate": 4.726492250354946e-05, "loss": 1.8287, "step": 37000 }, { "epoch": 0.55, "learning_rate": 4.725752780407005e-05, "loss": 1.7928, "step": 37100 }, { "epoch": 0.55, "learning_rate": 4.725013310459063e-05, "loss": 1.797, "step": 37200 }, { "epoch": 0.55, "learning_rate": 4.724273840511122e-05, "loss": 1.8524, "step": 37300 }, { "epoch": 0.55, "learning_rate": 4.7235343705631804e-05, "loss": 1.7954, "step": 37400 }, { "epoch": 0.55, "learning_rate": 4.7227949006152396e-05, "loss": 1.8148, "step": 37500 }, { "epoch": 0.56, "learning_rate": 4.722055430667298e-05, "loss": 1.808, "step": 37600 }, { "epoch": 0.56, "learning_rate": 4.721315960719356e-05, "loss": 1.7973, "step": 37700 }, { "epoch": 0.56, "learning_rate": 4.720576490771415e-05, "loss": 1.8017, "step": 37800 }, { "epoch": 0.56, "learning_rate": 4.719837020823474e-05, "loss": 1.8122, "step": 37900 }, { "epoch": 0.56, "learning_rate": 4.7190975508755325e-05, "loss": 1.8046, "step": 38000 }, { "epoch": 0.56, "learning_rate": 4.718358080927591e-05, "loss": 1.7849, "step": 38100 }, { "epoch": 0.56, "learning_rate": 4.71761861097965e-05, "loss": 1.8101, "step": 38200 }, { "epoch": 0.57, "learning_rate": 4.716879141031709e-05, "loss": 1.7937, "step": 38300 }, { "epoch": 0.57, "learning_rate": 4.7161396710837674e-05, "loss": 1.8027, "step": 38400 }, { "epoch": 0.57, "learning_rate": 4.715400201135826e-05, "loss": 1.8289, "step": 38500 }, { "epoch": 0.57, "learning_rate": 4.714660731187885e-05, "loss": 1.7869, "step": 38600 }, { "epoch": 0.57, "learning_rate": 4.713921261239944e-05, "loss": 1.7739, "step": 38700 }, { "epoch": 0.57, "learning_rate": 4.713181791292002e-05, "loss": 1.7466, "step": 38800 }, { "epoch": 0.58, "learning_rate": 4.712442321344061e-05, "loss": 1.8165, "step": 38900 }, { "epoch": 0.58, "learning_rate": 4.7117028513961194e-05, "loss": 1.7868, "step": 39000 }, { "epoch": 0.58, "learning_rate": 4.710963381448178e-05, "loss": 1.7665, "step": 39100 }, { "epoch": 0.58, "learning_rate": 4.710231306199716e-05, "loss": 1.7795, "step": 39200 }, { "epoch": 0.58, "learning_rate": 4.7094918362517746e-05, "loss": 1.7745, "step": 39300 }, { "epoch": 0.58, "learning_rate": 4.708752366303834e-05, "loss": 1.7927, "step": 39400 }, { "epoch": 0.58, "learning_rate": 4.7080128963558924e-05, "loss": 1.7867, "step": 39500 }, { "epoch": 0.59, "learning_rate": 4.707273426407951e-05, "loss": 1.7866, "step": 39600 }, { "epoch": 0.59, "learning_rate": 4.7065339564600095e-05, "loss": 1.7803, "step": 39700 }, { "epoch": 0.59, "learning_rate": 4.705794486512068e-05, "loss": 1.7881, "step": 39800 }, { "epoch": 0.59, "learning_rate": 4.7050550165641274e-05, "loss": 1.7949, "step": 39900 }, { "epoch": 0.59, "learning_rate": 4.704315546616186e-05, "loss": 1.7869, "step": 40000 }, { "epoch": 0.59, "learning_rate": 4.7035760766682445e-05, "loss": 1.7688, "step": 40100 }, { "epoch": 0.59, "learning_rate": 4.702836606720303e-05, "loss": 1.8111, "step": 40200 }, { "epoch": 0.6, "learning_rate": 4.7020971367723616e-05, "loss": 1.7692, "step": 40300 }, { "epoch": 0.6, "learning_rate": 4.70135766682442e-05, "loss": 1.7678, "step": 40400 }, { "epoch": 0.6, "learning_rate": 4.700618196876479e-05, "loss": 1.7915, "step": 40500 }, { "epoch": 0.6, "learning_rate": 4.699878726928538e-05, "loss": 1.761, "step": 40600 }, { "epoch": 0.6, "learning_rate": 4.6991392569805965e-05, "loss": 1.7747, "step": 40700 }, { "epoch": 0.6, "learning_rate": 4.698399787032655e-05, "loss": 1.7642, "step": 40800 }, { "epoch": 0.6, "learning_rate": 4.697660317084714e-05, "loss": 1.7973, "step": 40900 }, { "epoch": 0.61, "learning_rate": 4.696920847136773e-05, "loss": 1.7942, "step": 41000 }, { "epoch": 0.61, "learning_rate": 4.6961813771888315e-05, "loss": 1.7908, "step": 41100 }, { "epoch": 0.61, "learning_rate": 4.69544190724089e-05, "loss": 1.7836, "step": 41200 }, { "epoch": 0.61, "learning_rate": 4.6947024372929486e-05, "loss": 1.803, "step": 41300 }, { "epoch": 0.61, "learning_rate": 4.693962967345008e-05, "loss": 1.7714, "step": 41400 }, { "epoch": 0.61, "learning_rate": 4.6932234973970664e-05, "loss": 1.7912, "step": 41500 }, { "epoch": 0.62, "learning_rate": 4.692484027449125e-05, "loss": 1.7617, "step": 41600 }, { "epoch": 0.62, "learning_rate": 4.691744557501183e-05, "loss": 1.7715, "step": 41700 }, { "epoch": 0.62, "learning_rate": 4.6910124822527216e-05, "loss": 1.7794, "step": 41800 }, { "epoch": 0.62, "learning_rate": 4.69027301230478e-05, "loss": 1.7629, "step": 41900 }, { "epoch": 0.62, "learning_rate": 4.689533542356839e-05, "loss": 1.7977, "step": 42000 }, { "epoch": 0.62, "learning_rate": 4.688794072408897e-05, "loss": 1.7731, "step": 42100 }, { "epoch": 0.62, "learning_rate": 4.6880546024609565e-05, "loss": 1.7539, "step": 42200 }, { "epoch": 0.63, "learning_rate": 4.687315132513015e-05, "loss": 1.788, "step": 42300 }, { "epoch": 0.63, "learning_rate": 4.6865756625650736e-05, "loss": 1.7804, "step": 42400 }, { "epoch": 0.63, "learning_rate": 4.685836192617132e-05, "loss": 1.7973, "step": 42500 }, { "epoch": 0.63, "learning_rate": 4.685096722669191e-05, "loss": 1.8139, "step": 42600 }, { "epoch": 0.63, "learning_rate": 4.68435725272125e-05, "loss": 1.7456, "step": 42700 }, { "epoch": 0.63, "learning_rate": 4.6836177827733086e-05, "loss": 1.7873, "step": 42800 }, { "epoch": 0.63, "learning_rate": 4.682878312825367e-05, "loss": 1.7842, "step": 42900 }, { "epoch": 0.64, "learning_rate": 4.682138842877426e-05, "loss": 1.7672, "step": 43000 }, { "epoch": 0.64, "learning_rate": 4.681399372929484e-05, "loss": 1.7784, "step": 43100 }, { "epoch": 0.64, "learning_rate": 4.680659902981543e-05, "loss": 1.7817, "step": 43200 }, { "epoch": 0.64, "learning_rate": 4.6799204330336014e-05, "loss": 1.7407, "step": 43300 }, { "epoch": 0.64, "learning_rate": 4.6791809630856606e-05, "loss": 1.7646, "step": 43400 }, { "epoch": 0.64, "learning_rate": 4.678441493137719e-05, "loss": 1.7657, "step": 43500 }, { "epoch": 0.64, "learning_rate": 4.677702023189778e-05, "loss": 1.7983, "step": 43600 }, { "epoch": 0.65, "learning_rate": 4.676962553241836e-05, "loss": 1.743, "step": 43700 }, { "epoch": 0.65, "learning_rate": 4.6762230832938956e-05, "loss": 1.8009, "step": 43800 }, { "epoch": 0.65, "learning_rate": 4.6754910080454336e-05, "loss": 1.8079, "step": 43900 }, { "epoch": 0.65, "learning_rate": 4.674751538097492e-05, "loss": 1.7845, "step": 44000 }, { "epoch": 0.65, "learning_rate": 4.674012068149551e-05, "loss": 1.7908, "step": 44100 }, { "epoch": 0.65, "learning_rate": 4.673272598201609e-05, "loss": 1.7645, "step": 44200 }, { "epoch": 0.66, "learning_rate": 4.672533128253668e-05, "loss": 1.792, "step": 44300 }, { "epoch": 0.66, "learning_rate": 4.6717936583057264e-05, "loss": 1.7845, "step": 44400 }, { "epoch": 0.66, "learning_rate": 4.671054188357785e-05, "loss": 1.7503, "step": 44500 }, { "epoch": 0.66, "learning_rate": 4.670314718409844e-05, "loss": 1.7454, "step": 44600 }, { "epoch": 0.66, "learning_rate": 4.669575248461903e-05, "loss": 1.7689, "step": 44700 }, { "epoch": 0.66, "learning_rate": 4.6688357785139614e-05, "loss": 1.7625, "step": 44800 }, { "epoch": 0.66, "learning_rate": 4.66809630856602e-05, "loss": 1.7851, "step": 44900 }, { "epoch": 0.67, "learning_rate": 4.6673568386180785e-05, "loss": 1.7418, "step": 45000 }, { "epoch": 0.67, "learning_rate": 4.666617368670138e-05, "loss": 1.7588, "step": 45100 }, { "epoch": 0.67, "learning_rate": 4.665877898722196e-05, "loss": 1.7847, "step": 45200 }, { "epoch": 0.67, "learning_rate": 4.665138428774255e-05, "loss": 1.7753, "step": 45300 }, { "epoch": 0.67, "learning_rate": 4.6643989588263134e-05, "loss": 1.782, "step": 45400 }, { "epoch": 0.67, "learning_rate": 4.663659488878373e-05, "loss": 1.7734, "step": 45500 }, { "epoch": 0.67, "learning_rate": 4.6629200189304306e-05, "loss": 1.765, "step": 45600 }, { "epoch": 0.68, "learning_rate": 4.662180548982489e-05, "loss": 1.7741, "step": 45700 }, { "epoch": 0.68, "learning_rate": 4.6614410790345484e-05, "loss": 1.766, "step": 45800 }, { "epoch": 0.68, "learning_rate": 4.660701609086607e-05, "loss": 1.7594, "step": 45900 }, { "epoch": 0.68, "learning_rate": 4.6599621391386655e-05, "loss": 1.781, "step": 46000 }, { "epoch": 0.68, "learning_rate": 4.659222669190724e-05, "loss": 1.7948, "step": 46100 }, { "epoch": 0.68, "learning_rate": 4.658483199242783e-05, "loss": 1.7626, "step": 46200 }, { "epoch": 0.68, "learning_rate": 4.657743729294842e-05, "loss": 1.7658, "step": 46300 }, { "epoch": 0.69, "learning_rate": 4.6570042593469004e-05, "loss": 1.7611, "step": 46400 }, { "epoch": 0.69, "learning_rate": 4.656264789398959e-05, "loss": 1.7607, "step": 46500 }, { "epoch": 0.69, "learning_rate": 4.655532714150497e-05, "loss": 1.7708, "step": 46600 }, { "epoch": 0.69, "learning_rate": 4.654793244202556e-05, "loss": 1.7282, "step": 46700 }, { "epoch": 0.69, "learning_rate": 4.654053774254615e-05, "loss": 1.7611, "step": 46800 }, { "epoch": 0.69, "learning_rate": 4.653314304306673e-05, "loss": 1.7682, "step": 46900 }, { "epoch": 0.7, "learning_rate": 4.652574834358732e-05, "loss": 1.7663, "step": 47000 }, { "epoch": 0.7, "learning_rate": 4.6518353644107905e-05, "loss": 1.7679, "step": 47100 }, { "epoch": 0.7, "learning_rate": 4.651095894462849e-05, "loss": 1.7655, "step": 47200 }, { "epoch": 0.7, "learning_rate": 4.6503564245149076e-05, "loss": 1.7753, "step": 47300 }, { "epoch": 0.7, "learning_rate": 4.649616954566967e-05, "loss": 1.7546, "step": 47400 }, { "epoch": 0.7, "learning_rate": 4.6488774846190255e-05, "loss": 1.7677, "step": 47500 }, { "epoch": 0.7, "learning_rate": 4.648138014671084e-05, "loss": 1.786, "step": 47600 }, { "epoch": 0.71, "learning_rate": 4.6473985447231426e-05, "loss": 1.7512, "step": 47700 }, { "epoch": 0.71, "learning_rate": 4.646659074775201e-05, "loss": 1.7708, "step": 47800 }, { "epoch": 0.71, "learning_rate": 4.6459196048272604e-05, "loss": 1.7607, "step": 47900 }, { "epoch": 0.71, "learning_rate": 4.645180134879319e-05, "loss": 1.78, "step": 48000 }, { "epoch": 0.71, "learning_rate": 4.6444406649313775e-05, "loss": 1.794, "step": 48100 }, { "epoch": 0.71, "learning_rate": 4.643701194983436e-05, "loss": 1.7604, "step": 48200 }, { "epoch": 0.71, "learning_rate": 4.6429617250354946e-05, "loss": 1.7383, "step": 48300 }, { "epoch": 0.72, "learning_rate": 4.642222255087553e-05, "loss": 1.7602, "step": 48400 }, { "epoch": 0.72, "learning_rate": 4.641482785139612e-05, "loss": 1.7424, "step": 48500 }, { "epoch": 0.72, "learning_rate": 4.640743315191671e-05, "loss": 1.7605, "step": 48600 }, { "epoch": 0.72, "learning_rate": 4.640011239943209e-05, "loss": 1.771, "step": 48700 }, { "epoch": 0.72, "learning_rate": 4.6392717699952676e-05, "loss": 1.7781, "step": 48800 }, { "epoch": 0.72, "learning_rate": 4.638532300047326e-05, "loss": 1.7859, "step": 48900 }, { "epoch": 0.72, "learning_rate": 4.637792830099385e-05, "loss": 1.7836, "step": 49000 }, { "epoch": 0.73, "learning_rate": 4.637053360151444e-05, "loss": 1.7654, "step": 49100 }, { "epoch": 0.73, "learning_rate": 4.6363138902035026e-05, "loss": 1.7866, "step": 49200 }, { "epoch": 0.73, "learning_rate": 4.635574420255561e-05, "loss": 1.7538, "step": 49300 }, { "epoch": 0.73, "learning_rate": 4.63483495030762e-05, "loss": 1.7573, "step": 49400 }, { "epoch": 0.73, "learning_rate": 4.634095480359679e-05, "loss": 1.7741, "step": 49500 }, { "epoch": 0.73, "learning_rate": 4.633356010411737e-05, "loss": 1.7288, "step": 49600 }, { "epoch": 0.74, "learning_rate": 4.6326165404637954e-05, "loss": 1.789, "step": 49700 }, { "epoch": 0.74, "learning_rate": 4.6318770705158546e-05, "loss": 1.7548, "step": 49800 }, { "epoch": 0.74, "learning_rate": 4.631137600567913e-05, "loss": 1.75, "step": 49900 }, { "epoch": 0.74, "learning_rate": 4.630398130619972e-05, "loss": 1.7316, "step": 50000 }, { "epoch": 0.74, "learning_rate": 4.62965866067203e-05, "loss": 1.7567, "step": 50100 }, { "epoch": 0.74, "learning_rate": 4.628919190724089e-05, "loss": 1.7842, "step": 50200 }, { "epoch": 0.74, "learning_rate": 4.628179720776148e-05, "loss": 1.7256, "step": 50300 }, { "epoch": 0.75, "learning_rate": 4.627440250828207e-05, "loss": 1.8103, "step": 50400 }, { "epoch": 0.75, "learning_rate": 4.626700780880265e-05, "loss": 1.753, "step": 50500 }, { "epoch": 0.75, "learning_rate": 4.625961310932324e-05, "loss": 1.7531, "step": 50600 }, { "epoch": 0.75, "learning_rate": 4.6252292356838625e-05, "loss": 1.7673, "step": 50700 }, { "epoch": 0.75, "learning_rate": 4.624489765735921e-05, "loss": 1.6973, "step": 50800 }, { "epoch": 0.75, "learning_rate": 4.623750295787979e-05, "loss": 1.75, "step": 50900 }, { "epoch": 0.75, "learning_rate": 4.6230108258400375e-05, "loss": 1.7644, "step": 51000 }, { "epoch": 0.76, "learning_rate": 4.622271355892097e-05, "loss": 1.7536, "step": 51100 }, { "epoch": 0.76, "learning_rate": 4.6215318859441553e-05, "loss": 1.7453, "step": 51200 }, { "epoch": 0.76, "learning_rate": 4.620792415996214e-05, "loss": 1.7729, "step": 51300 }, { "epoch": 0.76, "learning_rate": 4.6200529460482725e-05, "loss": 1.7514, "step": 51400 }, { "epoch": 0.76, "learning_rate": 4.619313476100332e-05, "loss": 1.7211, "step": 51500 }, { "epoch": 0.76, "learning_rate": 4.61857400615239e-05, "loss": 1.777, "step": 51600 }, { "epoch": 0.76, "learning_rate": 4.617834536204449e-05, "loss": 1.7453, "step": 51700 }, { "epoch": 0.77, "learning_rate": 4.6170950662565074e-05, "loss": 1.7488, "step": 51800 }, { "epoch": 0.77, "learning_rate": 4.6163555963085666e-05, "loss": 1.7438, "step": 51900 }, { "epoch": 0.77, "learning_rate": 4.615616126360625e-05, "loss": 1.7071, "step": 52000 }, { "epoch": 0.77, "learning_rate": 4.614876656412684e-05, "loss": 1.7346, "step": 52100 }, { "epoch": 0.77, "learning_rate": 4.614137186464742e-05, "loss": 1.7436, "step": 52200 }, { "epoch": 0.77, "learning_rate": 4.613397716516801e-05, "loss": 1.761, "step": 52300 }, { "epoch": 0.77, "learning_rate": 4.6126582465688595e-05, "loss": 1.7495, "step": 52400 }, { "epoch": 0.78, "learning_rate": 4.611918776620918e-05, "loss": 1.7501, "step": 52500 }, { "epoch": 0.78, "learning_rate": 4.6111793066729766e-05, "loss": 1.7356, "step": 52600 }, { "epoch": 0.78, "learning_rate": 4.610439836725036e-05, "loss": 1.7315, "step": 52700 }, { "epoch": 0.78, "learning_rate": 4.6097003667770944e-05, "loss": 1.7563, "step": 52800 }, { "epoch": 0.78, "learning_rate": 4.608960896829153e-05, "loss": 1.7512, "step": 52900 }, { "epoch": 0.78, "learning_rate": 4.6082214268812115e-05, "loss": 1.7517, "step": 53000 }, { "epoch": 0.79, "learning_rate": 4.607481956933271e-05, "loss": 1.7447, "step": 53100 }, { "epoch": 0.79, "learning_rate": 4.606742486985329e-05, "loss": 1.7181, "step": 53200 }, { "epoch": 0.79, "learning_rate": 4.606003017037388e-05, "loss": 1.7548, "step": 53300 }, { "epoch": 0.79, "learning_rate": 4.6052635470894465e-05, "loss": 1.7634, "step": 53400 }, { "epoch": 0.79, "learning_rate": 4.604524077141506e-05, "loss": 1.761, "step": 53500 }, { "epoch": 0.79, "learning_rate": 4.6037846071935636e-05, "loss": 1.7367, "step": 53600 }, { "epoch": 0.79, "learning_rate": 4.603045137245622e-05, "loss": 1.7466, "step": 53700 }, { "epoch": 0.8, "learning_rate": 4.6023056672976814e-05, "loss": 1.7266, "step": 53800 }, { "epoch": 0.8, "learning_rate": 4.60156619734974e-05, "loss": 1.7491, "step": 53900 }, { "epoch": 0.8, "learning_rate": 4.6008267274017985e-05, "loss": 1.7448, "step": 54000 }, { "epoch": 0.8, "learning_rate": 4.600087257453857e-05, "loss": 1.7238, "step": 54100 }, { "epoch": 0.8, "learning_rate": 4.599347787505916e-05, "loss": 1.7352, "step": 54200 }, { "epoch": 0.8, "learning_rate": 4.598608317557975e-05, "loss": 1.7386, "step": 54300 }, { "epoch": 0.8, "learning_rate": 4.5978688476100334e-05, "loss": 1.7512, "step": 54400 }, { "epoch": 0.81, "learning_rate": 4.597129377662092e-05, "loss": 1.7432, "step": 54500 }, { "epoch": 0.81, "learning_rate": 4.5963899077141506e-05, "loss": 1.7693, "step": 54600 }, { "epoch": 0.81, "learning_rate": 4.59565043776621e-05, "loss": 1.7302, "step": 54700 }, { "epoch": 0.81, "learning_rate": 4.5949109678182684e-05, "loss": 1.751, "step": 54800 }, { "epoch": 0.81, "learning_rate": 4.594171497870326e-05, "loss": 1.7279, "step": 54900 }, { "epoch": 0.81, "learning_rate": 4.5934320279223855e-05, "loss": 1.746, "step": 55000 }, { "epoch": 0.81, "learning_rate": 4.592692557974444e-05, "loss": 1.7433, "step": 55100 }, { "epoch": 0.82, "learning_rate": 4.5919530880265026e-05, "loss": 1.76, "step": 55200 }, { "epoch": 0.82, "learning_rate": 4.591213618078561e-05, "loss": 1.7278, "step": 55300 }, { "epoch": 0.82, "learning_rate": 4.590481542830099e-05, "loss": 1.7517, "step": 55400 }, { "epoch": 0.82, "learning_rate": 4.5897420728821585e-05, "loss": 1.7349, "step": 55500 }, { "epoch": 0.82, "learning_rate": 4.589002602934217e-05, "loss": 1.7422, "step": 55600 }, { "epoch": 0.82, "learning_rate": 4.5882631329862756e-05, "loss": 1.7348, "step": 55700 }, { "epoch": 0.83, "learning_rate": 4.587523663038334e-05, "loss": 1.7244, "step": 55800 }, { "epoch": 0.83, "learning_rate": 4.5867841930903934e-05, "loss": 1.7639, "step": 55900 }, { "epoch": 0.83, "learning_rate": 4.586044723142452e-05, "loss": 1.7663, "step": 56000 }, { "epoch": 0.83, "learning_rate": 4.5853052531945105e-05, "loss": 1.7351, "step": 56100 }, { "epoch": 0.83, "learning_rate": 4.584565783246569e-05, "loss": 1.719, "step": 56200 }, { "epoch": 0.83, "learning_rate": 4.583826313298628e-05, "loss": 1.7408, "step": 56300 }, { "epoch": 0.83, "learning_rate": 4.583086843350686e-05, "loss": 1.7307, "step": 56400 }, { "epoch": 0.84, "learning_rate": 4.582347373402745e-05, "loss": 1.7479, "step": 56500 }, { "epoch": 0.84, "learning_rate": 4.581607903454804e-05, "loss": 1.7718, "step": 56600 }, { "epoch": 0.84, "learning_rate": 4.5808684335068626e-05, "loss": 1.7378, "step": 56700 }, { "epoch": 0.84, "learning_rate": 4.580128963558921e-05, "loss": 1.741, "step": 56800 }, { "epoch": 0.84, "learning_rate": 4.57938949361098e-05, "loss": 1.7423, "step": 56900 }, { "epoch": 0.84, "learning_rate": 4.578650023663038e-05, "loss": 1.7297, "step": 57000 }, { "epoch": 0.84, "learning_rate": 4.5779105537150975e-05, "loss": 1.7669, "step": 57100 }, { "epoch": 0.85, "learning_rate": 4.577171083767156e-05, "loss": 1.7625, "step": 57200 }, { "epoch": 0.85, "learning_rate": 4.576431613819215e-05, "loss": 1.745, "step": 57300 }, { "epoch": 0.85, "learning_rate": 4.575692143871273e-05, "loss": 1.7359, "step": 57400 }, { "epoch": 0.85, "learning_rate": 4.574952673923332e-05, "loss": 1.724, "step": 57500 }, { "epoch": 0.85, "learning_rate": 4.5742132039753904e-05, "loss": 1.698, "step": 57600 }, { "epoch": 0.85, "learning_rate": 4.573473734027449e-05, "loss": 1.7201, "step": 57700 }, { "epoch": 0.85, "learning_rate": 4.572734264079508e-05, "loss": 1.75, "step": 57800 }, { "epoch": 0.86, "learning_rate": 4.571994794131567e-05, "loss": 1.7518, "step": 57900 }, { "epoch": 0.86, "learning_rate": 4.571255324183625e-05, "loss": 1.7491, "step": 58000 }, { "epoch": 0.86, "learning_rate": 4.570515854235684e-05, "loss": 1.7018, "step": 58100 }, { "epoch": 0.86, "learning_rate": 4.569776384287743e-05, "loss": 1.7489, "step": 58200 }, { "epoch": 0.86, "learning_rate": 4.5690369143398017e-05, "loss": 1.7427, "step": 58300 }, { "epoch": 0.86, "learning_rate": 4.56830483909134e-05, "loss": 1.748, "step": 58400 }, { "epoch": 0.87, "learning_rate": 4.567565369143398e-05, "loss": 1.7254, "step": 58500 }, { "epoch": 0.87, "learning_rate": 4.566825899195457e-05, "loss": 1.748, "step": 58600 }, { "epoch": 0.87, "learning_rate": 4.566086429247516e-05, "loss": 1.7413, "step": 58700 }, { "epoch": 0.87, "learning_rate": 4.565346959299574e-05, "loss": 1.7595, "step": 58800 }, { "epoch": 0.87, "learning_rate": 4.5646074893516325e-05, "loss": 1.7217, "step": 58900 }, { "epoch": 0.87, "learning_rate": 4.563868019403692e-05, "loss": 1.7287, "step": 59000 }, { "epoch": 0.87, "learning_rate": 4.56312854945575e-05, "loss": 1.7052, "step": 59100 }, { "epoch": 0.88, "learning_rate": 4.562389079507809e-05, "loss": 1.721, "step": 59200 }, { "epoch": 0.88, "learning_rate": 4.5616496095598675e-05, "loss": 1.7357, "step": 59300 }, { "epoch": 0.88, "learning_rate": 4.560910139611927e-05, "loss": 1.74, "step": 59400 }, { "epoch": 0.88, "learning_rate": 4.560170669663985e-05, "loss": 1.7229, "step": 59500 }, { "epoch": 0.88, "learning_rate": 4.559431199716044e-05, "loss": 1.7402, "step": 59600 }, { "epoch": 0.88, "learning_rate": 4.5586917297681024e-05, "loss": 1.7735, "step": 59700 }, { "epoch": 0.88, "learning_rate": 4.557952259820161e-05, "loss": 1.7196, "step": 59800 }, { "epoch": 0.89, "learning_rate": 4.55721278987222e-05, "loss": 1.7319, "step": 59900 }, { "epoch": 0.89, "learning_rate": 4.556473319924279e-05, "loss": 1.7132, "step": 60000 }, { "epoch": 0.89, "learning_rate": 4.555733849976337e-05, "loss": 1.737, "step": 60100 }, { "epoch": 0.89, "learning_rate": 4.554994380028396e-05, "loss": 1.7046, "step": 60200 }, { "epoch": 0.89, "learning_rate": 4.5542549100804544e-05, "loss": 1.7156, "step": 60300 }, { "epoch": 0.89, "learning_rate": 4.553515440132513e-05, "loss": 1.7484, "step": 60400 }, { "epoch": 0.89, "learning_rate": 4.5527759701845716e-05, "loss": 1.6991, "step": 60500 }, { "epoch": 0.9, "learning_rate": 4.552036500236631e-05, "loss": 1.7218, "step": 60600 }, { "epoch": 0.9, "learning_rate": 4.5512970302886894e-05, "loss": 1.7342, "step": 60700 }, { "epoch": 0.9, "learning_rate": 4.5505649550402274e-05, "loss": 1.7479, "step": 60800 }, { "epoch": 0.9, "learning_rate": 4.549825485092286e-05, "loss": 1.7051, "step": 60900 }, { "epoch": 0.9, "learning_rate": 4.5490860151443446e-05, "loss": 1.7051, "step": 61000 }, { "epoch": 0.9, "learning_rate": 4.548346545196404e-05, "loss": 1.7394, "step": 61100 }, { "epoch": 0.91, "learning_rate": 4.5476070752484624e-05, "loss": 1.7232, "step": 61200 }, { "epoch": 0.91, "learning_rate": 4.546867605300521e-05, "loss": 1.7034, "step": 61300 }, { "epoch": 0.91, "learning_rate": 4.5461281353525795e-05, "loss": 1.7138, "step": 61400 }, { "epoch": 0.91, "learning_rate": 4.545388665404638e-05, "loss": 1.728, "step": 61500 }, { "epoch": 0.91, "learning_rate": 4.5446491954566966e-05, "loss": 1.7295, "step": 61600 }, { "epoch": 0.91, "learning_rate": 4.543909725508755e-05, "loss": 1.7085, "step": 61700 }, { "epoch": 0.91, "learning_rate": 4.5431702555608144e-05, "loss": 1.7272, "step": 61800 }, { "epoch": 0.92, "learning_rate": 4.542430785612873e-05, "loss": 1.744, "step": 61900 }, { "epoch": 0.92, "learning_rate": 4.5416913156649315e-05, "loss": 1.739, "step": 62000 }, { "epoch": 0.92, "learning_rate": 4.54095184571699e-05, "loss": 1.7242, "step": 62100 }, { "epoch": 0.92, "learning_rate": 4.540212375769049e-05, "loss": 1.7257, "step": 62200 }, { "epoch": 0.92, "learning_rate": 4.539472905821108e-05, "loss": 1.7315, "step": 62300 }, { "epoch": 0.92, "learning_rate": 4.5387334358731665e-05, "loss": 1.7385, "step": 62400 }, { "epoch": 0.92, "learning_rate": 4.537993965925225e-05, "loss": 1.7208, "step": 62500 }, { "epoch": 0.93, "learning_rate": 4.5372544959772836e-05, "loss": 1.7236, "step": 62600 }, { "epoch": 0.93, "learning_rate": 4.536515026029343e-05, "loss": 1.7295, "step": 62700 }, { "epoch": 0.93, "learning_rate": 4.535775556081401e-05, "loss": 1.6798, "step": 62800 }, { "epoch": 0.93, "learning_rate": 4.535036086133459e-05, "loss": 1.7287, "step": 62900 }, { "epoch": 0.93, "learning_rate": 4.5343040108849973e-05, "loss": 1.7127, "step": 63000 }, { "epoch": 0.93, "learning_rate": 4.5335645409370566e-05, "loss": 1.7257, "step": 63100 }, { "epoch": 0.93, "learning_rate": 4.532825070989115e-05, "loss": 1.7089, "step": 63200 }, { "epoch": 0.94, "learning_rate": 4.532085601041174e-05, "loss": 1.7232, "step": 63300 }, { "epoch": 0.94, "learning_rate": 4.531346131093232e-05, "loss": 1.6997, "step": 63400 }, { "epoch": 0.94, "learning_rate": 4.5306066611452915e-05, "loss": 1.7407, "step": 63500 }, { "epoch": 0.94, "learning_rate": 4.52986719119735e-05, "loss": 1.7252, "step": 63600 }, { "epoch": 0.94, "learning_rate": 4.5291277212494086e-05, "loss": 1.7325, "step": 63700 }, { "epoch": 0.94, "learning_rate": 4.528388251301467e-05, "loss": 1.703, "step": 63800 }, { "epoch": 0.95, "learning_rate": 4.5276487813535265e-05, "loss": 1.7222, "step": 63900 }, { "epoch": 0.95, "learning_rate": 4.526909311405585e-05, "loss": 1.7445, "step": 64000 }, { "epoch": 0.95, "learning_rate": 4.526169841457643e-05, "loss": 1.7151, "step": 64100 }, { "epoch": 0.95, "learning_rate": 4.525430371509702e-05, "loss": 1.7084, "step": 64200 }, { "epoch": 0.95, "learning_rate": 4.524690901561761e-05, "loss": 1.7056, "step": 64300 }, { "epoch": 0.95, "learning_rate": 4.523951431613819e-05, "loss": 1.7038, "step": 64400 }, { "epoch": 0.95, "learning_rate": 4.523219356365357e-05, "loss": 1.7043, "step": 64500 }, { "epoch": 0.96, "learning_rate": 4.522479886417416e-05, "loss": 1.7337, "step": 64600 }, { "epoch": 0.96, "learning_rate": 4.521740416469475e-05, "loss": 1.7563, "step": 64700 }, { "epoch": 0.96, "learning_rate": 4.521000946521534e-05, "loss": 1.7321, "step": 64800 }, { "epoch": 0.96, "learning_rate": 4.520261476573592e-05, "loss": 1.6985, "step": 64900 }, { "epoch": 0.96, "learning_rate": 4.519522006625651e-05, "loss": 1.7249, "step": 65000 }, { "epoch": 0.96, "learning_rate": 4.51878253667771e-05, "loss": 1.7415, "step": 65100 }, { "epoch": 0.96, "learning_rate": 4.5180430667297686e-05, "loss": 1.7054, "step": 65200 }, { "epoch": 0.97, "learning_rate": 4.517303596781827e-05, "loss": 1.6725, "step": 65300 }, { "epoch": 0.97, "learning_rate": 4.516564126833886e-05, "loss": 1.6966, "step": 65400 }, { "epoch": 0.97, "learning_rate": 4.515824656885944e-05, "loss": 1.7048, "step": 65500 }, { "epoch": 0.97, "learning_rate": 4.515085186938003e-05, "loss": 1.7154, "step": 65600 }, { "epoch": 0.97, "learning_rate": 4.5143457169900614e-05, "loss": 1.6992, "step": 65700 }, { "epoch": 0.97, "learning_rate": 4.51360624704212e-05, "loss": 1.7224, "step": 65800 }, { "epoch": 0.97, "learning_rate": 4.512866777094179e-05, "loss": 1.6985, "step": 65900 }, { "epoch": 0.98, "learning_rate": 4.512127307146238e-05, "loss": 1.7052, "step": 66000 }, { "epoch": 0.98, "learning_rate": 4.5113878371982964e-05, "loss": 1.7244, "step": 66100 }, { "epoch": 0.98, "learning_rate": 4.510648367250355e-05, "loss": 1.7161, "step": 66200 }, { "epoch": 0.98, "learning_rate": 4.509908897302414e-05, "loss": 1.7339, "step": 66300 }, { "epoch": 0.98, "learning_rate": 4.509169427354473e-05, "loss": 1.7057, "step": 66400 }, { "epoch": 0.98, "learning_rate": 4.508429957406531e-05, "loss": 1.7268, "step": 66500 }, { "epoch": 0.98, "learning_rate": 4.50769048745859e-05, "loss": 1.6977, "step": 66600 }, { "epoch": 0.99, "learning_rate": 4.506951017510649e-05, "loss": 1.7667, "step": 66700 }, { "epoch": 0.99, "learning_rate": 4.506211547562707e-05, "loss": 1.7237, "step": 66800 }, { "epoch": 0.99, "learning_rate": 4.5054720776147656e-05, "loss": 1.6985, "step": 66900 }, { "epoch": 0.99, "learning_rate": 4.504732607666825e-05, "loss": 1.7174, "step": 67000 }, { "epoch": 0.99, "learning_rate": 4.5039931377188834e-05, "loss": 1.7205, "step": 67100 }, { "epoch": 0.99, "learning_rate": 4.503253667770942e-05, "loss": 1.7076, "step": 67200 }, { "epoch": 1.0, "learning_rate": 4.5025141978230005e-05, "loss": 1.6836, "step": 67300 }, { "epoch": 1.0, "learning_rate": 4.501774727875059e-05, "loss": 1.7388, "step": 67400 }, { "epoch": 1.0, "learning_rate": 4.501035257927118e-05, "loss": 1.7164, "step": 67500 }, { "epoch": 1.0, "learning_rate": 4.500295787979177e-05, "loss": 1.7383, "step": 67600 } ], "max_steps": 676160, "num_train_epochs": 10, "total_flos": 4.132202643259392e+17, "trial_name": null, "trial_params": null }