{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.9990358348063637, "global_step": 4665, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5.000000000000001e-07, "loss": 2.8558, "step": 1 }, { "epoch": 0.01, "learning_rate": 5e-06, "loss": 2.897, "step": 10 }, { "epoch": 0.01, "learning_rate": 1e-05, "loss": 2.8778, "step": 20 }, { "epoch": 0.02, "learning_rate": 1.5e-05, "loss": 2.9056, "step": 30 }, { "epoch": 0.03, "learning_rate": 2e-05, "loss": 2.8865, "step": 40 }, { "epoch": 0.03, "learning_rate": 2.5e-05, "loss": 2.8852, "step": 50 }, { "epoch": 0.04, "learning_rate": 3e-05, "loss": 2.9159, "step": 60 }, { "epoch": 0.04, "learning_rate": 3.5e-05, "loss": 2.8774, "step": 70 }, { "epoch": 0.05, "learning_rate": 4e-05, "loss": 2.8732, "step": 80 }, { "epoch": 0.06, "learning_rate": 4.5e-05, "loss": 2.8753, "step": 90 }, { "epoch": 0.06, "learning_rate": 5e-05, "loss": 2.8977, "step": 100 }, { "epoch": 0.07, "learning_rate": 4.9890470974808325e-05, "loss": 2.9238, "step": 110 }, { "epoch": 0.08, "learning_rate": 4.978094194961665e-05, "loss": 2.8907, "step": 120 }, { "epoch": 0.08, "learning_rate": 4.967141292442498e-05, "loss": 2.8952, "step": 130 }, { "epoch": 0.09, "learning_rate": 4.9561883899233295e-05, "loss": 2.8704, "step": 140 }, { "epoch": 0.1, "learning_rate": 4.9452354874041625e-05, "loss": 2.9006, "step": 150 }, { "epoch": 0.1, "learning_rate": 4.934282584884995e-05, "loss": 2.8912, "step": 160 }, { "epoch": 0.11, "learning_rate": 4.923329682365827e-05, "loss": 2.8691, "step": 170 }, { "epoch": 0.12, "learning_rate": 4.9123767798466594e-05, "loss": 2.8859, "step": 180 }, { "epoch": 0.12, "learning_rate": 4.901423877327492e-05, "loss": 2.8636, "step": 190 }, { "epoch": 0.13, "learning_rate": 4.890470974808325e-05, "loss": 2.8825, "step": 200 }, { "epoch": 0.13, "learning_rate": 4.879518072289157e-05, "loss": 2.8866, "step": 210 }, { "epoch": 0.14, "learning_rate": 4.868565169769989e-05, "loss": 2.9145, "step": 220 }, { "epoch": 0.15, "learning_rate": 4.8576122672508216e-05, "loss": 2.8909, "step": 230 }, { "epoch": 0.15, "learning_rate": 4.8466593647316546e-05, "loss": 2.8754, "step": 240 }, { "epoch": 0.16, "learning_rate": 4.835706462212486e-05, "loss": 2.8581, "step": 250 }, { "epoch": 0.17, "learning_rate": 4.824753559693319e-05, "loss": 2.8592, "step": 260 }, { "epoch": 0.17, "learning_rate": 4.813800657174151e-05, "loss": 2.9211, "step": 270 }, { "epoch": 0.18, "learning_rate": 4.802847754654984e-05, "loss": 2.851, "step": 280 }, { "epoch": 0.19, "learning_rate": 4.791894852135816e-05, "loss": 2.8774, "step": 290 }, { "epoch": 0.19, "learning_rate": 4.7809419496166485e-05, "loss": 2.8707, "step": 300 }, { "epoch": 0.2, "learning_rate": 4.7699890470974815e-05, "loss": 2.8243, "step": 310 }, { "epoch": 0.21, "learning_rate": 4.759036144578313e-05, "loss": 2.8519, "step": 320 }, { "epoch": 0.21, "learning_rate": 4.748083242059146e-05, "loss": 2.8624, "step": 330 }, { "epoch": 0.22, "learning_rate": 4.7371303395399784e-05, "loss": 2.8551, "step": 340 }, { "epoch": 0.22, "learning_rate": 4.726177437020811e-05, "loss": 2.8593, "step": 350 }, { "epoch": 0.23, "learning_rate": 4.715224534501643e-05, "loss": 2.9018, "step": 360 }, { "epoch": 0.24, "learning_rate": 4.704271631982475e-05, "loss": 2.8707, "step": 370 }, { "epoch": 0.24, "learning_rate": 4.6933187294633076e-05, "loss": 2.8595, "step": 380 }, { "epoch": 0.25, "learning_rate": 4.6823658269441406e-05, "loss": 2.8713, "step": 390 }, { "epoch": 0.26, "learning_rate": 4.671412924424973e-05, "loss": 2.8765, "step": 400 }, { "epoch": 0.26, "learning_rate": 4.660460021905805e-05, "loss": 2.8672, "step": 410 }, { "epoch": 0.27, "learning_rate": 4.6495071193866376e-05, "loss": 2.8757, "step": 420 }, { "epoch": 0.28, "learning_rate": 4.63855421686747e-05, "loss": 2.8763, "step": 430 }, { "epoch": 0.28, "learning_rate": 4.627601314348303e-05, "loss": 2.8511, "step": 440 }, { "epoch": 0.29, "learning_rate": 4.6166484118291345e-05, "loss": 2.8628, "step": 450 }, { "epoch": 0.3, "learning_rate": 4.6056955093099675e-05, "loss": 2.8471, "step": 460 }, { "epoch": 0.3, "learning_rate": 4.5947426067908e-05, "loss": 2.8648, "step": 470 }, { "epoch": 0.31, "learning_rate": 4.583789704271632e-05, "loss": 2.8346, "step": 480 }, { "epoch": 0.31, "learning_rate": 4.572836801752465e-05, "loss": 2.8836, "step": 490 }, { "epoch": 0.32, "learning_rate": 4.561883899233297e-05, "loss": 2.8472, "step": 500 }, { "epoch": 0.33, "learning_rate": 4.55093099671413e-05, "loss": 2.8497, "step": 510 }, { "epoch": 0.33, "learning_rate": 4.539978094194962e-05, "loss": 2.8357, "step": 520 }, { "epoch": 0.34, "learning_rate": 4.529025191675794e-05, "loss": 2.8375, "step": 530 }, { "epoch": 0.35, "learning_rate": 4.5180722891566266e-05, "loss": 2.8139, "step": 540 }, { "epoch": 0.35, "learning_rate": 4.507119386637459e-05, "loss": 2.8254, "step": 550 }, { "epoch": 0.36, "learning_rate": 4.496166484118291e-05, "loss": 2.8252, "step": 560 }, { "epoch": 0.37, "learning_rate": 4.485213581599124e-05, "loss": 2.8282, "step": 570 }, { "epoch": 0.37, "learning_rate": 4.4742606790799566e-05, "loss": 2.8091, "step": 580 }, { "epoch": 0.38, "learning_rate": 4.463307776560789e-05, "loss": 2.831, "step": 590 }, { "epoch": 0.39, "learning_rate": 4.452354874041621e-05, "loss": 2.8862, "step": 600 }, { "epoch": 0.39, "learning_rate": 4.4414019715224535e-05, "loss": 2.8276, "step": 610 }, { "epoch": 0.4, "learning_rate": 4.4304490690032865e-05, "loss": 2.8614, "step": 620 }, { "epoch": 0.4, "learning_rate": 4.419496166484118e-05, "loss": 2.7905, "step": 630 }, { "epoch": 0.41, "learning_rate": 4.408543263964951e-05, "loss": 2.8561, "step": 640 }, { "epoch": 0.42, "learning_rate": 4.3975903614457834e-05, "loss": 2.837, "step": 650 }, { "epoch": 0.42, "learning_rate": 4.386637458926616e-05, "loss": 2.8084, "step": 660 }, { "epoch": 0.43, "learning_rate": 4.375684556407448e-05, "loss": 2.7978, "step": 670 }, { "epoch": 0.44, "learning_rate": 4.36473165388828e-05, "loss": 2.8264, "step": 680 }, { "epoch": 0.44, "learning_rate": 4.353778751369113e-05, "loss": 2.7847, "step": 690 }, { "epoch": 0.45, "learning_rate": 4.3428258488499456e-05, "loss": 2.8349, "step": 700 }, { "epoch": 0.46, "learning_rate": 4.331872946330778e-05, "loss": 2.8281, "step": 710 }, { "epoch": 0.46, "learning_rate": 4.32092004381161e-05, "loss": 2.8142, "step": 720 }, { "epoch": 0.47, "learning_rate": 4.3099671412924426e-05, "loss": 2.8364, "step": 730 }, { "epoch": 0.48, "learning_rate": 4.299014238773275e-05, "loss": 2.8077, "step": 740 }, { "epoch": 0.48, "learning_rate": 4.288061336254108e-05, "loss": 2.7992, "step": 750 }, { "epoch": 0.49, "learning_rate": 4.27710843373494e-05, "loss": 2.8266, "step": 760 }, { "epoch": 0.49, "learning_rate": 4.2661555312157725e-05, "loss": 2.8064, "step": 770 }, { "epoch": 0.5, "learning_rate": 4.255202628696605e-05, "loss": 2.8159, "step": 780 }, { "epoch": 0.51, "learning_rate": 4.244249726177437e-05, "loss": 2.8324, "step": 790 }, { "epoch": 0.51, "learning_rate": 4.23329682365827e-05, "loss": 2.8898, "step": 800 }, { "epoch": 0.52, "learning_rate": 4.222343921139102e-05, "loss": 2.8212, "step": 810 }, { "epoch": 0.53, "learning_rate": 4.211391018619935e-05, "loss": 2.8234, "step": 820 }, { "epoch": 0.53, "learning_rate": 4.2004381161007663e-05, "loss": 2.8306, "step": 830 }, { "epoch": 0.54, "learning_rate": 4.189485213581599e-05, "loss": 2.8276, "step": 840 }, { "epoch": 0.55, "learning_rate": 4.1785323110624316e-05, "loss": 2.8373, "step": 850 }, { "epoch": 0.55, "learning_rate": 4.167579408543264e-05, "loss": 2.8032, "step": 860 }, { "epoch": 0.56, "learning_rate": 4.156626506024097e-05, "loss": 2.8054, "step": 870 }, { "epoch": 0.57, "learning_rate": 4.145673603504929e-05, "loss": 2.806, "step": 880 }, { "epoch": 0.57, "learning_rate": 4.1347207009857616e-05, "loss": 2.822, "step": 890 }, { "epoch": 0.58, "learning_rate": 4.123767798466594e-05, "loss": 2.8198, "step": 900 }, { "epoch": 0.58, "learning_rate": 4.112814895947426e-05, "loss": 2.8682, "step": 910 }, { "epoch": 0.59, "learning_rate": 4.1018619934282585e-05, "loss": 2.8474, "step": 920 }, { "epoch": 0.6, "learning_rate": 4.0909090909090915e-05, "loss": 2.8419, "step": 930 }, { "epoch": 0.6, "learning_rate": 4.079956188389923e-05, "loss": 2.8299, "step": 940 }, { "epoch": 0.61, "learning_rate": 4.069003285870756e-05, "loss": 2.8078, "step": 950 }, { "epoch": 0.62, "learning_rate": 4.0580503833515884e-05, "loss": 2.8012, "step": 960 }, { "epoch": 0.62, "learning_rate": 4.047097480832421e-05, "loss": 2.793, "step": 970 }, { "epoch": 0.63, "learning_rate": 4.036144578313254e-05, "loss": 2.8289, "step": 980 }, { "epoch": 0.64, "learning_rate": 4.025191675794085e-05, "loss": 2.8005, "step": 990 }, { "epoch": 0.64, "learning_rate": 4.014238773274918e-05, "loss": 2.8124, "step": 1000 }, { "epoch": 0.65, "learning_rate": 4.00328587075575e-05, "loss": 2.7928, "step": 1010 }, { "epoch": 0.66, "learning_rate": 3.992332968236583e-05, "loss": 2.8334, "step": 1020 }, { "epoch": 0.66, "learning_rate": 3.981380065717415e-05, "loss": 2.8234, "step": 1030 }, { "epoch": 0.67, "learning_rate": 3.9704271631982476e-05, "loss": 2.8509, "step": 1040 }, { "epoch": 0.67, "learning_rate": 3.9594742606790806e-05, "loss": 2.7912, "step": 1050 }, { "epoch": 0.68, "learning_rate": 3.948521358159912e-05, "loss": 2.8191, "step": 1060 }, { "epoch": 0.69, "learning_rate": 3.937568455640745e-05, "loss": 2.7925, "step": 1070 }, { "epoch": 0.69, "learning_rate": 3.9266155531215775e-05, "loss": 2.8202, "step": 1080 }, { "epoch": 0.7, "learning_rate": 3.91566265060241e-05, "loss": 2.8411, "step": 1090 }, { "epoch": 0.71, "learning_rate": 3.904709748083242e-05, "loss": 2.8074, "step": 1100 }, { "epoch": 0.71, "learning_rate": 3.893756845564075e-05, "loss": 2.8486, "step": 1110 }, { "epoch": 0.72, "learning_rate": 3.882803943044907e-05, "loss": 2.8377, "step": 1120 }, { "epoch": 0.73, "learning_rate": 3.87185104052574e-05, "loss": 2.7922, "step": 1130 }, { "epoch": 0.73, "learning_rate": 3.860898138006572e-05, "loss": 2.7544, "step": 1140 }, { "epoch": 0.74, "learning_rate": 3.849945235487404e-05, "loss": 2.8116, "step": 1150 }, { "epoch": 0.75, "learning_rate": 3.838992332968237e-05, "loss": 2.8353, "step": 1160 }, { "epoch": 0.75, "learning_rate": 3.828039430449069e-05, "loss": 2.7945, "step": 1170 }, { "epoch": 0.76, "learning_rate": 3.817086527929902e-05, "loss": 2.8528, "step": 1180 }, { "epoch": 0.76, "learning_rate": 3.8061336254107336e-05, "loss": 2.7954, "step": 1190 }, { "epoch": 0.77, "learning_rate": 3.7951807228915666e-05, "loss": 2.8015, "step": 1200 }, { "epoch": 0.78, "learning_rate": 3.784227820372399e-05, "loss": 2.7983, "step": 1210 }, { "epoch": 0.78, "learning_rate": 3.773274917853231e-05, "loss": 2.8139, "step": 1220 }, { "epoch": 0.79, "learning_rate": 3.7623220153340635e-05, "loss": 2.8045, "step": 1230 }, { "epoch": 0.8, "learning_rate": 3.751369112814896e-05, "loss": 2.8147, "step": 1240 }, { "epoch": 0.8, "learning_rate": 3.740416210295729e-05, "loss": 2.8169, "step": 1250 }, { "epoch": 0.81, "learning_rate": 3.729463307776561e-05, "loss": 2.8122, "step": 1260 }, { "epoch": 0.82, "learning_rate": 3.7185104052573934e-05, "loss": 2.8188, "step": 1270 }, { "epoch": 0.82, "learning_rate": 3.707557502738226e-05, "loss": 2.8409, "step": 1280 }, { "epoch": 0.83, "learning_rate": 3.696604600219058e-05, "loss": 2.7812, "step": 1290 }, { "epoch": 0.84, "learning_rate": 3.6856516976998903e-05, "loss": 2.8083, "step": 1300 }, { "epoch": 0.84, "learning_rate": 3.674698795180723e-05, "loss": 2.7675, "step": 1310 }, { "epoch": 0.85, "learning_rate": 3.6637458926615556e-05, "loss": 2.8239, "step": 1320 }, { "epoch": 0.85, "learning_rate": 3.652792990142388e-05, "loss": 2.8222, "step": 1330 }, { "epoch": 0.86, "learning_rate": 3.64184008762322e-05, "loss": 2.7656, "step": 1340 }, { "epoch": 0.87, "learning_rate": 3.6308871851040526e-05, "loss": 2.8149, "step": 1350 }, { "epoch": 0.87, "learning_rate": 3.6199342825848856e-05, "loss": 2.7705, "step": 1360 }, { "epoch": 0.88, "learning_rate": 3.608981380065717e-05, "loss": 2.7853, "step": 1370 }, { "epoch": 0.89, "learning_rate": 3.59802847754655e-05, "loss": 2.7719, "step": 1380 }, { "epoch": 0.89, "learning_rate": 3.5870755750273825e-05, "loss": 2.755, "step": 1390 }, { "epoch": 0.9, "learning_rate": 3.576122672508215e-05, "loss": 2.8628, "step": 1400 }, { "epoch": 0.91, "learning_rate": 3.565169769989047e-05, "loss": 2.7986, "step": 1410 }, { "epoch": 0.91, "learning_rate": 3.5542168674698794e-05, "loss": 2.8426, "step": 1420 }, { "epoch": 0.92, "learning_rate": 3.5432639649507124e-05, "loss": 2.7956, "step": 1430 }, { "epoch": 0.93, "learning_rate": 3.532311062431545e-05, "loss": 2.7683, "step": 1440 }, { "epoch": 0.93, "learning_rate": 3.521358159912377e-05, "loss": 2.7692, "step": 1450 }, { "epoch": 0.94, "learning_rate": 3.510405257393209e-05, "loss": 2.8034, "step": 1460 }, { "epoch": 0.94, "learning_rate": 3.4994523548740416e-05, "loss": 2.7739, "step": 1470 }, { "epoch": 0.95, "learning_rate": 3.488499452354874e-05, "loss": 2.7754, "step": 1480 }, { "epoch": 0.96, "learning_rate": 3.477546549835707e-05, "loss": 2.7881, "step": 1490 }, { "epoch": 0.96, "learning_rate": 3.4665936473165386e-05, "loss": 2.8022, "step": 1500 }, { "epoch": 0.97, "learning_rate": 3.4556407447973716e-05, "loss": 2.8127, "step": 1510 }, { "epoch": 0.98, "learning_rate": 3.444687842278204e-05, "loss": 2.7977, "step": 1520 }, { "epoch": 0.98, "learning_rate": 3.433734939759036e-05, "loss": 2.8056, "step": 1530 }, { "epoch": 0.99, "learning_rate": 3.422782037239869e-05, "loss": 2.7901, "step": 1540 }, { "epoch": 1.0, "learning_rate": 3.411829134720701e-05, "loss": 2.8467, "step": 1550 }, { "epoch": 1.0, "learning_rate": 3.400876232201534e-05, "loss": 2.7446, "step": 1560 }, { "epoch": 1.01, "learning_rate": 3.389923329682366e-05, "loss": 2.7269, "step": 1570 }, { "epoch": 1.02, "learning_rate": 3.3789704271631984e-05, "loss": 2.7249, "step": 1580 }, { "epoch": 1.02, "learning_rate": 3.368017524644031e-05, "loss": 2.7381, "step": 1590 }, { "epoch": 1.03, "learning_rate": 3.357064622124863e-05, "loss": 2.7462, "step": 1600 }, { "epoch": 1.03, "learning_rate": 3.346111719605696e-05, "loss": 2.761, "step": 1610 }, { "epoch": 1.04, "learning_rate": 3.335158817086528e-05, "loss": 2.7193, "step": 1620 }, { "epoch": 1.05, "learning_rate": 3.3242059145673606e-05, "loss": 2.7272, "step": 1630 }, { "epoch": 1.05, "learning_rate": 3.313253012048193e-05, "loss": 2.7436, "step": 1640 }, { "epoch": 1.06, "learning_rate": 3.302300109529025e-05, "loss": 2.738, "step": 1650 }, { "epoch": 1.07, "learning_rate": 3.2913472070098576e-05, "loss": 2.7364, "step": 1660 }, { "epoch": 1.07, "learning_rate": 3.2803943044906906e-05, "loss": 2.7036, "step": 1670 }, { "epoch": 1.08, "learning_rate": 3.269441401971522e-05, "loss": 2.7632, "step": 1680 }, { "epoch": 1.09, "learning_rate": 3.258488499452355e-05, "loss": 2.749, "step": 1690 }, { "epoch": 1.09, "learning_rate": 3.2475355969331875e-05, "loss": 2.7418, "step": 1700 }, { "epoch": 1.1, "learning_rate": 3.23658269441402e-05, "loss": 2.718, "step": 1710 }, { "epoch": 1.11, "learning_rate": 3.225629791894853e-05, "loss": 2.6969, "step": 1720 }, { "epoch": 1.11, "learning_rate": 3.2146768893756844e-05, "loss": 2.7334, "step": 1730 }, { "epoch": 1.12, "learning_rate": 3.2037239868565174e-05, "loss": 2.7445, "step": 1740 }, { "epoch": 1.12, "learning_rate": 3.192771084337349e-05, "loss": 2.7153, "step": 1750 }, { "epoch": 1.13, "learning_rate": 3.181818181818182e-05, "loss": 2.7625, "step": 1760 }, { "epoch": 1.14, "learning_rate": 3.1708652792990143e-05, "loss": 2.7302, "step": 1770 }, { "epoch": 1.14, "learning_rate": 3.1599123767798467e-05, "loss": 2.7519, "step": 1780 }, { "epoch": 1.15, "learning_rate": 3.148959474260679e-05, "loss": 2.7566, "step": 1790 }, { "epoch": 1.16, "learning_rate": 3.138006571741512e-05, "loss": 2.7113, "step": 1800 }, { "epoch": 1.16, "learning_rate": 3.127053669222344e-05, "loss": 2.761, "step": 1810 }, { "epoch": 1.17, "learning_rate": 3.1161007667031766e-05, "loss": 2.7201, "step": 1820 }, { "epoch": 1.18, "learning_rate": 3.105147864184009e-05, "loss": 2.7285, "step": 1830 }, { "epoch": 1.18, "learning_rate": 3.094194961664841e-05, "loss": 2.7033, "step": 1840 }, { "epoch": 1.19, "learning_rate": 3.083242059145674e-05, "loss": 2.697, "step": 1850 }, { "epoch": 1.2, "learning_rate": 3.072289156626506e-05, "loss": 2.7614, "step": 1860 }, { "epoch": 1.2, "learning_rate": 3.061336254107339e-05, "loss": 2.7281, "step": 1870 }, { "epoch": 1.21, "learning_rate": 3.0503833515881708e-05, "loss": 2.7169, "step": 1880 }, { "epoch": 1.21, "learning_rate": 3.0394304490690034e-05, "loss": 2.7, "step": 1890 }, { "epoch": 1.22, "learning_rate": 3.028477546549836e-05, "loss": 2.7139, "step": 1900 }, { "epoch": 1.23, "learning_rate": 3.017524644030668e-05, "loss": 2.6886, "step": 1910 }, { "epoch": 1.23, "learning_rate": 3.0065717415115007e-05, "loss": 2.7662, "step": 1920 }, { "epoch": 1.24, "learning_rate": 2.995618838992333e-05, "loss": 2.7315, "step": 1930 }, { "epoch": 1.25, "learning_rate": 2.9846659364731656e-05, "loss": 2.7221, "step": 1940 }, { "epoch": 1.25, "learning_rate": 2.9737130339539983e-05, "loss": 2.6996, "step": 1950 }, { "epoch": 1.26, "learning_rate": 2.9627601314348303e-05, "loss": 2.7292, "step": 1960 }, { "epoch": 1.27, "learning_rate": 2.951807228915663e-05, "loss": 2.7306, "step": 1970 }, { "epoch": 1.27, "learning_rate": 2.940854326396495e-05, "loss": 2.7258, "step": 1980 }, { "epoch": 1.28, "learning_rate": 2.9299014238773275e-05, "loss": 2.7372, "step": 1990 }, { "epoch": 1.29, "learning_rate": 2.9189485213581602e-05, "loss": 2.7249, "step": 2000 }, { "epoch": 1.29, "learning_rate": 2.907995618838992e-05, "loss": 2.7432, "step": 2010 }, { "epoch": 1.3, "learning_rate": 2.8970427163198248e-05, "loss": 2.716, "step": 2020 }, { "epoch": 1.3, "learning_rate": 2.8860898138006575e-05, "loss": 2.7356, "step": 2030 }, { "epoch": 1.31, "learning_rate": 2.8751369112814898e-05, "loss": 2.7392, "step": 2040 }, { "epoch": 1.32, "learning_rate": 2.8641840087623224e-05, "loss": 2.7202, "step": 2050 }, { "epoch": 1.32, "learning_rate": 2.8532311062431544e-05, "loss": 2.751, "step": 2060 }, { "epoch": 1.33, "learning_rate": 2.842278203723987e-05, "loss": 2.7147, "step": 2070 }, { "epoch": 1.34, "learning_rate": 2.8313253012048197e-05, "loss": 2.766, "step": 2080 }, { "epoch": 1.34, "learning_rate": 2.8203723986856517e-05, "loss": 2.7035, "step": 2090 }, { "epoch": 1.35, "learning_rate": 2.8094194961664843e-05, "loss": 2.7446, "step": 2100 }, { "epoch": 1.36, "learning_rate": 2.7984665936473166e-05, "loss": 2.7258, "step": 2110 }, { "epoch": 1.36, "learning_rate": 2.787513691128149e-05, "loss": 2.7014, "step": 2120 }, { "epoch": 1.37, "learning_rate": 2.7765607886089816e-05, "loss": 2.7351, "step": 2130 }, { "epoch": 1.38, "learning_rate": 2.765607886089814e-05, "loss": 2.714, "step": 2140 }, { "epoch": 1.38, "learning_rate": 2.7546549835706465e-05, "loss": 2.7449, "step": 2150 }, { "epoch": 1.39, "learning_rate": 2.7437020810514785e-05, "loss": 2.7405, "step": 2160 }, { "epoch": 1.39, "learning_rate": 2.732749178532311e-05, "loss": 2.7395, "step": 2170 }, { "epoch": 1.4, "learning_rate": 2.7217962760131438e-05, "loss": 2.748, "step": 2180 }, { "epoch": 1.41, "learning_rate": 2.7108433734939758e-05, "loss": 2.7138, "step": 2190 }, { "epoch": 1.41, "learning_rate": 2.6998904709748084e-05, "loss": 2.7807, "step": 2200 }, { "epoch": 1.42, "learning_rate": 2.6889375684556407e-05, "loss": 2.759, "step": 2210 }, { "epoch": 1.43, "learning_rate": 2.6779846659364734e-05, "loss": 2.7023, "step": 2220 }, { "epoch": 1.43, "learning_rate": 2.667031763417306e-05, "loss": 2.7701, "step": 2230 }, { "epoch": 1.44, "learning_rate": 2.656078860898138e-05, "loss": 2.7268, "step": 2240 }, { "epoch": 1.45, "learning_rate": 2.6451259583789707e-05, "loss": 2.7427, "step": 2250 }, { "epoch": 1.45, "learning_rate": 2.6341730558598033e-05, "loss": 2.6808, "step": 2260 }, { "epoch": 1.46, "learning_rate": 2.6232201533406353e-05, "loss": 2.7167, "step": 2270 }, { "epoch": 1.47, "learning_rate": 2.612267250821468e-05, "loss": 2.71, "step": 2280 }, { "epoch": 1.47, "learning_rate": 2.6013143483023e-05, "loss": 2.7486, "step": 2290 }, { "epoch": 1.48, "learning_rate": 2.5903614457831325e-05, "loss": 2.7368, "step": 2300 }, { "epoch": 1.48, "learning_rate": 2.5794085432639652e-05, "loss": 2.7369, "step": 2310 }, { "epoch": 1.49, "learning_rate": 2.5684556407447975e-05, "loss": 2.7162, "step": 2320 }, { "epoch": 1.5, "learning_rate": 2.55750273822563e-05, "loss": 2.7446, "step": 2330 }, { "epoch": 1.5, "learning_rate": 2.546549835706462e-05, "loss": 2.7379, "step": 2340 }, { "epoch": 1.51, "learning_rate": 2.5355969331872948e-05, "loss": 2.7436, "step": 2350 }, { "epoch": 1.52, "learning_rate": 2.5246440306681274e-05, "loss": 2.6969, "step": 2360 }, { "epoch": 1.52, "learning_rate": 2.5136911281489594e-05, "loss": 2.7595, "step": 2370 }, { "epoch": 1.53, "learning_rate": 2.502738225629792e-05, "loss": 2.7417, "step": 2380 }, { "epoch": 1.54, "learning_rate": 2.4917853231106244e-05, "loss": 2.6993, "step": 2390 }, { "epoch": 1.54, "learning_rate": 2.4808324205914567e-05, "loss": 2.6811, "step": 2400 }, { "epoch": 1.55, "learning_rate": 2.4698795180722893e-05, "loss": 2.7363, "step": 2410 }, { "epoch": 1.56, "learning_rate": 2.458926615553122e-05, "loss": 2.7304, "step": 2420 }, { "epoch": 1.56, "learning_rate": 2.4479737130339543e-05, "loss": 2.706, "step": 2430 }, { "epoch": 1.57, "learning_rate": 2.4370208105147866e-05, "loss": 2.7094, "step": 2440 }, { "epoch": 1.57, "learning_rate": 2.426067907995619e-05, "loss": 2.7501, "step": 2450 }, { "epoch": 1.58, "learning_rate": 2.4151150054764512e-05, "loss": 2.7244, "step": 2460 }, { "epoch": 1.59, "learning_rate": 2.404162102957284e-05, "loss": 2.7131, "step": 2470 }, { "epoch": 1.59, "learning_rate": 2.393209200438116e-05, "loss": 2.7198, "step": 2480 }, { "epoch": 1.6, "learning_rate": 2.3822562979189485e-05, "loss": 2.7474, "step": 2490 }, { "epoch": 1.61, "learning_rate": 2.371303395399781e-05, "loss": 2.7216, "step": 2500 }, { "epoch": 1.61, "learning_rate": 2.3603504928806138e-05, "loss": 2.7487, "step": 2510 }, { "epoch": 1.62, "learning_rate": 2.349397590361446e-05, "loss": 2.7063, "step": 2520 }, { "epoch": 1.63, "learning_rate": 2.3384446878422784e-05, "loss": 2.7108, "step": 2530 }, { "epoch": 1.63, "learning_rate": 2.3274917853231107e-05, "loss": 2.7471, "step": 2540 }, { "epoch": 1.64, "learning_rate": 2.316538882803943e-05, "loss": 2.6941, "step": 2550 }, { "epoch": 1.65, "learning_rate": 2.3055859802847757e-05, "loss": 2.7476, "step": 2560 }, { "epoch": 1.65, "learning_rate": 2.294633077765608e-05, "loss": 2.6889, "step": 2570 }, { "epoch": 1.66, "learning_rate": 2.2836801752464403e-05, "loss": 2.7739, "step": 2580 }, { "epoch": 1.66, "learning_rate": 2.272727272727273e-05, "loss": 2.7491, "step": 2590 }, { "epoch": 1.67, "learning_rate": 2.2617743702081052e-05, "loss": 2.725, "step": 2600 }, { "epoch": 1.68, "learning_rate": 2.250821467688938e-05, "loss": 2.7143, "step": 2610 }, { "epoch": 1.68, "learning_rate": 2.2398685651697702e-05, "loss": 2.7077, "step": 2620 }, { "epoch": 1.69, "learning_rate": 2.2289156626506025e-05, "loss": 2.7286, "step": 2630 }, { "epoch": 1.7, "learning_rate": 2.2179627601314348e-05, "loss": 2.7379, "step": 2640 }, { "epoch": 1.7, "learning_rate": 2.2070098576122675e-05, "loss": 2.7116, "step": 2650 }, { "epoch": 1.71, "learning_rate": 2.1960569550930998e-05, "loss": 2.7165, "step": 2660 }, { "epoch": 1.72, "learning_rate": 2.185104052573932e-05, "loss": 2.7406, "step": 2670 }, { "epoch": 1.72, "learning_rate": 2.1741511500547644e-05, "loss": 2.7019, "step": 2680 }, { "epoch": 1.73, "learning_rate": 2.163198247535597e-05, "loss": 2.738, "step": 2690 }, { "epoch": 1.74, "learning_rate": 2.1522453450164297e-05, "loss": 2.688, "step": 2700 }, { "epoch": 1.74, "learning_rate": 2.141292442497262e-05, "loss": 2.7302, "step": 2710 }, { "epoch": 1.75, "learning_rate": 2.1303395399780943e-05, "loss": 2.7449, "step": 2720 }, { "epoch": 1.75, "learning_rate": 2.1193866374589266e-05, "loss": 2.7179, "step": 2730 }, { "epoch": 1.76, "learning_rate": 2.1084337349397593e-05, "loss": 2.6978, "step": 2740 }, { "epoch": 1.77, "learning_rate": 2.0974808324205916e-05, "loss": 2.718, "step": 2750 }, { "epoch": 1.77, "learning_rate": 2.086527929901424e-05, "loss": 2.7595, "step": 2760 }, { "epoch": 1.78, "learning_rate": 2.0755750273822562e-05, "loss": 2.7445, "step": 2770 }, { "epoch": 1.79, "learning_rate": 2.064622124863089e-05, "loss": 2.6969, "step": 2780 }, { "epoch": 1.79, "learning_rate": 2.0536692223439215e-05, "loss": 2.7149, "step": 2790 }, { "epoch": 1.8, "learning_rate": 2.0427163198247538e-05, "loss": 2.6722, "step": 2800 }, { "epoch": 1.81, "learning_rate": 2.031763417305586e-05, "loss": 2.7336, "step": 2810 }, { "epoch": 1.81, "learning_rate": 2.0208105147864184e-05, "loss": 2.7387, "step": 2820 }, { "epoch": 1.82, "learning_rate": 2.009857612267251e-05, "loss": 2.753, "step": 2830 }, { "epoch": 1.83, "learning_rate": 1.9989047097480834e-05, "loss": 2.7185, "step": 2840 }, { "epoch": 1.83, "learning_rate": 1.9879518072289157e-05, "loss": 2.7218, "step": 2850 }, { "epoch": 1.84, "learning_rate": 1.976998904709748e-05, "loss": 2.7595, "step": 2860 }, { "epoch": 1.85, "learning_rate": 1.9660460021905803e-05, "loss": 2.7301, "step": 2870 }, { "epoch": 1.85, "learning_rate": 1.955093099671413e-05, "loss": 2.7108, "step": 2880 }, { "epoch": 1.86, "learning_rate": 1.9441401971522456e-05, "loss": 2.7146, "step": 2890 }, { "epoch": 1.86, "learning_rate": 1.933187294633078e-05, "loss": 2.7083, "step": 2900 }, { "epoch": 1.87, "learning_rate": 1.9222343921139102e-05, "loss": 2.6983, "step": 2910 }, { "epoch": 1.88, "learning_rate": 1.9112814895947426e-05, "loss": 2.6832, "step": 2920 }, { "epoch": 1.88, "learning_rate": 1.9003285870755752e-05, "loss": 2.7189, "step": 2930 }, { "epoch": 1.89, "learning_rate": 1.8893756845564075e-05, "loss": 2.7567, "step": 2940 }, { "epoch": 1.9, "learning_rate": 1.8784227820372398e-05, "loss": 2.6915, "step": 2950 }, { "epoch": 1.9, "learning_rate": 1.867469879518072e-05, "loss": 2.7368, "step": 2960 }, { "epoch": 1.91, "learning_rate": 1.8565169769989048e-05, "loss": 2.7478, "step": 2970 }, { "epoch": 1.92, "learning_rate": 1.8455640744797374e-05, "loss": 2.71, "step": 2980 }, { "epoch": 1.92, "learning_rate": 1.8346111719605697e-05, "loss": 2.7042, "step": 2990 }, { "epoch": 1.93, "learning_rate": 1.823658269441402e-05, "loss": 2.7431, "step": 3000 }, { "epoch": 1.94, "learning_rate": 1.8127053669222344e-05, "loss": 2.7226, "step": 3010 }, { "epoch": 1.94, "learning_rate": 1.801752464403067e-05, "loss": 2.7644, "step": 3020 }, { "epoch": 1.95, "learning_rate": 1.7907995618838993e-05, "loss": 2.7904, "step": 3030 }, { "epoch": 1.95, "learning_rate": 1.7798466593647316e-05, "loss": 2.686, "step": 3040 }, { "epoch": 1.96, "learning_rate": 1.768893756845564e-05, "loss": 2.6975, "step": 3050 }, { "epoch": 1.97, "learning_rate": 1.7579408543263966e-05, "loss": 2.7183, "step": 3060 }, { "epoch": 1.97, "learning_rate": 1.7469879518072292e-05, "loss": 2.7052, "step": 3070 }, { "epoch": 1.98, "learning_rate": 1.7360350492880616e-05, "loss": 2.7286, "step": 3080 }, { "epoch": 1.99, "learning_rate": 1.725082146768894e-05, "loss": 2.7089, "step": 3090 }, { "epoch": 1.99, "learning_rate": 1.7141292442497262e-05, "loss": 2.7264, "step": 3100 }, { "epoch": 2.0, "learning_rate": 1.7031763417305588e-05, "loss": 2.7051, "step": 3110 }, { "epoch": 2.01, "learning_rate": 1.692223439211391e-05, "loss": 2.6688, "step": 3120 }, { "epoch": 2.01, "learning_rate": 1.6812705366922234e-05, "loss": 2.6887, "step": 3130 }, { "epoch": 2.02, "learning_rate": 1.6703176341730558e-05, "loss": 2.6779, "step": 3140 }, { "epoch": 2.03, "learning_rate": 1.659364731653888e-05, "loss": 2.6838, "step": 3150 }, { "epoch": 2.03, "learning_rate": 1.6484118291347207e-05, "loss": 2.6666, "step": 3160 }, { "epoch": 2.04, "learning_rate": 1.6374589266155534e-05, "loss": 2.6558, "step": 3170 }, { "epoch": 2.04, "learning_rate": 1.6265060240963857e-05, "loss": 2.6693, "step": 3180 }, { "epoch": 2.05, "learning_rate": 1.615553121577218e-05, "loss": 2.6539, "step": 3190 }, { "epoch": 2.06, "learning_rate": 1.6046002190580506e-05, "loss": 2.6365, "step": 3200 }, { "epoch": 2.06, "learning_rate": 1.593647316538883e-05, "loss": 2.6712, "step": 3210 }, { "epoch": 2.07, "learning_rate": 1.5826944140197152e-05, "loss": 2.6668, "step": 3220 }, { "epoch": 2.08, "learning_rate": 1.5717415115005476e-05, "loss": 2.7002, "step": 3230 }, { "epoch": 2.08, "learning_rate": 1.56078860898138e-05, "loss": 2.641, "step": 3240 }, { "epoch": 2.09, "learning_rate": 1.5498357064622125e-05, "loss": 2.6985, "step": 3250 }, { "epoch": 2.1, "learning_rate": 1.538882803943045e-05, "loss": 2.6769, "step": 3260 }, { "epoch": 2.1, "learning_rate": 1.5279299014238775e-05, "loss": 2.6481, "step": 3270 }, { "epoch": 2.11, "learning_rate": 1.5169769989047098e-05, "loss": 2.6592, "step": 3280 }, { "epoch": 2.12, "learning_rate": 1.5060240963855424e-05, "loss": 2.6225, "step": 3290 }, { "epoch": 2.12, "learning_rate": 1.4950711938663747e-05, "loss": 2.661, "step": 3300 }, { "epoch": 2.13, "learning_rate": 1.484118291347207e-05, "loss": 2.6897, "step": 3310 }, { "epoch": 2.13, "learning_rate": 1.4731653888280394e-05, "loss": 2.6965, "step": 3320 }, { "epoch": 2.14, "learning_rate": 1.4622124863088718e-05, "loss": 2.6726, "step": 3330 }, { "epoch": 2.15, "learning_rate": 1.4512595837897045e-05, "loss": 2.6131, "step": 3340 }, { "epoch": 2.15, "learning_rate": 1.4403066812705368e-05, "loss": 2.6998, "step": 3350 }, { "epoch": 2.16, "learning_rate": 1.4293537787513691e-05, "loss": 2.6998, "step": 3360 }, { "epoch": 2.17, "learning_rate": 1.4184008762322016e-05, "loss": 2.6854, "step": 3370 }, { "epoch": 2.17, "learning_rate": 1.4074479737130339e-05, "loss": 2.681, "step": 3380 }, { "epoch": 2.18, "learning_rate": 1.3964950711938666e-05, "loss": 2.6531, "step": 3390 }, { "epoch": 2.19, "learning_rate": 1.3855421686746989e-05, "loss": 2.7184, "step": 3400 }, { "epoch": 2.19, "learning_rate": 1.3745892661555312e-05, "loss": 2.7018, "step": 3410 }, { "epoch": 2.2, "learning_rate": 1.3636363636363637e-05, "loss": 2.6654, "step": 3420 }, { "epoch": 2.21, "learning_rate": 1.3526834611171963e-05, "loss": 2.6593, "step": 3430 }, { "epoch": 2.21, "learning_rate": 1.3417305585980286e-05, "loss": 2.6994, "step": 3440 }, { "epoch": 2.22, "learning_rate": 1.330777656078861e-05, "loss": 2.653, "step": 3450 }, { "epoch": 2.22, "learning_rate": 1.3198247535596932e-05, "loss": 2.6511, "step": 3460 }, { "epoch": 2.23, "learning_rate": 1.3088718510405257e-05, "loss": 2.6701, "step": 3470 }, { "epoch": 2.24, "learning_rate": 1.2979189485213584e-05, "loss": 2.6723, "step": 3480 }, { "epoch": 2.24, "learning_rate": 1.2869660460021907e-05, "loss": 2.6456, "step": 3490 }, { "epoch": 2.25, "learning_rate": 1.276013143483023e-05, "loss": 2.6542, "step": 3500 }, { "epoch": 2.26, "learning_rate": 1.2650602409638555e-05, "loss": 2.6936, "step": 3510 }, { "epoch": 2.26, "learning_rate": 1.2541073384446881e-05, "loss": 2.6799, "step": 3520 }, { "epoch": 2.27, "learning_rate": 1.2431544359255204e-05, "loss": 2.6849, "step": 3530 }, { "epoch": 2.28, "learning_rate": 1.2322015334063527e-05, "loss": 2.692, "step": 3540 }, { "epoch": 2.28, "learning_rate": 1.221248630887185e-05, "loss": 2.6228, "step": 3550 }, { "epoch": 2.29, "learning_rate": 1.2102957283680175e-05, "loss": 2.7349, "step": 3560 }, { "epoch": 2.3, "learning_rate": 1.19934282584885e-05, "loss": 2.6931, "step": 3570 }, { "epoch": 2.3, "learning_rate": 1.1883899233296825e-05, "loss": 2.6742, "step": 3580 }, { "epoch": 2.31, "learning_rate": 1.1774370208105148e-05, "loss": 2.685, "step": 3590 }, { "epoch": 2.31, "learning_rate": 1.1664841182913473e-05, "loss": 2.7104, "step": 3600 }, { "epoch": 2.32, "learning_rate": 1.1555312157721798e-05, "loss": 2.6751, "step": 3610 }, { "epoch": 2.33, "learning_rate": 1.144578313253012e-05, "loss": 2.6685, "step": 3620 }, { "epoch": 2.33, "learning_rate": 1.1336254107338445e-05, "loss": 2.6653, "step": 3630 }, { "epoch": 2.34, "learning_rate": 1.1226725082146769e-05, "loss": 2.7115, "step": 3640 }, { "epoch": 2.35, "learning_rate": 1.1117196056955093e-05, "loss": 2.702, "step": 3650 }, { "epoch": 2.35, "learning_rate": 1.1007667031763418e-05, "loss": 2.6685, "step": 3660 }, { "epoch": 2.36, "learning_rate": 1.0898138006571743e-05, "loss": 2.6765, "step": 3670 }, { "epoch": 2.37, "learning_rate": 1.0788608981380066e-05, "loss": 2.6778, "step": 3680 }, { "epoch": 2.37, "learning_rate": 1.0679079956188389e-05, "loss": 2.6618, "step": 3690 }, { "epoch": 2.38, "learning_rate": 1.0569550930996714e-05, "loss": 2.6208, "step": 3700 }, { "epoch": 2.39, "learning_rate": 1.0460021905805039e-05, "loss": 2.6446, "step": 3710 }, { "epoch": 2.39, "learning_rate": 1.0350492880613364e-05, "loss": 2.7144, "step": 3720 }, { "epoch": 2.4, "learning_rate": 1.0240963855421687e-05, "loss": 2.6631, "step": 3730 }, { "epoch": 2.4, "learning_rate": 1.0131434830230011e-05, "loss": 2.6769, "step": 3740 }, { "epoch": 2.41, "learning_rate": 1.0021905805038336e-05, "loss": 2.7273, "step": 3750 }, { "epoch": 2.42, "learning_rate": 9.912376779846661e-06, "loss": 2.6302, "step": 3760 }, { "epoch": 2.42, "learning_rate": 9.802847754654984e-06, "loss": 2.6877, "step": 3770 }, { "epoch": 2.43, "learning_rate": 9.693318729463307e-06, "loss": 2.658, "step": 3780 }, { "epoch": 2.44, "learning_rate": 9.583789704271632e-06, "loss": 2.7129, "step": 3790 }, { "epoch": 2.44, "learning_rate": 9.474260679079957e-06, "loss": 2.7103, "step": 3800 }, { "epoch": 2.45, "learning_rate": 9.364731653888282e-06, "loss": 2.6904, "step": 3810 }, { "epoch": 2.46, "learning_rate": 9.255202628696605e-06, "loss": 2.6706, "step": 3820 }, { "epoch": 2.46, "learning_rate": 9.14567360350493e-06, "loss": 2.6615, "step": 3830 }, { "epoch": 2.47, "learning_rate": 9.036144578313253e-06, "loss": 2.7039, "step": 3840 }, { "epoch": 2.48, "learning_rate": 8.926615553121577e-06, "loss": 2.7451, "step": 3850 }, { "epoch": 2.48, "learning_rate": 8.817086527929902e-06, "loss": 2.6681, "step": 3860 }, { "epoch": 2.49, "learning_rate": 8.707557502738225e-06, "loss": 2.6745, "step": 3870 }, { "epoch": 2.49, "learning_rate": 8.59802847754655e-06, "loss": 2.707, "step": 3880 }, { "epoch": 2.5, "learning_rate": 8.488499452354875e-06, "loss": 2.672, "step": 3890 }, { "epoch": 2.51, "learning_rate": 8.3789704271632e-06, "loss": 2.647, "step": 3900 }, { "epoch": 2.51, "learning_rate": 8.269441401971523e-06, "loss": 2.6512, "step": 3910 }, { "epoch": 2.52, "learning_rate": 8.159912376779848e-06, "loss": 2.6808, "step": 3920 }, { "epoch": 2.53, "learning_rate": 8.05038335158817e-06, "loss": 2.6612, "step": 3930 }, { "epoch": 2.53, "learning_rate": 7.940854326396495e-06, "loss": 2.6702, "step": 3940 }, { "epoch": 2.54, "learning_rate": 7.83132530120482e-06, "loss": 2.661, "step": 3950 }, { "epoch": 2.55, "learning_rate": 7.721796276013143e-06, "loss": 2.716, "step": 3960 }, { "epoch": 2.55, "learning_rate": 7.612267250821468e-06, "loss": 2.6917, "step": 3970 }, { "epoch": 2.56, "learning_rate": 7.502738225629792e-06, "loss": 2.6875, "step": 3980 }, { "epoch": 2.57, "learning_rate": 7.393209200438117e-06, "loss": 2.704, "step": 3990 }, { "epoch": 2.57, "learning_rate": 7.283680175246441e-06, "loss": 2.6584, "step": 4000 }, { "epoch": 2.58, "learning_rate": 7.174151150054764e-06, "loss": 2.6978, "step": 4010 }, { "epoch": 2.58, "learning_rate": 7.06462212486309e-06, "loss": 2.6987, "step": 4020 }, { "epoch": 2.59, "learning_rate": 6.955093099671413e-06, "loss": 2.6952, "step": 4030 }, { "epoch": 2.6, "learning_rate": 6.8455640744797375e-06, "loss": 2.6737, "step": 4040 }, { "epoch": 2.6, "learning_rate": 6.7360350492880615e-06, "loss": 2.6625, "step": 4050 }, { "epoch": 2.61, "learning_rate": 6.626506024096386e-06, "loss": 2.6267, "step": 4060 }, { "epoch": 2.62, "learning_rate": 6.51697699890471e-06, "loss": 2.6574, "step": 4070 }, { "epoch": 2.62, "learning_rate": 6.407447973713033e-06, "loss": 2.6545, "step": 4080 }, { "epoch": 2.63, "learning_rate": 6.297918948521359e-06, "loss": 2.687, "step": 4090 }, { "epoch": 2.64, "learning_rate": 6.188389923329683e-06, "loss": 2.7007, "step": 4100 }, { "epoch": 2.64, "learning_rate": 6.078860898138007e-06, "loss": 2.6227, "step": 4110 }, { "epoch": 2.65, "learning_rate": 5.969331872946331e-06, "loss": 2.6677, "step": 4120 }, { "epoch": 2.66, "learning_rate": 5.859802847754655e-06, "loss": 2.6416, "step": 4130 }, { "epoch": 2.66, "learning_rate": 5.7502738225629795e-06, "loss": 2.681, "step": 4140 }, { "epoch": 2.67, "learning_rate": 5.6407447973713035e-06, "loss": 2.6312, "step": 4150 }, { "epoch": 2.67, "learning_rate": 5.531215772179628e-06, "loss": 2.6801, "step": 4160 }, { "epoch": 2.68, "learning_rate": 5.421686746987952e-06, "loss": 2.6593, "step": 4170 }, { "epoch": 2.69, "learning_rate": 5.312157721796276e-06, "loss": 2.6815, "step": 4180 }, { "epoch": 2.69, "learning_rate": 5.202628696604601e-06, "loss": 2.618, "step": 4190 }, { "epoch": 2.7, "learning_rate": 5.093099671412924e-06, "loss": 2.6588, "step": 4200 }, { "epoch": 2.71, "learning_rate": 4.983570646221249e-06, "loss": 2.6882, "step": 4210 }, { "epoch": 2.71, "learning_rate": 4.874041621029573e-06, "loss": 2.6454, "step": 4220 }, { "epoch": 2.72, "learning_rate": 4.764512595837898e-06, "loss": 2.6749, "step": 4230 }, { "epoch": 2.73, "learning_rate": 4.6549835706462216e-06, "loss": 2.6714, "step": 4240 }, { "epoch": 2.73, "learning_rate": 4.5454545454545455e-06, "loss": 2.6335, "step": 4250 }, { "epoch": 2.74, "learning_rate": 4.43592552026287e-06, "loss": 2.6734, "step": 4260 }, { "epoch": 2.75, "learning_rate": 4.326396495071194e-06, "loss": 2.6764, "step": 4270 }, { "epoch": 2.75, "learning_rate": 4.216867469879518e-06, "loss": 2.6899, "step": 4280 }, { "epoch": 2.76, "learning_rate": 4.107338444687842e-06, "loss": 2.7273, "step": 4290 }, { "epoch": 2.76, "learning_rate": 3.997809419496167e-06, "loss": 2.6337, "step": 4300 }, { "epoch": 2.77, "learning_rate": 3.888280394304491e-06, "loss": 2.6679, "step": 4310 }, { "epoch": 2.78, "learning_rate": 3.7787513691128153e-06, "loss": 2.6505, "step": 4320 }, { "epoch": 2.78, "learning_rate": 3.6692223439211392e-06, "loss": 2.6732, "step": 4330 }, { "epoch": 2.79, "learning_rate": 3.5596933187294636e-06, "loss": 2.6806, "step": 4340 }, { "epoch": 2.8, "learning_rate": 3.450164293537788e-06, "loss": 2.6498, "step": 4350 }, { "epoch": 2.8, "learning_rate": 3.3406352683461115e-06, "loss": 2.6626, "step": 4360 }, { "epoch": 2.81, "learning_rate": 3.231106243154436e-06, "loss": 2.6762, "step": 4370 }, { "epoch": 2.82, "learning_rate": 3.1215772179627602e-06, "loss": 2.6767, "step": 4380 }, { "epoch": 2.82, "learning_rate": 3.0120481927710846e-06, "loss": 2.6712, "step": 4390 }, { "epoch": 2.83, "learning_rate": 2.9025191675794086e-06, "loss": 2.6342, "step": 4400 }, { "epoch": 2.84, "learning_rate": 2.792990142387733e-06, "loss": 2.7124, "step": 4410 }, { "epoch": 2.84, "learning_rate": 2.683461117196057e-06, "loss": 2.6639, "step": 4420 }, { "epoch": 2.85, "learning_rate": 2.5739320920043813e-06, "loss": 2.6494, "step": 4430 }, { "epoch": 2.85, "learning_rate": 2.4644030668127056e-06, "loss": 2.6704, "step": 4440 }, { "epoch": 2.86, "learning_rate": 2.35487404162103e-06, "loss": 2.67, "step": 4450 }, { "epoch": 2.87, "learning_rate": 2.245345016429354e-06, "loss": 2.66, "step": 4460 }, { "epoch": 2.87, "learning_rate": 2.135815991237678e-06, "loss": 2.6743, "step": 4470 }, { "epoch": 2.88, "learning_rate": 2.0262869660460023e-06, "loss": 2.6514, "step": 4480 }, { "epoch": 2.89, "learning_rate": 1.9167579408543262e-06, "loss": 2.6848, "step": 4490 }, { "epoch": 2.89, "learning_rate": 1.8072289156626506e-06, "loss": 2.7078, "step": 4500 }, { "epoch": 2.9, "learning_rate": 1.697699890470975e-06, "loss": 2.6466, "step": 4510 }, { "epoch": 2.91, "learning_rate": 1.5881708652792991e-06, "loss": 2.6073, "step": 4520 }, { "epoch": 2.91, "learning_rate": 1.4786418400876233e-06, "loss": 2.7111, "step": 4530 }, { "epoch": 2.92, "learning_rate": 1.3691128148959475e-06, "loss": 2.688, "step": 4540 }, { "epoch": 2.93, "learning_rate": 1.2595837897042716e-06, "loss": 2.6719, "step": 4550 }, { "epoch": 2.93, "learning_rate": 1.1500547645125958e-06, "loss": 2.638, "step": 4560 }, { "epoch": 2.94, "learning_rate": 1.0405257393209202e-06, "loss": 2.6686, "step": 4570 }, { "epoch": 2.94, "learning_rate": 9.309967141292442e-07, "loss": 2.7157, "step": 4580 }, { "epoch": 2.95, "learning_rate": 8.214676889375685e-07, "loss": 2.6681, "step": 4590 }, { "epoch": 2.96, "learning_rate": 7.119386637458927e-07, "loss": 2.6724, "step": 4600 }, { "epoch": 2.96, "learning_rate": 6.024096385542169e-07, "loss": 2.6502, "step": 4610 }, { "epoch": 2.97, "learning_rate": 4.928806133625412e-07, "loss": 2.6625, "step": 4620 }, { "epoch": 2.98, "learning_rate": 3.833515881708653e-07, "loss": 2.6823, "step": 4630 }, { "epoch": 2.98, "learning_rate": 2.738225629791895e-07, "loss": 2.6877, "step": 4640 }, { "epoch": 2.99, "learning_rate": 1.642935377875137e-07, "loss": 2.6878, "step": 4650 }, { "epoch": 3.0, "learning_rate": 5.47645125958379e-08, "loss": 2.7003, "step": 4660 }, { "epoch": 3.0, "step": 4665, "total_flos": 2.0794722991315354e+17, "train_loss": 1.1010990204151803, "train_runtime": 40118.9781, "train_samples_per_second": 1.396, "train_steps_per_second": 0.116 } ], "max_steps": 4665, "num_train_epochs": 3, "total_flos": 2.0794722991315354e+17, "trial_name": null, "trial_params": null }