{ "best_metric": null, "best_model_checkpoint": null, "epoch": 9.690504807692308, "global_step": 64500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.02, "learning_rate": 9.984975961538462e-05, "loss": 9.5457, "step": 100 }, { "epoch": 0.03, "learning_rate": 9.969951923076923e-05, "loss": 6.3223, "step": 200 }, { "epoch": 0.05, "learning_rate": 9.954927884615385e-05, "loss": 5.9118, "step": 300 }, { "epoch": 0.06, "learning_rate": 9.939903846153847e-05, "loss": 5.6779, "step": 400 }, { "epoch": 0.08, "learning_rate": 9.924879807692308e-05, "loss": 5.4467, "step": 500 }, { "epoch": 0.09, "learning_rate": 9.909855769230769e-05, "loss": 5.3733, "step": 600 }, { "epoch": 0.11, "learning_rate": 9.894831730769232e-05, "loss": 5.3544, "step": 700 }, { "epoch": 0.12, "learning_rate": 9.879807692307693e-05, "loss": 5.2784, "step": 800 }, { "epoch": 0.14, "learning_rate": 9.864783653846154e-05, "loss": 5.1503, "step": 900 }, { "epoch": 0.15, "learning_rate": 9.849759615384617e-05, "loss": 5.0107, "step": 1000 }, { "epoch": 0.17, "learning_rate": 9.834735576923078e-05, "loss": 5.1408, "step": 1100 }, { "epoch": 0.18, "learning_rate": 9.819711538461539e-05, "loss": 4.9742, "step": 1200 }, { "epoch": 0.2, "learning_rate": 9.8046875e-05, "loss": 4.8672, "step": 1300 }, { "epoch": 0.21, "learning_rate": 9.789663461538462e-05, "loss": 4.858, "step": 1400 }, { "epoch": 0.23, "learning_rate": 9.774639423076923e-05, "loss": 4.8912, "step": 1500 }, { "epoch": 0.24, "learning_rate": 9.759615384615385e-05, "loss": 4.8654, "step": 1600 }, { "epoch": 0.26, "learning_rate": 9.744591346153847e-05, "loss": 4.8192, "step": 1700 }, { "epoch": 0.27, "learning_rate": 9.729567307692308e-05, "loss": 4.7405, "step": 1800 }, { "epoch": 0.29, "learning_rate": 9.714543269230769e-05, "loss": 4.69, "step": 1900 }, { "epoch": 0.3, "learning_rate": 9.699519230769232e-05, "loss": 4.5945, "step": 2000 }, { "epoch": 0.32, "learning_rate": 9.684495192307693e-05, "loss": 4.621, "step": 2100 }, { "epoch": 0.33, "learning_rate": 9.669471153846155e-05, "loss": 4.68, "step": 2200 }, { "epoch": 0.35, "learning_rate": 9.654447115384616e-05, "loss": 4.5514, "step": 2300 }, { "epoch": 0.36, "learning_rate": 9.639423076923078e-05, "loss": 4.6192, "step": 2400 }, { "epoch": 0.38, "learning_rate": 9.624399038461539e-05, "loss": 4.4672, "step": 2500 }, { "epoch": 0.39, "learning_rate": 9.609375e-05, "loss": 4.5006, "step": 2600 }, { "epoch": 0.41, "learning_rate": 9.594350961538461e-05, "loss": 4.5668, "step": 2700 }, { "epoch": 0.42, "learning_rate": 9.579326923076924e-05, "loss": 4.5345, "step": 2800 }, { "epoch": 0.44, "learning_rate": 9.564302884615385e-05, "loss": 4.4806, "step": 2900 }, { "epoch": 0.45, "learning_rate": 9.549278846153846e-05, "loss": 4.3853, "step": 3000 }, { "epoch": 0.47, "learning_rate": 9.534254807692308e-05, "loss": 4.4482, "step": 3100 }, { "epoch": 0.48, "learning_rate": 9.519230769230769e-05, "loss": 4.4437, "step": 3200 }, { "epoch": 0.5, "learning_rate": 9.504206730769231e-05, "loss": 4.3163, "step": 3300 }, { "epoch": 0.51, "learning_rate": 9.489182692307693e-05, "loss": 4.3702, "step": 3400 }, { "epoch": 0.53, "learning_rate": 9.474158653846155e-05, "loss": 4.3657, "step": 3500 }, { "epoch": 0.54, "learning_rate": 9.459134615384616e-05, "loss": 4.4059, "step": 3600 }, { "epoch": 0.56, "learning_rate": 9.444110576923078e-05, "loss": 4.3584, "step": 3700 }, { "epoch": 0.57, "learning_rate": 9.429086538461539e-05, "loss": 4.2906, "step": 3800 }, { "epoch": 0.59, "learning_rate": 9.4140625e-05, "loss": 4.2862, "step": 3900 }, { "epoch": 0.6, "learning_rate": 9.399038461538463e-05, "loss": 4.3426, "step": 4000 }, { "epoch": 0.62, "learning_rate": 9.384014423076924e-05, "loss": 4.2767, "step": 4100 }, { "epoch": 0.63, "learning_rate": 9.368990384615385e-05, "loss": 4.3542, "step": 4200 }, { "epoch": 0.65, "learning_rate": 9.353966346153846e-05, "loss": 4.3341, "step": 4300 }, { "epoch": 0.66, "learning_rate": 9.338942307692307e-05, "loss": 4.2464, "step": 4400 }, { "epoch": 0.68, "learning_rate": 9.323918269230769e-05, "loss": 4.3169, "step": 4500 }, { "epoch": 0.69, "learning_rate": 9.308894230769231e-05, "loss": 4.1644, "step": 4600 }, { "epoch": 0.71, "learning_rate": 9.293870192307694e-05, "loss": 4.3213, "step": 4700 }, { "epoch": 0.72, "learning_rate": 9.278846153846155e-05, "loss": 4.2788, "step": 4800 }, { "epoch": 0.74, "learning_rate": 9.263822115384616e-05, "loss": 4.2807, "step": 4900 }, { "epoch": 0.75, "learning_rate": 9.248798076923077e-05, "loss": 4.1281, "step": 5000 }, { "epoch": 0.77, "learning_rate": 9.233774038461539e-05, "loss": 4.2481, "step": 5100 }, { "epoch": 0.78, "learning_rate": 9.21875e-05, "loss": 4.163, "step": 5200 }, { "epoch": 0.8, "learning_rate": 9.203725961538462e-05, "loss": 4.2567, "step": 5300 }, { "epoch": 0.81, "learning_rate": 9.188701923076924e-05, "loss": 4.2224, "step": 5400 }, { "epoch": 0.83, "learning_rate": 9.173677884615385e-05, "loss": 4.132, "step": 5500 }, { "epoch": 0.84, "learning_rate": 9.158653846153846e-05, "loss": 4.1601, "step": 5600 }, { "epoch": 0.86, "learning_rate": 9.143629807692307e-05, "loss": 4.0697, "step": 5700 }, { "epoch": 0.87, "learning_rate": 9.12860576923077e-05, "loss": 4.1043, "step": 5800 }, { "epoch": 0.89, "learning_rate": 9.113581730769232e-05, "loss": 4.1087, "step": 5900 }, { "epoch": 0.9, "learning_rate": 9.098557692307694e-05, "loss": 3.9814, "step": 6000 }, { "epoch": 0.92, "learning_rate": 9.083533653846155e-05, "loss": 4.0993, "step": 6100 }, { "epoch": 0.93, "learning_rate": 9.068509615384616e-05, "loss": 4.1064, "step": 6200 }, { "epoch": 0.95, "learning_rate": 9.053485576923077e-05, "loss": 4.065, "step": 6300 }, { "epoch": 0.96, "learning_rate": 9.038461538461538e-05, "loss": 4.0022, "step": 6400 }, { "epoch": 0.98, "learning_rate": 9.023437500000001e-05, "loss": 4.0968, "step": 6500 }, { "epoch": 0.99, "learning_rate": 9.008413461538462e-05, "loss": 4.1214, "step": 6600 }, { "epoch": 1.01, "learning_rate": 8.993389423076923e-05, "loss": 4.0841, "step": 6700 }, { "epoch": 1.02, "learning_rate": 8.978365384615385e-05, "loss": 4.0666, "step": 6800 }, { "epoch": 1.04, "learning_rate": 8.963341346153846e-05, "loss": 3.9632, "step": 6900 }, { "epoch": 1.05, "learning_rate": 8.948317307692307e-05, "loss": 3.9496, "step": 7000 }, { "epoch": 1.07, "learning_rate": 8.93329326923077e-05, "loss": 3.9757, "step": 7100 }, { "epoch": 1.08, "learning_rate": 8.918269230769232e-05, "loss": 3.9959, "step": 7200 }, { "epoch": 1.1, "learning_rate": 8.903245192307693e-05, "loss": 3.9594, "step": 7300 }, { "epoch": 1.11, "learning_rate": 8.888221153846155e-05, "loss": 3.9735, "step": 7400 }, { "epoch": 1.13, "learning_rate": 8.873197115384616e-05, "loss": 3.9647, "step": 7500 }, { "epoch": 1.14, "learning_rate": 8.858173076923077e-05, "loss": 3.8759, "step": 7600 }, { "epoch": 1.16, "learning_rate": 8.84314903846154e-05, "loss": 3.9956, "step": 7700 }, { "epoch": 1.17, "learning_rate": 8.828125000000001e-05, "loss": 3.9526, "step": 7800 }, { "epoch": 1.19, "learning_rate": 8.813100961538462e-05, "loss": 3.9409, "step": 7900 }, { "epoch": 1.2, "learning_rate": 8.798076923076923e-05, "loss": 3.9545, "step": 8000 }, { "epoch": 1.22, "learning_rate": 8.783052884615384e-05, "loss": 3.8668, "step": 8100 }, { "epoch": 1.23, "learning_rate": 8.768028846153846e-05, "loss": 3.8675, "step": 8200 }, { "epoch": 1.25, "learning_rate": 8.753004807692308e-05, "loss": 3.9234, "step": 8300 }, { "epoch": 1.26, "learning_rate": 8.73798076923077e-05, "loss": 3.9181, "step": 8400 }, { "epoch": 1.28, "learning_rate": 8.722956730769232e-05, "loss": 3.9288, "step": 8500 }, { "epoch": 1.29, "learning_rate": 8.707932692307693e-05, "loss": 3.9103, "step": 8600 }, { "epoch": 1.31, "learning_rate": 8.692908653846154e-05, "loss": 3.9452, "step": 8700 }, { "epoch": 1.32, "learning_rate": 8.677884615384616e-05, "loss": 3.8211, "step": 8800 }, { "epoch": 1.34, "learning_rate": 8.662860576923077e-05, "loss": 3.7997, "step": 8900 }, { "epoch": 1.35, "learning_rate": 8.64783653846154e-05, "loss": 3.8444, "step": 9000 }, { "epoch": 1.37, "learning_rate": 8.6328125e-05, "loss": 3.8875, "step": 9100 }, { "epoch": 1.38, "learning_rate": 8.617788461538462e-05, "loss": 3.8792, "step": 9200 }, { "epoch": 1.4, "learning_rate": 8.602764423076923e-05, "loss": 3.8994, "step": 9300 }, { "epoch": 1.41, "learning_rate": 8.587740384615384e-05, "loss": 3.8862, "step": 9400 }, { "epoch": 1.43, "learning_rate": 8.572716346153847e-05, "loss": 3.824, "step": 9500 }, { "epoch": 1.44, "learning_rate": 8.557692307692308e-05, "loss": 3.7989, "step": 9600 }, { "epoch": 1.46, "learning_rate": 8.542668269230769e-05, "loss": 3.8998, "step": 9700 }, { "epoch": 1.47, "learning_rate": 8.527644230769232e-05, "loss": 3.8514, "step": 9800 }, { "epoch": 1.49, "learning_rate": 8.512620192307693e-05, "loss": 3.8324, "step": 9900 }, { "epoch": 1.5, "learning_rate": 8.497596153846154e-05, "loss": 3.8852, "step": 10000 }, { "epoch": 1.52, "learning_rate": 8.482572115384615e-05, "loss": 3.7941, "step": 10100 }, { "epoch": 1.53, "learning_rate": 8.467548076923078e-05, "loss": 3.9133, "step": 10200 }, { "epoch": 1.55, "learning_rate": 8.452524038461539e-05, "loss": 3.7357, "step": 10300 }, { "epoch": 1.56, "learning_rate": 8.4375e-05, "loss": 3.7986, "step": 10400 }, { "epoch": 1.58, "learning_rate": 8.422475961538462e-05, "loss": 3.8262, "step": 10500 }, { "epoch": 1.59, "learning_rate": 8.407451923076923e-05, "loss": 3.888, "step": 10600 }, { "epoch": 1.61, "learning_rate": 8.392427884615384e-05, "loss": 3.7276, "step": 10700 }, { "epoch": 1.62, "learning_rate": 8.377403846153847e-05, "loss": 3.8952, "step": 10800 }, { "epoch": 1.64, "learning_rate": 8.362379807692308e-05, "loss": 3.89, "step": 10900 }, { "epoch": 1.65, "learning_rate": 8.347355769230769e-05, "loss": 3.8234, "step": 11000 }, { "epoch": 1.67, "learning_rate": 8.332331730769232e-05, "loss": 3.7485, "step": 11100 }, { "epoch": 1.68, "learning_rate": 8.317307692307693e-05, "loss": 3.7877, "step": 11200 }, { "epoch": 1.7, "learning_rate": 8.302283653846154e-05, "loss": 3.7776, "step": 11300 }, { "epoch": 1.71, "learning_rate": 8.287259615384617e-05, "loss": 3.7945, "step": 11400 }, { "epoch": 1.73, "learning_rate": 8.272235576923078e-05, "loss": 3.8574, "step": 11500 }, { "epoch": 1.74, "learning_rate": 8.257211538461539e-05, "loss": 3.7562, "step": 11600 }, { "epoch": 1.76, "learning_rate": 8.2421875e-05, "loss": 3.7713, "step": 11700 }, { "epoch": 1.77, "learning_rate": 8.227163461538461e-05, "loss": 3.7257, "step": 11800 }, { "epoch": 1.79, "learning_rate": 8.212139423076923e-05, "loss": 3.7542, "step": 11900 }, { "epoch": 1.8, "learning_rate": 8.197115384615385e-05, "loss": 3.6902, "step": 12000 }, { "epoch": 1.82, "learning_rate": 8.182091346153846e-05, "loss": 3.7354, "step": 12100 }, { "epoch": 1.83, "learning_rate": 8.167067307692308e-05, "loss": 3.745, "step": 12200 }, { "epoch": 1.85, "learning_rate": 8.152043269230769e-05, "loss": 3.7681, "step": 12300 }, { "epoch": 1.86, "learning_rate": 8.137019230769231e-05, "loss": 3.8431, "step": 12400 }, { "epoch": 1.88, "learning_rate": 8.121995192307693e-05, "loss": 3.7285, "step": 12500 }, { "epoch": 1.89, "learning_rate": 8.106971153846155e-05, "loss": 3.6801, "step": 12600 }, { "epoch": 1.91, "learning_rate": 8.091947115384616e-05, "loss": 3.6456, "step": 12700 }, { "epoch": 1.92, "learning_rate": 8.076923076923078e-05, "loss": 3.764, "step": 12800 }, { "epoch": 1.94, "learning_rate": 8.061899038461539e-05, "loss": 3.7537, "step": 12900 }, { "epoch": 1.95, "learning_rate": 8.046875e-05, "loss": 3.5876, "step": 13000 }, { "epoch": 1.97, "learning_rate": 8.031850961538461e-05, "loss": 3.8075, "step": 13100 }, { "epoch": 1.98, "learning_rate": 8.016826923076924e-05, "loss": 3.6953, "step": 13200 }, { "epoch": 2.0, "learning_rate": 8.001802884615385e-05, "loss": 3.6577, "step": 13300 }, { "epoch": 2.01, "learning_rate": 7.986778846153846e-05, "loss": 3.7064, "step": 13400 }, { "epoch": 2.03, "learning_rate": 7.971754807692308e-05, "loss": 3.6817, "step": 13500 }, { "epoch": 2.04, "learning_rate": 7.956730769230769e-05, "loss": 3.6729, "step": 13600 }, { "epoch": 2.06, "learning_rate": 7.941706730769231e-05, "loss": 3.6737, "step": 13700 }, { "epoch": 2.07, "learning_rate": 7.926682692307693e-05, "loss": 3.6222, "step": 13800 }, { "epoch": 2.09, "learning_rate": 7.911658653846155e-05, "loss": 3.6704, "step": 13900 }, { "epoch": 2.1, "learning_rate": 7.896634615384616e-05, "loss": 3.7445, "step": 14000 }, { "epoch": 2.12, "learning_rate": 7.881610576923078e-05, "loss": 3.6227, "step": 14100 }, { "epoch": 2.13, "learning_rate": 7.866586538461539e-05, "loss": 3.6822, "step": 14200 }, { "epoch": 2.15, "learning_rate": 7.8515625e-05, "loss": 3.6178, "step": 14300 }, { "epoch": 2.16, "learning_rate": 7.836538461538462e-05, "loss": 3.6329, "step": 14400 }, { "epoch": 2.18, "learning_rate": 7.821514423076924e-05, "loss": 3.627, "step": 14500 }, { "epoch": 2.19, "learning_rate": 7.806490384615385e-05, "loss": 3.7061, "step": 14600 }, { "epoch": 2.21, "learning_rate": 7.791466346153846e-05, "loss": 3.7037, "step": 14700 }, { "epoch": 2.22, "learning_rate": 7.776442307692307e-05, "loss": 3.5797, "step": 14800 }, { "epoch": 2.24, "learning_rate": 7.761418269230769e-05, "loss": 3.6865, "step": 14900 }, { "epoch": 2.25, "learning_rate": 7.746394230769231e-05, "loss": 3.7016, "step": 15000 }, { "epoch": 2.27, "learning_rate": 7.731370192307694e-05, "loss": 3.6149, "step": 15100 }, { "epoch": 2.28, "learning_rate": 7.716346153846155e-05, "loss": 3.614, "step": 15200 }, { "epoch": 2.3, "learning_rate": 7.701322115384616e-05, "loss": 3.5511, "step": 15300 }, { "epoch": 2.31, "learning_rate": 7.686298076923077e-05, "loss": 3.5887, "step": 15400 }, { "epoch": 2.33, "learning_rate": 7.671274038461539e-05, "loss": 3.6247, "step": 15500 }, { "epoch": 2.34, "learning_rate": 7.65625e-05, "loss": 3.5886, "step": 15600 }, { "epoch": 2.36, "learning_rate": 7.641225961538462e-05, "loss": 3.4512, "step": 15700 }, { "epoch": 2.37, "learning_rate": 7.626201923076924e-05, "loss": 3.4764, "step": 15800 }, { "epoch": 2.39, "learning_rate": 7.611177884615385e-05, "loss": 3.6115, "step": 15900 }, { "epoch": 2.4, "learning_rate": 7.596153846153846e-05, "loss": 3.5766, "step": 16000 }, { "epoch": 2.42, "learning_rate": 7.581129807692307e-05, "loss": 3.6717, "step": 16100 }, { "epoch": 2.43, "learning_rate": 7.56610576923077e-05, "loss": 3.4975, "step": 16200 }, { "epoch": 2.45, "learning_rate": 7.551081730769232e-05, "loss": 3.572, "step": 16300 }, { "epoch": 2.46, "learning_rate": 7.536057692307694e-05, "loss": 3.5285, "step": 16400 }, { "epoch": 2.48, "learning_rate": 7.521033653846155e-05, "loss": 3.634, "step": 16500 }, { "epoch": 2.49, "learning_rate": 7.506009615384616e-05, "loss": 3.5654, "step": 16600 }, { "epoch": 2.51, "learning_rate": 7.490985576923077e-05, "loss": 3.5617, "step": 16700 }, { "epoch": 2.52, "learning_rate": 7.475961538461538e-05, "loss": 3.6074, "step": 16800 }, { "epoch": 2.54, "learning_rate": 7.460937500000001e-05, "loss": 3.5804, "step": 16900 }, { "epoch": 2.55, "learning_rate": 7.445913461538462e-05, "loss": 3.5686, "step": 17000 }, { "epoch": 2.57, "learning_rate": 7.430889423076923e-05, "loss": 3.5895, "step": 17100 }, { "epoch": 2.58, "learning_rate": 7.415865384615385e-05, "loss": 3.5292, "step": 17200 }, { "epoch": 2.6, "learning_rate": 7.400841346153846e-05, "loss": 3.553, "step": 17300 }, { "epoch": 2.61, "learning_rate": 7.385817307692307e-05, "loss": 3.6004, "step": 17400 }, { "epoch": 2.63, "learning_rate": 7.37079326923077e-05, "loss": 3.5905, "step": 17500 }, { "epoch": 2.64, "learning_rate": 7.355769230769232e-05, "loss": 3.4882, "step": 17600 }, { "epoch": 2.66, "learning_rate": 7.340745192307693e-05, "loss": 3.7265, "step": 17700 }, { "epoch": 2.67, "learning_rate": 7.325721153846155e-05, "loss": 3.4666, "step": 17800 }, { "epoch": 2.69, "learning_rate": 7.310697115384616e-05, "loss": 3.5222, "step": 17900 }, { "epoch": 2.7, "learning_rate": 7.295673076923077e-05, "loss": 3.5472, "step": 18000 }, { "epoch": 2.72, "learning_rate": 7.28064903846154e-05, "loss": 3.4465, "step": 18100 }, { "epoch": 2.73, "learning_rate": 7.265625000000001e-05, "loss": 3.4855, "step": 18200 }, { "epoch": 2.75, "learning_rate": 7.250600961538462e-05, "loss": 3.4352, "step": 18300 }, { "epoch": 2.76, "learning_rate": 7.235576923076923e-05, "loss": 3.5702, "step": 18400 }, { "epoch": 2.78, "learning_rate": 7.220552884615384e-05, "loss": 3.5483, "step": 18500 }, { "epoch": 2.79, "learning_rate": 7.205528846153846e-05, "loss": 3.4681, "step": 18600 }, { "epoch": 2.81, "learning_rate": 7.190504807692308e-05, "loss": 3.4907, "step": 18700 }, { "epoch": 2.82, "learning_rate": 7.17548076923077e-05, "loss": 3.6045, "step": 18800 }, { "epoch": 2.84, "learning_rate": 7.160456730769232e-05, "loss": 3.4191, "step": 18900 }, { "epoch": 2.85, "learning_rate": 7.145432692307693e-05, "loss": 3.5984, "step": 19000 }, { "epoch": 2.87, "learning_rate": 7.130408653846154e-05, "loss": 3.4533, "step": 19100 }, { "epoch": 2.88, "learning_rate": 7.115384615384616e-05, "loss": 3.4745, "step": 19200 }, { "epoch": 2.9, "learning_rate": 7.100360576923077e-05, "loss": 3.4572, "step": 19300 }, { "epoch": 2.91, "learning_rate": 7.08533653846154e-05, "loss": 3.5035, "step": 19400 }, { "epoch": 2.93, "learning_rate": 7.0703125e-05, "loss": 3.5592, "step": 19500 }, { "epoch": 2.94, "learning_rate": 7.055288461538462e-05, "loss": 3.4138, "step": 19600 }, { "epoch": 2.96, "learning_rate": 7.040264423076923e-05, "loss": 3.4603, "step": 19700 }, { "epoch": 2.97, "learning_rate": 7.025240384615384e-05, "loss": 3.5511, "step": 19800 }, { "epoch": 2.99, "learning_rate": 7.010216346153847e-05, "loss": 3.4787, "step": 19900 }, { "epoch": 3.0, "learning_rate": 6.995192307692308e-05, "loss": 3.5999, "step": 20000 }, { "epoch": 3.02, "learning_rate": 6.980168269230769e-05, "loss": 3.4902, "step": 20100 }, { "epoch": 3.03, "learning_rate": 6.965144230769232e-05, "loss": 3.5083, "step": 20200 }, { "epoch": 3.05, "learning_rate": 6.950120192307693e-05, "loss": 3.5033, "step": 20300 }, { "epoch": 3.06, "learning_rate": 6.935096153846154e-05, "loss": 3.5604, "step": 20400 }, { "epoch": 3.08, "learning_rate": 6.920072115384615e-05, "loss": 3.5009, "step": 20500 }, { "epoch": 3.09, "learning_rate": 6.905048076923078e-05, "loss": 3.5076, "step": 20600 }, { "epoch": 3.11, "learning_rate": 6.890024038461539e-05, "loss": 3.4699, "step": 20700 }, { "epoch": 3.12, "learning_rate": 6.875e-05, "loss": 3.4902, "step": 20800 }, { "epoch": 3.14, "learning_rate": 6.859975961538462e-05, "loss": 3.4528, "step": 20900 }, { "epoch": 3.16, "learning_rate": 6.844951923076923e-05, "loss": 3.5253, "step": 21000 }, { "epoch": 3.17, "learning_rate": 6.829927884615384e-05, "loss": 3.48, "step": 21100 }, { "epoch": 3.19, "learning_rate": 6.814903846153847e-05, "loss": 3.3898, "step": 21200 }, { "epoch": 3.2, "learning_rate": 6.799879807692308e-05, "loss": 3.4181, "step": 21300 }, { "epoch": 3.22, "learning_rate": 6.784855769230769e-05, "loss": 3.4118, "step": 21400 }, { "epoch": 3.23, "learning_rate": 6.769831730769232e-05, "loss": 3.442, "step": 21500 }, { "epoch": 3.25, "learning_rate": 6.754807692307693e-05, "loss": 3.4115, "step": 21600 }, { "epoch": 3.26, "learning_rate": 6.739783653846154e-05, "loss": 3.4211, "step": 21700 }, { "epoch": 3.28, "learning_rate": 6.724759615384617e-05, "loss": 3.4422, "step": 21800 }, { "epoch": 3.29, "learning_rate": 6.709735576923078e-05, "loss": 3.5675, "step": 21900 }, { "epoch": 3.31, "learning_rate": 6.694711538461539e-05, "loss": 3.3725, "step": 22000 }, { "epoch": 3.32, "learning_rate": 6.6796875e-05, "loss": 3.4709, "step": 22100 }, { "epoch": 3.34, "learning_rate": 6.664663461538461e-05, "loss": 3.4247, "step": 22200 }, { "epoch": 3.35, "learning_rate": 6.649639423076923e-05, "loss": 3.411, "step": 22300 }, { "epoch": 3.37, "learning_rate": 6.634615384615385e-05, "loss": 3.3555, "step": 22400 }, { "epoch": 3.38, "learning_rate": 6.619591346153846e-05, "loss": 3.3727, "step": 22500 }, { "epoch": 3.4, "learning_rate": 6.604567307692308e-05, "loss": 3.4175, "step": 22600 }, { "epoch": 3.41, "learning_rate": 6.589543269230769e-05, "loss": 3.4083, "step": 22700 }, { "epoch": 3.43, "learning_rate": 6.574519230769231e-05, "loss": 3.4315, "step": 22800 }, { "epoch": 3.44, "learning_rate": 6.559495192307693e-05, "loss": 3.4095, "step": 22900 }, { "epoch": 3.46, "learning_rate": 6.544471153846154e-05, "loss": 3.4292, "step": 23000 }, { "epoch": 3.47, "learning_rate": 6.529447115384616e-05, "loss": 3.4097, "step": 23100 }, { "epoch": 3.49, "learning_rate": 6.514423076923078e-05, "loss": 3.5273, "step": 23200 }, { "epoch": 3.5, "learning_rate": 6.499399038461539e-05, "loss": 3.4549, "step": 23300 }, { "epoch": 3.52, "learning_rate": 6.484375e-05, "loss": 3.3474, "step": 23400 }, { "epoch": 3.53, "learning_rate": 6.469350961538461e-05, "loss": 3.3646, "step": 23500 }, { "epoch": 3.55, "learning_rate": 6.454326923076924e-05, "loss": 3.4457, "step": 23600 }, { "epoch": 3.56, "learning_rate": 6.439302884615385e-05, "loss": 3.4549, "step": 23700 }, { "epoch": 3.58, "learning_rate": 6.424278846153846e-05, "loss": 3.4032, "step": 23800 }, { "epoch": 3.59, "learning_rate": 6.409254807692307e-05, "loss": 3.3047, "step": 23900 }, { "epoch": 3.61, "learning_rate": 6.394230769230769e-05, "loss": 3.4121, "step": 24000 }, { "epoch": 3.62, "learning_rate": 6.379206730769231e-05, "loss": 3.4355, "step": 24100 }, { "epoch": 3.64, "learning_rate": 6.364182692307692e-05, "loss": 3.4153, "step": 24200 }, { "epoch": 3.65, "learning_rate": 6.349158653846155e-05, "loss": 3.3583, "step": 24300 }, { "epoch": 3.67, "learning_rate": 6.334134615384616e-05, "loss": 3.3435, "step": 24400 }, { "epoch": 3.68, "learning_rate": 6.319110576923077e-05, "loss": 3.3166, "step": 24500 }, { "epoch": 3.7, "learning_rate": 6.304086538461539e-05, "loss": 3.3751, "step": 24600 }, { "epoch": 3.71, "learning_rate": 6.2890625e-05, "loss": 3.4042, "step": 24700 }, { "epoch": 3.73, "learning_rate": 6.274038461538461e-05, "loss": 3.3774, "step": 24800 }, { "epoch": 3.74, "learning_rate": 6.259014423076924e-05, "loss": 3.4476, "step": 24900 }, { "epoch": 3.76, "learning_rate": 6.243990384615385e-05, "loss": 3.3689, "step": 25000 }, { "epoch": 3.77, "learning_rate": 6.228966346153846e-05, "loss": 3.4963, "step": 25100 }, { "epoch": 3.79, "learning_rate": 6.213942307692307e-05, "loss": 3.3515, "step": 25200 }, { "epoch": 3.8, "learning_rate": 6.198918269230769e-05, "loss": 3.329, "step": 25300 }, { "epoch": 3.82, "learning_rate": 6.183894230769231e-05, "loss": 3.4401, "step": 25400 }, { "epoch": 3.83, "learning_rate": 6.168870192307694e-05, "loss": 3.3613, "step": 25500 }, { "epoch": 3.85, "learning_rate": 6.153846153846155e-05, "loss": 3.3463, "step": 25600 }, { "epoch": 3.86, "learning_rate": 6.138822115384616e-05, "loss": 3.3742, "step": 25700 }, { "epoch": 3.88, "learning_rate": 6.123798076923077e-05, "loss": 3.3858, "step": 25800 }, { "epoch": 3.89, "learning_rate": 6.108774038461539e-05, "loss": 3.4393, "step": 25900 }, { "epoch": 3.91, "learning_rate": 6.0937500000000004e-05, "loss": 3.4649, "step": 26000 }, { "epoch": 3.92, "learning_rate": 6.0787259615384616e-05, "loss": 3.411, "step": 26100 }, { "epoch": 3.94, "learning_rate": 6.0637019230769235e-05, "loss": 3.4315, "step": 26200 }, { "epoch": 3.95, "learning_rate": 6.048677884615385e-05, "loss": 3.3696, "step": 26300 }, { "epoch": 3.97, "learning_rate": 6.033653846153846e-05, "loss": 3.4611, "step": 26400 }, { "epoch": 3.98, "learning_rate": 6.018629807692308e-05, "loss": 3.4147, "step": 26500 }, { "epoch": 4.0, "learning_rate": 6.003605769230769e-05, "loss": 3.3671, "step": 26600 }, { "epoch": 4.01, "learning_rate": 5.9885817307692316e-05, "loss": 3.2922, "step": 26700 }, { "epoch": 4.03, "learning_rate": 5.973557692307693e-05, "loss": 3.4404, "step": 26800 }, { "epoch": 4.04, "learning_rate": 5.958533653846155e-05, "loss": 3.2936, "step": 26900 }, { "epoch": 4.06, "learning_rate": 5.943509615384616e-05, "loss": 3.3042, "step": 27000 }, { "epoch": 4.07, "learning_rate": 5.928485576923077e-05, "loss": 3.3803, "step": 27100 }, { "epoch": 4.09, "learning_rate": 5.913461538461539e-05, "loss": 3.3067, "step": 27200 }, { "epoch": 4.1, "learning_rate": 5.8984375e-05, "loss": 3.2929, "step": 27300 }, { "epoch": 4.12, "learning_rate": 5.8834134615384614e-05, "loss": 3.3419, "step": 27400 }, { "epoch": 4.13, "learning_rate": 5.868389423076923e-05, "loss": 3.3165, "step": 27500 }, { "epoch": 4.15, "learning_rate": 5.8533653846153845e-05, "loss": 3.3961, "step": 27600 }, { "epoch": 4.16, "learning_rate": 5.8383413461538464e-05, "loss": 3.3305, "step": 27700 }, { "epoch": 4.18, "learning_rate": 5.8233173076923076e-05, "loss": 3.3619, "step": 27800 }, { "epoch": 4.19, "learning_rate": 5.808293269230769e-05, "loss": 3.3138, "step": 27900 }, { "epoch": 4.21, "learning_rate": 5.7932692307692314e-05, "loss": 3.3557, "step": 28000 }, { "epoch": 4.22, "learning_rate": 5.778245192307693e-05, "loss": 3.4086, "step": 28100 }, { "epoch": 4.24, "learning_rate": 5.7632211538461545e-05, "loss": 3.2462, "step": 28200 }, { "epoch": 4.25, "learning_rate": 5.748197115384616e-05, "loss": 3.3991, "step": 28300 }, { "epoch": 4.27, "learning_rate": 5.7331730769230776e-05, "loss": 3.3424, "step": 28400 }, { "epoch": 4.28, "learning_rate": 5.718149038461539e-05, "loss": 3.345, "step": 28500 }, { "epoch": 4.3, "learning_rate": 5.703125e-05, "loss": 3.3057, "step": 28600 }, { "epoch": 4.31, "learning_rate": 5.688100961538462e-05, "loss": 3.339, "step": 28700 }, { "epoch": 4.33, "learning_rate": 5.673076923076923e-05, "loss": 3.3355, "step": 28800 }, { "epoch": 4.34, "learning_rate": 5.6580528846153844e-05, "loss": 3.2835, "step": 28900 }, { "epoch": 4.36, "learning_rate": 5.643028846153846e-05, "loss": 3.1553, "step": 29000 }, { "epoch": 4.37, "learning_rate": 5.6280048076923075e-05, "loss": 3.3456, "step": 29100 }, { "epoch": 4.39, "learning_rate": 5.612980769230769e-05, "loss": 3.2839, "step": 29200 }, { "epoch": 4.4, "learning_rate": 5.597956730769231e-05, "loss": 3.3465, "step": 29300 }, { "epoch": 4.42, "learning_rate": 5.582932692307693e-05, "loss": 3.3, "step": 29400 }, { "epoch": 4.43, "learning_rate": 5.5679086538461544e-05, "loss": 3.395, "step": 29500 }, { "epoch": 4.45, "learning_rate": 5.552884615384616e-05, "loss": 3.3677, "step": 29600 }, { "epoch": 4.46, "learning_rate": 5.5378605769230775e-05, "loss": 3.2638, "step": 29700 }, { "epoch": 4.48, "learning_rate": 5.522836538461539e-05, "loss": 3.2915, "step": 29800 }, { "epoch": 4.49, "learning_rate": 5.5078125000000006e-05, "loss": 3.3927, "step": 29900 }, { "epoch": 4.51, "learning_rate": 5.492788461538462e-05, "loss": 3.2758, "step": 30000 }, { "epoch": 4.52, "learning_rate": 5.477764423076923e-05, "loss": 3.2846, "step": 30100 }, { "epoch": 4.54, "learning_rate": 5.462740384615385e-05, "loss": 3.2502, "step": 30200 }, { "epoch": 4.55, "learning_rate": 5.447716346153846e-05, "loss": 3.345, "step": 30300 }, { "epoch": 4.57, "learning_rate": 5.432692307692307e-05, "loss": 3.3753, "step": 30400 }, { "epoch": 4.58, "learning_rate": 5.417668269230769e-05, "loss": 3.2171, "step": 30500 }, { "epoch": 4.6, "learning_rate": 5.402644230769232e-05, "loss": 3.2815, "step": 30600 }, { "epoch": 4.61, "learning_rate": 5.387620192307693e-05, "loss": 3.3164, "step": 30700 }, { "epoch": 4.63, "learning_rate": 5.372596153846154e-05, "loss": 3.3076, "step": 30800 }, { "epoch": 4.64, "learning_rate": 5.357572115384616e-05, "loss": 3.3597, "step": 30900 }, { "epoch": 4.66, "learning_rate": 5.342548076923077e-05, "loss": 3.2528, "step": 31000 }, { "epoch": 4.67, "learning_rate": 5.3275240384615385e-05, "loss": 3.3317, "step": 31100 }, { "epoch": 4.69, "learning_rate": 5.3125000000000004e-05, "loss": 3.4087, "step": 31200 }, { "epoch": 4.7, "learning_rate": 5.2974759615384616e-05, "loss": 3.3444, "step": 31300 }, { "epoch": 4.72, "learning_rate": 5.2824519230769235e-05, "loss": 3.3282, "step": 31400 }, { "epoch": 4.73, "learning_rate": 5.267427884615385e-05, "loss": 3.2904, "step": 31500 }, { "epoch": 4.75, "learning_rate": 5.252403846153846e-05, "loss": 3.3006, "step": 31600 }, { "epoch": 4.76, "learning_rate": 5.237379807692308e-05, "loss": 3.3654, "step": 31700 }, { "epoch": 4.78, "learning_rate": 5.222355769230769e-05, "loss": 3.2973, "step": 31800 }, { "epoch": 4.79, "learning_rate": 5.2073317307692316e-05, "loss": 3.3668, "step": 31900 }, { "epoch": 4.81, "learning_rate": 5.192307692307693e-05, "loss": 3.3186, "step": 32000 }, { "epoch": 4.82, "learning_rate": 5.177283653846155e-05, "loss": 3.1989, "step": 32100 }, { "epoch": 4.84, "learning_rate": 5.162259615384616e-05, "loss": 3.3046, "step": 32200 }, { "epoch": 4.85, "learning_rate": 5.147235576923077e-05, "loss": 3.2191, "step": 32300 }, { "epoch": 4.87, "learning_rate": 5.132211538461539e-05, "loss": 3.2571, "step": 32400 }, { "epoch": 4.88, "learning_rate": 5.1171875e-05, "loss": 3.2072, "step": 32500 }, { "epoch": 4.9, "learning_rate": 5.1021634615384614e-05, "loss": 3.322, "step": 32600 }, { "epoch": 4.91, "learning_rate": 5.087139423076923e-05, "loss": 3.299, "step": 32700 }, { "epoch": 4.93, "learning_rate": 5.0721153846153845e-05, "loss": 3.2845, "step": 32800 }, { "epoch": 4.94, "learning_rate": 5.057091346153846e-05, "loss": 3.2909, "step": 32900 }, { "epoch": 4.96, "learning_rate": 5.0420673076923076e-05, "loss": 3.3085, "step": 33000 }, { "epoch": 4.97, "learning_rate": 5.027043269230769e-05, "loss": 3.3532, "step": 33100 }, { "epoch": 4.99, "learning_rate": 5.0120192307692314e-05, "loss": 3.2598, "step": 33200 }, { "epoch": 5.0, "learning_rate": 4.9969951923076926e-05, "loss": 3.259, "step": 33300 }, { "epoch": 5.02, "learning_rate": 4.981971153846154e-05, "loss": 3.194, "step": 33400 }, { "epoch": 5.03, "learning_rate": 4.966947115384616e-05, "loss": 3.202, "step": 33500 }, { "epoch": 5.05, "learning_rate": 4.9519230769230776e-05, "loss": 3.1681, "step": 33600 }, { "epoch": 5.06, "learning_rate": 4.936899038461539e-05, "loss": 3.256, "step": 33700 }, { "epoch": 5.08, "learning_rate": 4.921875e-05, "loss": 3.2368, "step": 33800 }, { "epoch": 5.09, "learning_rate": 4.906850961538462e-05, "loss": 3.3036, "step": 33900 }, { "epoch": 5.11, "learning_rate": 4.891826923076923e-05, "loss": 3.2212, "step": 34000 }, { "epoch": 5.12, "learning_rate": 4.8768028846153843e-05, "loss": 3.3221, "step": 34100 }, { "epoch": 5.14, "learning_rate": 4.861778846153847e-05, "loss": 3.2859, "step": 34200 }, { "epoch": 5.15, "learning_rate": 4.846754807692308e-05, "loss": 3.1449, "step": 34300 }, { "epoch": 5.17, "learning_rate": 4.8317307692307693e-05, "loss": 3.2694, "step": 34400 }, { "epoch": 5.18, "learning_rate": 4.816706730769231e-05, "loss": 3.2395, "step": 34500 }, { "epoch": 5.2, "learning_rate": 4.8016826923076924e-05, "loss": 3.2135, "step": 34600 }, { "epoch": 5.21, "learning_rate": 4.7866586538461537e-05, "loss": 3.2219, "step": 34700 }, { "epoch": 5.23, "learning_rate": 4.7716346153846155e-05, "loss": 3.2907, "step": 34800 }, { "epoch": 5.24, "learning_rate": 4.7566105769230774e-05, "loss": 3.3046, "step": 34900 }, { "epoch": 5.26, "learning_rate": 4.7415865384615386e-05, "loss": 3.2011, "step": 35000 }, { "epoch": 5.27, "learning_rate": 4.7265625000000005e-05, "loss": 3.2394, "step": 35100 }, { "epoch": 5.29, "learning_rate": 4.711538461538462e-05, "loss": 3.2004, "step": 35200 }, { "epoch": 5.3, "learning_rate": 4.696514423076923e-05, "loss": 3.2539, "step": 35300 }, { "epoch": 5.32, "learning_rate": 4.681490384615385e-05, "loss": 3.3304, "step": 35400 }, { "epoch": 5.33, "learning_rate": 4.666466346153847e-05, "loss": 3.2046, "step": 35500 }, { "epoch": 5.35, "learning_rate": 4.651442307692308e-05, "loss": 3.3181, "step": 35600 }, { "epoch": 5.36, "learning_rate": 4.636418269230769e-05, "loss": 3.1461, "step": 35700 }, { "epoch": 5.38, "learning_rate": 4.621394230769231e-05, "loss": 3.2141, "step": 35800 }, { "epoch": 5.39, "learning_rate": 4.606370192307692e-05, "loss": 3.2212, "step": 35900 }, { "epoch": 5.41, "learning_rate": 4.591346153846154e-05, "loss": 3.2862, "step": 36000 }, { "epoch": 5.42, "learning_rate": 4.576322115384616e-05, "loss": 3.1912, "step": 36100 }, { "epoch": 5.44, "learning_rate": 4.561298076923077e-05, "loss": 3.3004, "step": 36200 }, { "epoch": 5.45, "learning_rate": 4.5462740384615385e-05, "loss": 3.1604, "step": 36300 }, { "epoch": 5.47, "learning_rate": 4.5312500000000004e-05, "loss": 3.1868, "step": 36400 }, { "epoch": 5.48, "learning_rate": 4.5162259615384616e-05, "loss": 3.2882, "step": 36500 }, { "epoch": 5.5, "learning_rate": 4.501201923076923e-05, "loss": 3.2211, "step": 36600 }, { "epoch": 5.51, "learning_rate": 4.486177884615385e-05, "loss": 3.2298, "step": 36700 }, { "epoch": 5.53, "learning_rate": 4.4711538461538466e-05, "loss": 3.2108, "step": 36800 }, { "epoch": 5.54, "learning_rate": 4.456129807692308e-05, "loss": 3.2887, "step": 36900 }, { "epoch": 5.56, "learning_rate": 4.44110576923077e-05, "loss": 3.1887, "step": 37000 }, { "epoch": 5.57, "learning_rate": 4.426081730769231e-05, "loss": 3.3018, "step": 37100 }, { "epoch": 5.59, "learning_rate": 4.411057692307692e-05, "loss": 3.2316, "step": 37200 }, { "epoch": 5.6, "learning_rate": 4.396033653846154e-05, "loss": 3.3247, "step": 37300 }, { "epoch": 5.62, "learning_rate": 4.381009615384616e-05, "loss": 3.2524, "step": 37400 }, { "epoch": 5.63, "learning_rate": 4.365985576923077e-05, "loss": 3.4009, "step": 37500 }, { "epoch": 5.65, "learning_rate": 4.350961538461539e-05, "loss": 3.1606, "step": 37600 }, { "epoch": 5.66, "learning_rate": 4.3359375e-05, "loss": 3.3663, "step": 37700 }, { "epoch": 5.68, "learning_rate": 4.3209134615384614e-05, "loss": 3.2013, "step": 37800 }, { "epoch": 5.69, "learning_rate": 4.305889423076923e-05, "loss": 3.2427, "step": 37900 }, { "epoch": 5.71, "learning_rate": 4.2908653846153845e-05, "loss": 3.2398, "step": 38000 }, { "epoch": 5.72, "learning_rate": 4.2758413461538464e-05, "loss": 3.293, "step": 38100 }, { "epoch": 5.74, "learning_rate": 4.260817307692308e-05, "loss": 3.2233, "step": 38200 }, { "epoch": 5.75, "learning_rate": 4.2457932692307695e-05, "loss": 3.2299, "step": 38300 }, { "epoch": 5.77, "learning_rate": 4.230769230769231e-05, "loss": 3.1866, "step": 38400 }, { "epoch": 5.78, "learning_rate": 4.2157451923076926e-05, "loss": 3.2959, "step": 38500 }, { "epoch": 5.8, "learning_rate": 4.200721153846154e-05, "loss": 3.2326, "step": 38600 }, { "epoch": 5.81, "learning_rate": 4.185697115384616e-05, "loss": 3.2325, "step": 38700 }, { "epoch": 5.83, "learning_rate": 4.1706730769230776e-05, "loss": 3.312, "step": 38800 }, { "epoch": 5.84, "learning_rate": 4.155649038461539e-05, "loss": 3.2872, "step": 38900 }, { "epoch": 5.86, "learning_rate": 4.140625e-05, "loss": 3.3211, "step": 39000 }, { "epoch": 5.87, "learning_rate": 4.125600961538462e-05, "loss": 3.267, "step": 39100 }, { "epoch": 5.89, "learning_rate": 4.110576923076923e-05, "loss": 3.2381, "step": 39200 }, { "epoch": 5.9, "learning_rate": 4.095552884615384e-05, "loss": 3.1223, "step": 39300 }, { "epoch": 5.92, "learning_rate": 4.080528846153847e-05, "loss": 3.1428, "step": 39400 }, { "epoch": 5.93, "learning_rate": 4.065504807692308e-05, "loss": 3.3539, "step": 39500 }, { "epoch": 5.95, "learning_rate": 4.050480769230769e-05, "loss": 3.1743, "step": 39600 }, { "epoch": 5.96, "learning_rate": 4.035456730769231e-05, "loss": 3.1944, "step": 39700 }, { "epoch": 5.98, "learning_rate": 4.0204326923076924e-05, "loss": 3.2237, "step": 39800 }, { "epoch": 5.99, "learning_rate": 4.0054086538461536e-05, "loss": 3.3603, "step": 39900 }, { "epoch": 6.01, "learning_rate": 3.9903846153846155e-05, "loss": 3.1751, "step": 40000 }, { "epoch": 6.02, "learning_rate": 3.9753605769230774e-05, "loss": 3.1382, "step": 40100 }, { "epoch": 6.04, "learning_rate": 3.9603365384615386e-05, "loss": 3.2931, "step": 40200 }, { "epoch": 6.05, "learning_rate": 3.9453125000000005e-05, "loss": 3.1821, "step": 40300 }, { "epoch": 6.07, "learning_rate": 3.930288461538462e-05, "loss": 3.159, "step": 40400 }, { "epoch": 6.08, "learning_rate": 3.915264423076923e-05, "loss": 3.1653, "step": 40500 }, { "epoch": 6.1, "learning_rate": 3.900240384615385e-05, "loss": 3.2043, "step": 40600 }, { "epoch": 6.11, "learning_rate": 3.885216346153847e-05, "loss": 3.1382, "step": 40700 }, { "epoch": 6.13, "learning_rate": 3.870192307692308e-05, "loss": 3.1522, "step": 40800 }, { "epoch": 6.14, "learning_rate": 3.855168269230769e-05, "loss": 3.1957, "step": 40900 }, { "epoch": 6.16, "learning_rate": 3.840144230769231e-05, "loss": 3.1041, "step": 41000 }, { "epoch": 6.17, "learning_rate": 3.825120192307692e-05, "loss": 3.1838, "step": 41100 }, { "epoch": 6.19, "learning_rate": 3.810096153846154e-05, "loss": 3.1468, "step": 41200 }, { "epoch": 6.2, "learning_rate": 3.795072115384616e-05, "loss": 3.1075, "step": 41300 }, { "epoch": 6.22, "learning_rate": 3.780048076923077e-05, "loss": 3.1594, "step": 41400 }, { "epoch": 6.23, "learning_rate": 3.7650240384615385e-05, "loss": 3.1844, "step": 41500 }, { "epoch": 6.25, "learning_rate": 3.7500000000000003e-05, "loss": 3.3313, "step": 41600 }, { "epoch": 6.27, "learning_rate": 3.7349759615384616e-05, "loss": 3.1786, "step": 41700 }, { "epoch": 6.28, "learning_rate": 3.719951923076923e-05, "loss": 3.2375, "step": 41800 }, { "epoch": 6.3, "learning_rate": 3.704927884615385e-05, "loss": 3.1206, "step": 41900 }, { "epoch": 6.31, "learning_rate": 3.6899038461538466e-05, "loss": 3.1919, "step": 42000 }, { "epoch": 6.33, "learning_rate": 3.674879807692308e-05, "loss": 3.2956, "step": 42100 }, { "epoch": 6.34, "learning_rate": 3.6598557692307697e-05, "loss": 3.2287, "step": 42200 }, { "epoch": 6.36, "learning_rate": 3.644831730769231e-05, "loss": 3.1453, "step": 42300 }, { "epoch": 6.37, "learning_rate": 3.629807692307692e-05, "loss": 3.1535, "step": 42400 }, { "epoch": 6.39, "learning_rate": 3.614783653846154e-05, "loss": 3.165, "step": 42500 }, { "epoch": 6.4, "learning_rate": 3.599759615384616e-05, "loss": 3.1838, "step": 42600 }, { "epoch": 6.42, "learning_rate": 3.584735576923077e-05, "loss": 3.1737, "step": 42700 }, { "epoch": 6.43, "learning_rate": 3.569711538461539e-05, "loss": 3.1649, "step": 42800 }, { "epoch": 6.45, "learning_rate": 3.5546875e-05, "loss": 3.103, "step": 42900 }, { "epoch": 6.46, "learning_rate": 3.5396634615384614e-05, "loss": 3.1786, "step": 43000 }, { "epoch": 6.48, "learning_rate": 3.524639423076923e-05, "loss": 3.2883, "step": 43100 }, { "epoch": 6.49, "learning_rate": 3.5096153846153845e-05, "loss": 3.2017, "step": 43200 }, { "epoch": 6.51, "learning_rate": 3.4945913461538464e-05, "loss": 3.2765, "step": 43300 }, { "epoch": 6.52, "learning_rate": 3.479567307692308e-05, "loss": 3.1713, "step": 43400 }, { "epoch": 6.54, "learning_rate": 3.4645432692307695e-05, "loss": 3.1993, "step": 43500 }, { "epoch": 6.55, "learning_rate": 3.449519230769231e-05, "loss": 3.2256, "step": 43600 }, { "epoch": 6.57, "learning_rate": 3.4344951923076926e-05, "loss": 3.2242, "step": 43700 }, { "epoch": 6.58, "learning_rate": 3.419471153846154e-05, "loss": 3.256, "step": 43800 }, { "epoch": 6.6, "learning_rate": 3.404447115384616e-05, "loss": 3.1709, "step": 43900 }, { "epoch": 6.61, "learning_rate": 3.3894230769230776e-05, "loss": 3.1652, "step": 44000 }, { "epoch": 6.63, "learning_rate": 3.374399038461539e-05, "loss": 3.0949, "step": 44100 }, { "epoch": 6.64, "learning_rate": 3.359375e-05, "loss": 3.2357, "step": 44200 }, { "epoch": 6.66, "learning_rate": 3.344350961538462e-05, "loss": 3.1377, "step": 44300 }, { "epoch": 6.67, "learning_rate": 3.329326923076923e-05, "loss": 3.1905, "step": 44400 }, { "epoch": 6.69, "learning_rate": 3.314302884615384e-05, "loss": 3.1792, "step": 44500 }, { "epoch": 6.7, "learning_rate": 3.299278846153846e-05, "loss": 3.1817, "step": 44600 }, { "epoch": 6.72, "learning_rate": 3.284254807692308e-05, "loss": 3.1305, "step": 44700 }, { "epoch": 6.73, "learning_rate": 3.269230769230769e-05, "loss": 3.1406, "step": 44800 }, { "epoch": 6.75, "learning_rate": 3.254206730769231e-05, "loss": 3.2035, "step": 44900 }, { "epoch": 6.76, "learning_rate": 3.2391826923076924e-05, "loss": 3.115, "step": 45000 }, { "epoch": 6.78, "learning_rate": 3.2241586538461536e-05, "loss": 3.1376, "step": 45100 }, { "epoch": 6.79, "learning_rate": 3.2091346153846155e-05, "loss": 3.1234, "step": 45200 }, { "epoch": 6.81, "learning_rate": 3.1941105769230774e-05, "loss": 3.1036, "step": 45300 }, { "epoch": 6.82, "learning_rate": 3.1790865384615386e-05, "loss": 3.2273, "step": 45400 }, { "epoch": 6.84, "learning_rate": 3.1640625e-05, "loss": 3.1999, "step": 45500 }, { "epoch": 6.85, "learning_rate": 3.149038461538462e-05, "loss": 3.1843, "step": 45600 }, { "epoch": 6.87, "learning_rate": 3.134014423076923e-05, "loss": 3.1522, "step": 45700 }, { "epoch": 6.88, "learning_rate": 3.118990384615385e-05, "loss": 3.1722, "step": 45800 }, { "epoch": 6.9, "learning_rate": 3.103966346153847e-05, "loss": 3.2638, "step": 45900 }, { "epoch": 6.91, "learning_rate": 3.088942307692308e-05, "loss": 3.1483, "step": 46000 }, { "epoch": 6.93, "learning_rate": 3.073918269230769e-05, "loss": 3.2081, "step": 46100 }, { "epoch": 6.94, "learning_rate": 3.058894230769231e-05, "loss": 3.2939, "step": 46200 }, { "epoch": 6.96, "learning_rate": 3.0438701923076922e-05, "loss": 3.0819, "step": 46300 }, { "epoch": 6.97, "learning_rate": 3.0288461538461538e-05, "loss": 3.1972, "step": 46400 }, { "epoch": 6.99, "learning_rate": 3.0138221153846157e-05, "loss": 3.1313, "step": 46500 }, { "epoch": 7.0, "learning_rate": 2.9987980769230772e-05, "loss": 3.2346, "step": 46600 }, { "epoch": 7.02, "learning_rate": 2.9837740384615388e-05, "loss": 3.0856, "step": 46700 }, { "epoch": 7.03, "learning_rate": 2.96875e-05, "loss": 3.1688, "step": 46800 }, { "epoch": 7.05, "learning_rate": 2.9537259615384615e-05, "loss": 3.0855, "step": 46900 }, { "epoch": 7.06, "learning_rate": 2.938701923076923e-05, "loss": 3.107, "step": 47000 }, { "epoch": 7.08, "learning_rate": 2.9236778846153846e-05, "loss": 3.1544, "step": 47100 }, { "epoch": 7.09, "learning_rate": 2.9086538461538465e-05, "loss": 3.1732, "step": 47200 }, { "epoch": 7.11, "learning_rate": 2.893629807692308e-05, "loss": 3.1477, "step": 47300 }, { "epoch": 7.12, "learning_rate": 2.8786057692307693e-05, "loss": 3.137, "step": 47400 }, { "epoch": 7.14, "learning_rate": 2.863581730769231e-05, "loss": 3.1548, "step": 47500 }, { "epoch": 7.15, "learning_rate": 2.8485576923076924e-05, "loss": 3.2163, "step": 47600 }, { "epoch": 7.17, "learning_rate": 2.8335336538461536e-05, "loss": 3.1692, "step": 47700 }, { "epoch": 7.18, "learning_rate": 2.818509615384616e-05, "loss": 3.1894, "step": 47800 }, { "epoch": 7.2, "learning_rate": 2.8034855769230774e-05, "loss": 3.1581, "step": 47900 }, { "epoch": 7.21, "learning_rate": 2.7884615384615386e-05, "loss": 3.1314, "step": 48000 }, { "epoch": 7.23, "learning_rate": 2.7734375e-05, "loss": 3.1914, "step": 48100 }, { "epoch": 7.24, "learning_rate": 2.7584134615384617e-05, "loss": 3.193, "step": 48200 }, { "epoch": 7.26, "learning_rate": 2.743389423076923e-05, "loss": 3.1188, "step": 48300 }, { "epoch": 7.27, "learning_rate": 2.7283653846153845e-05, "loss": 3.1413, "step": 48400 }, { "epoch": 7.29, "learning_rate": 2.7133413461538464e-05, "loss": 3.1699, "step": 48500 }, { "epoch": 7.3, "learning_rate": 2.698317307692308e-05, "loss": 3.1479, "step": 48600 }, { "epoch": 7.32, "learning_rate": 2.6832932692307695e-05, "loss": 3.1354, "step": 48700 }, { "epoch": 7.33, "learning_rate": 2.668269230769231e-05, "loss": 3.1386, "step": 48800 }, { "epoch": 7.35, "learning_rate": 2.6532451923076922e-05, "loss": 3.1308, "step": 48900 }, { "epoch": 7.36, "learning_rate": 2.6382211538461538e-05, "loss": 3.1955, "step": 49000 }, { "epoch": 7.38, "learning_rate": 2.6231971153846157e-05, "loss": 3.0473, "step": 49100 }, { "epoch": 7.39, "learning_rate": 2.6081730769230772e-05, "loss": 3.1117, "step": 49200 }, { "epoch": 7.41, "learning_rate": 2.5931490384615388e-05, "loss": 3.127, "step": 49300 }, { "epoch": 7.42, "learning_rate": 2.578125e-05, "loss": 3.1794, "step": 49400 }, { "epoch": 7.44, "learning_rate": 2.5631009615384615e-05, "loss": 3.1297, "step": 49500 }, { "epoch": 7.45, "learning_rate": 2.548076923076923e-05, "loss": 3.204, "step": 49600 }, { "epoch": 7.47, "learning_rate": 2.5330528846153846e-05, "loss": 3.1469, "step": 49700 }, { "epoch": 7.48, "learning_rate": 2.5180288461538465e-05, "loss": 3.1524, "step": 49800 }, { "epoch": 7.5, "learning_rate": 2.503004807692308e-05, "loss": 3.1529, "step": 49900 }, { "epoch": 7.51, "learning_rate": 2.4879807692307693e-05, "loss": 3.209, "step": 50000 }, { "epoch": 7.53, "learning_rate": 2.472956730769231e-05, "loss": 3.1885, "step": 50100 }, { "epoch": 7.54, "learning_rate": 2.4579326923076924e-05, "loss": 3.1388, "step": 50200 }, { "epoch": 7.56, "learning_rate": 2.442908653846154e-05, "loss": 3.163, "step": 50300 }, { "epoch": 7.57, "learning_rate": 2.4278846153846155e-05, "loss": 3.1048, "step": 50400 }, { "epoch": 7.59, "learning_rate": 2.412860576923077e-05, "loss": 3.1563, "step": 50500 }, { "epoch": 7.6, "learning_rate": 2.3978365384615386e-05, "loss": 3.0561, "step": 50600 }, { "epoch": 7.62, "learning_rate": 2.3828125e-05, "loss": 3.1133, "step": 50700 }, { "epoch": 7.63, "learning_rate": 2.3677884615384617e-05, "loss": 3.0987, "step": 50800 }, { "epoch": 7.65, "learning_rate": 2.3527644230769233e-05, "loss": 3.1504, "step": 50900 }, { "epoch": 7.66, "learning_rate": 2.3377403846153848e-05, "loss": 3.057, "step": 51000 }, { "epoch": 7.68, "learning_rate": 2.3227163461538464e-05, "loss": 3.1533, "step": 51100 }, { "epoch": 7.69, "learning_rate": 2.307692307692308e-05, "loss": 3.1509, "step": 51200 }, { "epoch": 7.71, "learning_rate": 2.2926682692307695e-05, "loss": 3.1098, "step": 51300 }, { "epoch": 7.72, "learning_rate": 2.2776442307692307e-05, "loss": 3.1403, "step": 51400 }, { "epoch": 7.74, "learning_rate": 2.2626201923076922e-05, "loss": 3.175, "step": 51500 }, { "epoch": 7.75, "learning_rate": 2.247596153846154e-05, "loss": 3.082, "step": 51600 }, { "epoch": 7.77, "learning_rate": 2.2325721153846153e-05, "loss": 3.1393, "step": 51700 }, { "epoch": 7.78, "learning_rate": 2.217548076923077e-05, "loss": 3.0962, "step": 51800 }, { "epoch": 7.8, "learning_rate": 2.2025240384615388e-05, "loss": 3.1229, "step": 51900 }, { "epoch": 7.81, "learning_rate": 2.1875e-05, "loss": 3.1898, "step": 52000 }, { "epoch": 7.83, "learning_rate": 2.1724759615384615e-05, "loss": 3.1181, "step": 52100 }, { "epoch": 7.84, "learning_rate": 2.1574519230769234e-05, "loss": 3.1294, "step": 52200 }, { "epoch": 7.86, "learning_rate": 2.1424278846153846e-05, "loss": 3.1118, "step": 52300 }, { "epoch": 7.87, "learning_rate": 2.1274038461538462e-05, "loss": 3.2093, "step": 52400 }, { "epoch": 7.89, "learning_rate": 2.112379807692308e-05, "loss": 3.0968, "step": 52500 }, { "epoch": 7.9, "learning_rate": 2.0973557692307693e-05, "loss": 3.1043, "step": 52600 }, { "epoch": 7.92, "learning_rate": 2.082331730769231e-05, "loss": 3.1901, "step": 52700 }, { "epoch": 7.93, "learning_rate": 2.0673076923076924e-05, "loss": 3.1718, "step": 52800 }, { "epoch": 7.95, "learning_rate": 2.052283653846154e-05, "loss": 3.1754, "step": 52900 }, { "epoch": 7.96, "learning_rate": 2.0372596153846155e-05, "loss": 3.1013, "step": 53000 }, { "epoch": 7.98, "learning_rate": 2.022235576923077e-05, "loss": 3.1273, "step": 53100 }, { "epoch": 7.99, "learning_rate": 2.0072115384615386e-05, "loss": 3.0944, "step": 53200 }, { "epoch": 8.01, "learning_rate": 1.9921875e-05, "loss": 3.1341, "step": 53300 }, { "epoch": 8.02, "learning_rate": 1.9771634615384617e-05, "loss": 3.1024, "step": 53400 }, { "epoch": 8.04, "learning_rate": 1.9621394230769232e-05, "loss": 3.2147, "step": 53500 }, { "epoch": 8.05, "learning_rate": 1.9471153846153848e-05, "loss": 3.1102, "step": 53600 }, { "epoch": 8.07, "learning_rate": 1.9320913461538463e-05, "loss": 3.2247, "step": 53700 }, { "epoch": 8.08, "learning_rate": 1.917067307692308e-05, "loss": 3.1304, "step": 53800 }, { "epoch": 8.1, "learning_rate": 1.9020432692307695e-05, "loss": 3.1136, "step": 53900 }, { "epoch": 8.11, "learning_rate": 1.8870192307692307e-05, "loss": 3.1116, "step": 54000 }, { "epoch": 8.13, "learning_rate": 1.8719951923076922e-05, "loss": 3.0719, "step": 54100 }, { "epoch": 8.14, "learning_rate": 1.856971153846154e-05, "loss": 2.9952, "step": 54200 }, { "epoch": 8.16, "learning_rate": 1.8419471153846153e-05, "loss": 3.1379, "step": 54300 }, { "epoch": 8.17, "learning_rate": 1.826923076923077e-05, "loss": 3.0153, "step": 54400 }, { "epoch": 8.19, "learning_rate": 1.8118990384615388e-05, "loss": 3.0276, "step": 54500 }, { "epoch": 8.2, "learning_rate": 1.796875e-05, "loss": 3.1095, "step": 54600 }, { "epoch": 8.22, "learning_rate": 1.7818509615384615e-05, "loss": 3.173, "step": 54700 }, { "epoch": 8.23, "learning_rate": 1.7668269230769234e-05, "loss": 3.0593, "step": 54800 }, { "epoch": 8.25, "learning_rate": 1.7518028846153846e-05, "loss": 3.2008, "step": 54900 }, { "epoch": 8.26, "learning_rate": 1.7367788461538462e-05, "loss": 3.0976, "step": 55000 }, { "epoch": 8.28, "learning_rate": 1.721754807692308e-05, "loss": 3.0239, "step": 55100 }, { "epoch": 8.29, "learning_rate": 1.7067307692307693e-05, "loss": 2.9954, "step": 55200 }, { "epoch": 8.31, "learning_rate": 1.6917067307692308e-05, "loss": 3.0959, "step": 55300 }, { "epoch": 8.32, "learning_rate": 1.6766826923076924e-05, "loss": 3.1744, "step": 55400 }, { "epoch": 8.34, "learning_rate": 1.661658653846154e-05, "loss": 3.1568, "step": 55500 }, { "epoch": 8.35, "learning_rate": 1.6466346153846155e-05, "loss": 3.2058, "step": 55600 }, { "epoch": 8.37, "learning_rate": 1.631610576923077e-05, "loss": 3.2486, "step": 55700 }, { "epoch": 8.38, "learning_rate": 1.6165865384615386e-05, "loss": 3.0785, "step": 55800 }, { "epoch": 8.4, "learning_rate": 1.6015625e-05, "loss": 3.1703, "step": 55900 }, { "epoch": 8.41, "learning_rate": 1.5865384615384617e-05, "loss": 3.0091, "step": 56000 }, { "epoch": 8.43, "learning_rate": 1.5715144230769232e-05, "loss": 2.9814, "step": 56100 }, { "epoch": 8.44, "learning_rate": 1.5564903846153848e-05, "loss": 3.1474, "step": 56200 }, { "epoch": 8.46, "learning_rate": 1.541466346153846e-05, "loss": 3.0801, "step": 56300 }, { "epoch": 8.47, "learning_rate": 1.526442307692308e-05, "loss": 3.0667, "step": 56400 }, { "epoch": 8.49, "learning_rate": 1.5114182692307693e-05, "loss": 3.1068, "step": 56500 }, { "epoch": 8.5, "learning_rate": 1.4963942307692308e-05, "loss": 3.1035, "step": 56600 }, { "epoch": 8.52, "learning_rate": 1.4813701923076922e-05, "loss": 3.1616, "step": 56700 }, { "epoch": 8.53, "learning_rate": 1.466346153846154e-05, "loss": 3.2153, "step": 56800 }, { "epoch": 8.55, "learning_rate": 1.4513221153846155e-05, "loss": 3.1312, "step": 56900 }, { "epoch": 8.56, "learning_rate": 1.4362980769230769e-05, "loss": 3.1062, "step": 57000 }, { "epoch": 8.58, "learning_rate": 1.4212740384615386e-05, "loss": 3.092, "step": 57100 }, { "epoch": 8.59, "learning_rate": 1.4062500000000001e-05, "loss": 3.1349, "step": 57200 }, { "epoch": 8.61, "learning_rate": 1.3912259615384615e-05, "loss": 3.1018, "step": 57300 }, { "epoch": 8.62, "learning_rate": 1.3762019230769232e-05, "loss": 3.078, "step": 57400 }, { "epoch": 8.64, "learning_rate": 1.3611778846153848e-05, "loss": 3.0693, "step": 57500 }, { "epoch": 8.65, "learning_rate": 1.3461538461538462e-05, "loss": 3.0487, "step": 57600 }, { "epoch": 8.67, "learning_rate": 1.3311298076923079e-05, "loss": 3.0893, "step": 57700 }, { "epoch": 8.68, "learning_rate": 1.3161057692307693e-05, "loss": 3.0998, "step": 57800 }, { "epoch": 8.7, "learning_rate": 1.3010817307692308e-05, "loss": 3.0287, "step": 57900 }, { "epoch": 8.71, "learning_rate": 1.2860576923076922e-05, "loss": 3.0558, "step": 58000 }, { "epoch": 8.73, "learning_rate": 1.271033653846154e-05, "loss": 3.1865, "step": 58100 }, { "epoch": 8.74, "learning_rate": 1.2560096153846155e-05, "loss": 3.1941, "step": 58200 }, { "epoch": 8.76, "learning_rate": 1.240985576923077e-05, "loss": 3.1259, "step": 58300 }, { "epoch": 8.77, "learning_rate": 1.2259615384615384e-05, "loss": 3.0836, "step": 58400 }, { "epoch": 8.79, "learning_rate": 1.2109375000000001e-05, "loss": 3.2193, "step": 58500 }, { "epoch": 8.8, "learning_rate": 1.1959134615384617e-05, "loss": 3.091, "step": 58600 }, { "epoch": 8.82, "learning_rate": 1.180889423076923e-05, "loss": 3.1139, "step": 58700 }, { "epoch": 8.83, "learning_rate": 1.1658653846153846e-05, "loss": 3.1585, "step": 58800 }, { "epoch": 8.85, "learning_rate": 1.1508413461538462e-05, "loss": 3.1098, "step": 58900 }, { "epoch": 8.86, "learning_rate": 1.1358173076923077e-05, "loss": 3.0185, "step": 59000 }, { "epoch": 8.88, "learning_rate": 1.1207932692307693e-05, "loss": 3.1064, "step": 59100 }, { "epoch": 8.89, "learning_rate": 1.1057692307692308e-05, "loss": 3.1722, "step": 59200 }, { "epoch": 8.91, "learning_rate": 1.0907451923076924e-05, "loss": 3.107, "step": 59300 }, { "epoch": 8.92, "learning_rate": 1.075721153846154e-05, "loss": 3.0859, "step": 59400 }, { "epoch": 8.94, "learning_rate": 1.0606971153846155e-05, "loss": 3.1676, "step": 59500 }, { "epoch": 8.95, "learning_rate": 1.045673076923077e-05, "loss": 3.1134, "step": 59600 }, { "epoch": 8.97, "learning_rate": 1.0306490384615384e-05, "loss": 3.1622, "step": 59700 }, { "epoch": 8.98, "learning_rate": 1.0156250000000001e-05, "loss": 3.1357, "step": 59800 }, { "epoch": 9.0, "learning_rate": 1.0006009615384617e-05, "loss": 3.1807, "step": 59900 }, { "epoch": 9.01, "learning_rate": 9.85576923076923e-06, "loss": 3.1431, "step": 60000 }, { "epoch": 9.03, "learning_rate": 9.705528846153846e-06, "loss": 3.2043, "step": 60100 }, { "epoch": 9.04, "learning_rate": 9.555288461538462e-06, "loss": 3.0665, "step": 60200 }, { "epoch": 9.06, "learning_rate": 9.405048076923077e-06, "loss": 3.016, "step": 60300 }, { "epoch": 9.07, "learning_rate": 9.254807692307693e-06, "loss": 3.0571, "step": 60400 }, { "epoch": 9.09, "learning_rate": 9.104567307692308e-06, "loss": 3.1035, "step": 60500 }, { "epoch": 9.1, "learning_rate": 8.954326923076924e-06, "loss": 3.1543, "step": 60600 }, { "epoch": 9.12, "learning_rate": 8.804086538461539e-06, "loss": 3.1839, "step": 60700 }, { "epoch": 9.13, "learning_rate": 8.653846153846155e-06, "loss": 3.1786, "step": 60800 }, { "epoch": 9.15, "learning_rate": 8.50360576923077e-06, "loss": 3.0707, "step": 60900 }, { "epoch": 9.16, "learning_rate": 8.353365384615384e-06, "loss": 3.052, "step": 61000 }, { "epoch": 9.18, "learning_rate": 8.203125000000001e-06, "loss": 3.0002, "step": 61100 }, { "epoch": 9.19, "learning_rate": 8.052884615384617e-06, "loss": 3.0621, "step": 61200 }, { "epoch": 9.21, "learning_rate": 7.90264423076923e-06, "loss": 3.1472, "step": 61300 }, { "epoch": 9.22, "learning_rate": 7.752403846153846e-06, "loss": 3.2139, "step": 61400 }, { "epoch": 9.24, "learning_rate": 7.6021634615384615e-06, "loss": 3.1535, "step": 61500 }, { "epoch": 9.25, "learning_rate": 7.451923076923077e-06, "loss": 3.0481, "step": 61600 }, { "epoch": 9.27, "learning_rate": 7.301682692307693e-06, "loss": 3.1044, "step": 61700 }, { "epoch": 9.28, "learning_rate": 7.151442307692307e-06, "loss": 3.1502, "step": 61800 }, { "epoch": 9.3, "learning_rate": 7.001201923076924e-06, "loss": 3.16, "step": 61900 }, { "epoch": 9.31, "learning_rate": 6.850961538461539e-06, "loss": 3.1364, "step": 62000 }, { "epoch": 9.33, "learning_rate": 6.700721153846154e-06, "loss": 3.1023, "step": 62100 }, { "epoch": 9.34, "learning_rate": 6.55048076923077e-06, "loss": 3.0382, "step": 62200 }, { "epoch": 9.36, "learning_rate": 6.400240384615385e-06, "loss": 3.087, "step": 62300 }, { "epoch": 9.38, "learning_rate": 6.25e-06, "loss": 3.1023, "step": 62400 }, { "epoch": 9.39, "learning_rate": 6.099759615384616e-06, "loss": 3.1571, "step": 62500 }, { "epoch": 9.41, "learning_rate": 5.9495192307692305e-06, "loss": 3.0847, "step": 62600 }, { "epoch": 9.42, "learning_rate": 5.799278846153847e-06, "loss": 3.0513, "step": 62700 }, { "epoch": 9.44, "learning_rate": 5.6490384615384615e-06, "loss": 3.1568, "step": 62800 }, { "epoch": 9.45, "learning_rate": 5.498798076923077e-06, "loss": 3.1487, "step": 62900 }, { "epoch": 9.47, "learning_rate": 5.3485576923076925e-06, "loss": 3.1173, "step": 63000 }, { "epoch": 9.48, "learning_rate": 5.198317307692308e-06, "loss": 3.0394, "step": 63100 }, { "epoch": 9.5, "learning_rate": 5.0480769230769235e-06, "loss": 3.0607, "step": 63200 }, { "epoch": 9.51, "learning_rate": 4.897836538461539e-06, "loss": 3.0811, "step": 63300 }, { "epoch": 9.53, "learning_rate": 4.747596153846154e-06, "loss": 3.1086, "step": 63400 }, { "epoch": 9.54, "learning_rate": 4.59735576923077e-06, "loss": 3.0946, "step": 63500 }, { "epoch": 9.56, "learning_rate": 4.447115384615385e-06, "loss": 3.0247, "step": 63600 }, { "epoch": 9.57, "learning_rate": 4.296875e-06, "loss": 3.1085, "step": 63700 }, { "epoch": 9.59, "learning_rate": 4.146634615384616e-06, "loss": 3.1925, "step": 63800 }, { "epoch": 9.6, "learning_rate": 3.9963942307692304e-06, "loss": 3.1547, "step": 63900 }, { "epoch": 9.62, "learning_rate": 3.846153846153847e-06, "loss": 3.0974, "step": 64000 }, { "epoch": 9.63, "learning_rate": 3.695913461538462e-06, "loss": 3.0615, "step": 64100 }, { "epoch": 9.65, "learning_rate": 3.545673076923077e-06, "loss": 3.1745, "step": 64200 }, { "epoch": 9.66, "learning_rate": 3.395432692307692e-06, "loss": 3.0591, "step": 64300 }, { "epoch": 9.68, "learning_rate": 3.245192307692308e-06, "loss": 3.0274, "step": 64400 }, { "epoch": 9.69, "learning_rate": 3.0949519230769235e-06, "loss": 3.0092, "step": 64500 } ], "max_steps": 66560, "num_train_epochs": 10, "total_flos": 1.3030638807239885e+17, "trial_name": null, "trial_params": null }