{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 101625, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 3.0831802676462875e-07, "loss": 4.3168, "step": 100 }, { "epoch": 0.01, "learning_rate": 6.363159275780635e-07, "loss": 4.155, "step": 200 }, { "epoch": 0.01, "learning_rate": 9.643138283914982e-07, "loss": 3.7804, "step": 300 }, { "epoch": 0.01, "learning_rate": 1.2890317501967988e-06, "loss": 3.8055, "step": 400 }, { "epoch": 0.01, "learning_rate": 1.6170296510102337e-06, "loss": 3.6927, "step": 500 }, { "epoch": 0.02, "learning_rate": 1.9450275518236685e-06, "loss": 3.5158, "step": 600 }, { "epoch": 0.02, "learning_rate": 2.273025452637103e-06, "loss": 3.6069, "step": 700 }, { "epoch": 0.02, "learning_rate": 2.6010233534505377e-06, "loss": 3.6429, "step": 800 }, { "epoch": 0.03, "learning_rate": 2.929021254263973e-06, "loss": 3.6292, "step": 900 }, { "epoch": 0.03, "learning_rate": 3.257019155077408e-06, "loss": 3.4969, "step": 1000 }, { "epoch": 0.03, "learning_rate": 3.585017055890842e-06, "loss": 3.5082, "step": 1100 }, { "epoch": 0.04, "learning_rate": 3.9130149567042775e-06, "loss": 3.5132, "step": 1200 }, { "epoch": 0.04, "learning_rate": 4.2410128575177125e-06, "loss": 3.5413, "step": 1300 }, { "epoch": 0.04, "learning_rate": 4.569010758331147e-06, "loss": 3.4286, "step": 1400 }, { "epoch": 0.04, "learning_rate": 4.897008659144582e-06, "loss": 3.4228, "step": 1500 }, { "epoch": 0.05, "learning_rate": 5.225006559958016e-06, "loss": 3.4743, "step": 1600 }, { "epoch": 0.05, "learning_rate": 5.553004460771451e-06, "loss": 3.4929, "step": 1700 }, { "epoch": 0.05, "learning_rate": 5.881002361584886e-06, "loss": 3.4324, "step": 1800 }, { "epoch": 0.06, "learning_rate": 6.20900026239832e-06, "loss": 3.4384, "step": 1900 }, { "epoch": 0.06, "learning_rate": 6.536998163211756e-06, "loss": 3.488, "step": 2000 }, { "epoch": 0.06, "learning_rate": 6.86499606402519e-06, "loss": 3.3966, "step": 2100 }, { "epoch": 0.06, "learning_rate": 7.1929939648386246e-06, "loss": 3.4662, "step": 2200 }, { "epoch": 0.07, "learning_rate": 7.5209918656520605e-06, "loss": 3.5339, "step": 2300 }, { "epoch": 0.07, "learning_rate": 7.848989766465495e-06, "loss": 3.3541, "step": 2400 }, { "epoch": 0.07, "learning_rate": 8.17698766727893e-06, "loss": 3.3954, "step": 2500 }, { "epoch": 0.08, "learning_rate": 8.504985568092365e-06, "loss": 3.3898, "step": 2600 }, { "epoch": 0.08, "learning_rate": 8.8329834689058e-06, "loss": 3.431, "step": 2700 }, { "epoch": 0.08, "learning_rate": 9.160981369719235e-06, "loss": 3.4336, "step": 2800 }, { "epoch": 0.09, "learning_rate": 9.48897927053267e-06, "loss": 3.4402, "step": 2900 }, { "epoch": 0.09, "learning_rate": 9.816977171346103e-06, "loss": 3.3912, "step": 3000 }, { "epoch": 0.09, "learning_rate": 1.0144975072159538e-05, "loss": 3.3787, "step": 3100 }, { "epoch": 0.09, "learning_rate": 1.0472972972972973e-05, "loss": 3.3998, "step": 3200 }, { "epoch": 0.1, "learning_rate": 1.0800970873786408e-05, "loss": 3.4397, "step": 3300 }, { "epoch": 0.1, "learning_rate": 1.1125688795591708e-05, "loss": 3.313, "step": 3400 }, { "epoch": 0.1, "learning_rate": 1.1453686696405144e-05, "loss": 3.3708, "step": 3500 }, { "epoch": 0.11, "learning_rate": 1.1781684597218578e-05, "loss": 3.3924, "step": 3600 }, { "epoch": 0.11, "learning_rate": 1.2109682498032013e-05, "loss": 3.3792, "step": 3700 }, { "epoch": 0.11, "learning_rate": 1.2437680398845448e-05, "loss": 3.3953, "step": 3800 }, { "epoch": 0.12, "learning_rate": 1.2765678299658885e-05, "loss": 3.4286, "step": 3900 }, { "epoch": 0.12, "learning_rate": 1.3093676200472318e-05, "loss": 3.3751, "step": 4000 }, { "epoch": 0.12, "learning_rate": 1.3421674101285753e-05, "loss": 3.4209, "step": 4100 }, { "epoch": 0.12, "learning_rate": 1.3749672002099186e-05, "loss": 3.3963, "step": 4200 }, { "epoch": 0.13, "learning_rate": 1.4077669902912621e-05, "loss": 3.4004, "step": 4300 }, { "epoch": 0.13, "learning_rate": 1.4405667803726055e-05, "loss": 3.3787, "step": 4400 }, { "epoch": 0.13, "learning_rate": 1.4733665704539493e-05, "loss": 3.3521, "step": 4500 }, { "epoch": 0.14, "learning_rate": 1.5061663605352927e-05, "loss": 3.3486, "step": 4600 }, { "epoch": 0.14, "learning_rate": 1.538966150616636e-05, "loss": 3.3643, "step": 4700 }, { "epoch": 0.14, "learning_rate": 1.5717659406979797e-05, "loss": 3.4276, "step": 4800 }, { "epoch": 0.14, "learning_rate": 1.604565730779323e-05, "loss": 3.3843, "step": 4900 }, { "epoch": 0.15, "learning_rate": 1.6373655208606667e-05, "loss": 3.3808, "step": 5000 }, { "epoch": 0.15, "learning_rate": 1.67016531094201e-05, "loss": 3.4145, "step": 5100 }, { "epoch": 0.15, "learning_rate": 1.7029651010233537e-05, "loss": 3.4281, "step": 5200 }, { "epoch": 0.16, "learning_rate": 1.735764891104697e-05, "loss": 3.3493, "step": 5300 }, { "epoch": 0.16, "learning_rate": 1.7685646811860404e-05, "loss": 3.3366, "step": 5400 }, { "epoch": 0.16, "learning_rate": 1.801364471267384e-05, "loss": 3.3874, "step": 5500 }, { "epoch": 0.17, "learning_rate": 1.8341642613487277e-05, "loss": 3.3963, "step": 5600 }, { "epoch": 0.17, "learning_rate": 1.866964051430071e-05, "loss": 3.4015, "step": 5700 }, { "epoch": 0.17, "learning_rate": 1.8997638415114144e-05, "loss": 3.4283, "step": 5800 }, { "epoch": 0.17, "learning_rate": 1.9325636315927577e-05, "loss": 3.3884, "step": 5900 }, { "epoch": 0.18, "learning_rate": 1.9653634216741014e-05, "loss": 3.3508, "step": 6000 }, { "epoch": 0.18, "learning_rate": 1.998163211755445e-05, "loss": 3.3748, "step": 6100 }, { "epoch": 0.18, "learning_rate": 2.0309630018367884e-05, "loss": 3.3285, "step": 6200 }, { "epoch": 0.19, "learning_rate": 2.0637627919181317e-05, "loss": 3.3309, "step": 6300 }, { "epoch": 0.19, "learning_rate": 2.0965625819994754e-05, "loss": 3.3261, "step": 6400 }, { "epoch": 0.19, "learning_rate": 2.1293623720808187e-05, "loss": 3.3698, "step": 6500 }, { "epoch": 0.19, "learning_rate": 2.1621621621621624e-05, "loss": 3.4237, "step": 6600 }, { "epoch": 0.2, "learning_rate": 2.1949619522435057e-05, "loss": 3.3434, "step": 6700 }, { "epoch": 0.2, "learning_rate": 2.227761742324849e-05, "loss": 3.3825, "step": 6800 }, { "epoch": 0.2, "learning_rate": 2.2605615324061928e-05, "loss": 3.4024, "step": 6900 }, { "epoch": 0.21, "learning_rate": 2.293361322487536e-05, "loss": 3.4105, "step": 7000 }, { "epoch": 0.21, "learning_rate": 2.3261611125688798e-05, "loss": 3.3623, "step": 7100 }, { "epoch": 0.21, "learning_rate": 2.358960902650223e-05, "loss": 3.3796, "step": 7200 }, { "epoch": 0.22, "learning_rate": 2.3917606927315668e-05, "loss": 3.4282, "step": 7300 }, { "epoch": 0.22, "learning_rate": 2.42456048281291e-05, "loss": 3.3423, "step": 7400 }, { "epoch": 0.22, "learning_rate": 2.4573602728942534e-05, "loss": 3.3964, "step": 7500 }, { "epoch": 0.22, "learning_rate": 2.490160062975597e-05, "loss": 3.3467, "step": 7600 }, { "epoch": 0.23, "learning_rate": 2.5229598530569405e-05, "loss": 3.3749, "step": 7700 }, { "epoch": 0.23, "learning_rate": 2.555759643138284e-05, "loss": 3.3759, "step": 7800 }, { "epoch": 0.23, "learning_rate": 2.5885594332196278e-05, "loss": 3.3565, "step": 7900 }, { "epoch": 0.24, "learning_rate": 2.6213592233009708e-05, "loss": 3.3313, "step": 8000 }, { "epoch": 0.24, "learning_rate": 2.6541590133823148e-05, "loss": 3.4114, "step": 8100 }, { "epoch": 0.24, "learning_rate": 2.6869588034636578e-05, "loss": 3.4161, "step": 8200 }, { "epoch": 0.25, "learning_rate": 2.7197585935450015e-05, "loss": 3.4302, "step": 8300 }, { "epoch": 0.25, "learning_rate": 2.7525583836263448e-05, "loss": 3.443, "step": 8400 }, { "epoch": 0.25, "learning_rate": 2.7853581737076885e-05, "loss": 3.358, "step": 8500 }, { "epoch": 0.25, "learning_rate": 2.818157963789032e-05, "loss": 3.381, "step": 8600 }, { "epoch": 0.26, "learning_rate": 2.850957753870375e-05, "loss": 3.3726, "step": 8700 }, { "epoch": 0.26, "learning_rate": 2.883757543951719e-05, "loss": 3.494, "step": 8800 }, { "epoch": 0.26, "learning_rate": 2.9165573340330622e-05, "loss": 3.4925, "step": 8900 }, { "epoch": 0.27, "learning_rate": 2.949357124114406e-05, "loss": 3.4443, "step": 9000 }, { "epoch": 0.27, "learning_rate": 2.9821569141957495e-05, "loss": 3.476, "step": 9100 }, { "epoch": 0.27, "learning_rate": 3.0149567042770925e-05, "loss": 3.4143, "step": 9200 }, { "epoch": 0.27, "learning_rate": 3.0474284964576228e-05, "loss": 3.3816, "step": 9300 }, { "epoch": 0.28, "learning_rate": 3.080228286538966e-05, "loss": 3.493, "step": 9400 }, { "epoch": 0.28, "learning_rate": 3.1130280766203095e-05, "loss": 3.4608, "step": 9500 }, { "epoch": 0.28, "learning_rate": 3.1458278667016535e-05, "loss": 3.3971, "step": 9600 }, { "epoch": 0.29, "learning_rate": 3.178627656782997e-05, "loss": 3.4157, "step": 9700 }, { "epoch": 0.29, "learning_rate": 3.21142744686434e-05, "loss": 3.3921, "step": 9800 }, { "epoch": 0.29, "learning_rate": 3.244227236945684e-05, "loss": 3.3785, "step": 9900 }, { "epoch": 0.3, "learning_rate": 3.277027027027027e-05, "loss": 3.3634, "step": 10000 }, { "epoch": 0.3, "learning_rate": 3.309826817108371e-05, "loss": 3.3881, "step": 10100 }, { "epoch": 0.3, "learning_rate": 3.342626607189714e-05, "loss": 3.362, "step": 10200 }, { "epoch": 0.3, "learning_rate": 3.3754263972710575e-05, "loss": 3.4395, "step": 10300 }, { "epoch": 0.31, "learning_rate": 3.4082261873524015e-05, "loss": 3.4151, "step": 10400 }, { "epoch": 0.31, "learning_rate": 3.441025977433744e-05, "loss": 3.3813, "step": 10500 }, { "epoch": 0.31, "learning_rate": 3.473825767515088e-05, "loss": 3.3624, "step": 10600 }, { "epoch": 0.32, "learning_rate": 3.5066255575964315e-05, "loss": 3.5004, "step": 10700 }, { "epoch": 0.32, "learning_rate": 3.539425347677775e-05, "loss": 3.373, "step": 10800 }, { "epoch": 0.32, "learning_rate": 3.572225137759119e-05, "loss": 3.4307, "step": 10900 }, { "epoch": 0.32, "learning_rate": 3.605024927840462e-05, "loss": 3.451, "step": 11000 }, { "epoch": 0.33, "learning_rate": 3.6378247179218055e-05, "loss": 3.3964, "step": 11100 }, { "epoch": 0.33, "learning_rate": 3.670624508003149e-05, "loss": 3.3537, "step": 11200 }, { "epoch": 0.33, "learning_rate": 3.703424298084492e-05, "loss": 3.469, "step": 11300 }, { "epoch": 0.34, "learning_rate": 3.736224088165836e-05, "loss": 3.4307, "step": 11400 }, { "epoch": 0.34, "learning_rate": 3.7690238782471796e-05, "loss": 3.4911, "step": 11500 }, { "epoch": 0.34, "learning_rate": 3.801823668328523e-05, "loss": 3.3976, "step": 11600 }, { "epoch": 0.35, "learning_rate": 3.834623458409866e-05, "loss": 3.4258, "step": 11700 }, { "epoch": 0.35, "learning_rate": 3.8674232484912096e-05, "loss": 3.4217, "step": 11800 }, { "epoch": 0.35, "learning_rate": 3.9002230385725536e-05, "loss": 3.4396, "step": 11900 }, { "epoch": 0.35, "learning_rate": 3.933022828653897e-05, "loss": 3.4841, "step": 12000 }, { "epoch": 0.36, "learning_rate": 3.96582261873524e-05, "loss": 3.4864, "step": 12100 }, { "epoch": 0.36, "learning_rate": 3.9986224088165836e-05, "loss": 3.4933, "step": 12200 }, { "epoch": 0.36, "learning_rate": 4.0314221988979276e-05, "loss": 3.4383, "step": 12300 }, { "epoch": 0.37, "learning_rate": 4.064221988979271e-05, "loss": 3.3793, "step": 12400 }, { "epoch": 0.37, "learning_rate": 4.097021779060614e-05, "loss": 3.4841, "step": 12500 }, { "epoch": 0.37, "learning_rate": 4.1298215691419576e-05, "loss": 3.482, "step": 12600 }, { "epoch": 0.37, "learning_rate": 4.162621359223301e-05, "loss": 3.4459, "step": 12700 }, { "epoch": 0.38, "learning_rate": 4.195421149304645e-05, "loss": 3.4549, "step": 12800 }, { "epoch": 0.38, "learning_rate": 4.228220939385988e-05, "loss": 3.5482, "step": 12900 }, { "epoch": 0.38, "learning_rate": 4.2610207294673316e-05, "loss": 3.4926, "step": 13000 }, { "epoch": 0.39, "learning_rate": 4.293820519548675e-05, "loss": 3.5195, "step": 13100 }, { "epoch": 0.39, "learning_rate": 4.326620309630018e-05, "loss": 3.5106, "step": 13200 }, { "epoch": 0.39, "learning_rate": 4.359420099711362e-05, "loss": 3.4536, "step": 13300 }, { "epoch": 0.4, "learning_rate": 4.392219889792705e-05, "loss": 3.4985, "step": 13400 }, { "epoch": 0.4, "learning_rate": 4.425019679874049e-05, "loss": 3.4308, "step": 13500 }, { "epoch": 0.4, "learning_rate": 4.457491472054579e-05, "loss": 3.4666, "step": 13600 }, { "epoch": 0.4, "learning_rate": 4.4902912621359226e-05, "loss": 3.5624, "step": 13700 }, { "epoch": 0.41, "learning_rate": 4.523091052217266e-05, "loss": 3.4941, "step": 13800 }, { "epoch": 0.41, "learning_rate": 4.55589084229861e-05, "loss": 3.4589, "step": 13900 }, { "epoch": 0.41, "learning_rate": 4.5886906323799526e-05, "loss": 3.5033, "step": 14000 }, { "epoch": 0.42, "learning_rate": 4.6214904224612966e-05, "loss": 3.4525, "step": 14100 }, { "epoch": 0.42, "learning_rate": 4.65429021254264e-05, "loss": 3.5153, "step": 14200 }, { "epoch": 0.42, "learning_rate": 4.687090002623983e-05, "loss": 3.5456, "step": 14300 }, { "epoch": 0.43, "learning_rate": 4.719889792705327e-05, "loss": 3.4878, "step": 14400 }, { "epoch": 0.43, "learning_rate": 4.75268958278667e-05, "loss": 3.4686, "step": 14500 }, { "epoch": 0.43, "learning_rate": 4.785489372868014e-05, "loss": 3.4891, "step": 14600 }, { "epoch": 0.43, "learning_rate": 4.818289162949357e-05, "loss": 3.4157, "step": 14700 }, { "epoch": 0.44, "learning_rate": 4.8510889530307006e-05, "loss": 3.444, "step": 14800 }, { "epoch": 0.44, "learning_rate": 4.8838887431120446e-05, "loss": 3.5387, "step": 14900 }, { "epoch": 0.44, "learning_rate": 4.916688533193387e-05, "loss": 3.5091, "step": 15000 }, { "epoch": 0.45, "learning_rate": 4.949488323274731e-05, "loss": 3.5356, "step": 15100 }, { "epoch": 0.45, "learning_rate": 4.9822881133560746e-05, "loss": 3.4978, "step": 15200 }, { "epoch": 0.45, "learning_rate": 4.997337377432538e-05, "loss": 3.4566, "step": 15300 }, { "epoch": 0.45, "learning_rate": 4.991549067503271e-05, "loss": 3.5653, "step": 15400 }, { "epoch": 0.46, "learning_rate": 4.985760757574004e-05, "loss": 3.5342, "step": 15500 }, { "epoch": 0.46, "learning_rate": 4.979972447644737e-05, "loss": 3.5189, "step": 15600 }, { "epoch": 0.46, "learning_rate": 4.97418413771547e-05, "loss": 3.5025, "step": 15700 }, { "epoch": 0.47, "learning_rate": 4.968395827786203e-05, "loss": 3.5216, "step": 15800 }, { "epoch": 0.47, "learning_rate": 4.9626075178569366e-05, "loss": 3.5428, "step": 15900 }, { "epoch": 0.47, "learning_rate": 4.9568192079276696e-05, "loss": 3.5112, "step": 16000 }, { "epoch": 0.48, "learning_rate": 4.9510308979984026e-05, "loss": 3.4538, "step": 16100 }, { "epoch": 0.48, "learning_rate": 4.9452425880691356e-05, "loss": 3.5476, "step": 16200 }, { "epoch": 0.48, "learning_rate": 4.9394542781398686e-05, "loss": 3.5039, "step": 16300 }, { "epoch": 0.48, "learning_rate": 4.933665968210602e-05, "loss": 3.4226, "step": 16400 }, { "epoch": 0.49, "learning_rate": 4.927877658281335e-05, "loss": 3.5409, "step": 16500 }, { "epoch": 0.49, "learning_rate": 4.922089348352068e-05, "loss": 3.4823, "step": 16600 }, { "epoch": 0.49, "learning_rate": 4.916301038422801e-05, "loss": 3.5517, "step": 16700 }, { "epoch": 0.5, "learning_rate": 4.910512728493535e-05, "loss": 3.5905, "step": 16800 }, { "epoch": 0.5, "learning_rate": 4.9047244185642673e-05, "loss": 3.5193, "step": 16900 }, { "epoch": 0.5, "learning_rate": 4.898936108635001e-05, "loss": 3.5474, "step": 17000 }, { "epoch": 0.5, "learning_rate": 4.893147798705734e-05, "loss": 3.5083, "step": 17100 }, { "epoch": 0.51, "learning_rate": 4.887359488776467e-05, "loss": 3.5096, "step": 17200 }, { "epoch": 0.51, "learning_rate": 4.8815711788472e-05, "loss": 3.4946, "step": 17300 }, { "epoch": 0.51, "learning_rate": 4.875782868917934e-05, "loss": 3.5561, "step": 17400 }, { "epoch": 0.52, "learning_rate": 4.869994558988667e-05, "loss": 3.5484, "step": 17500 }, { "epoch": 0.52, "learning_rate": 4.864264132158692e-05, "loss": 3.473, "step": 17600 }, { "epoch": 0.52, "learning_rate": 4.858475822229426e-05, "loss": 3.4777, "step": 17700 }, { "epoch": 0.53, "learning_rate": 4.852687512300159e-05, "loss": 3.5096, "step": 17800 }, { "epoch": 0.53, "learning_rate": 4.846899202370892e-05, "loss": 3.5513, "step": 17900 }, { "epoch": 0.53, "learning_rate": 4.841110892441625e-05, "loss": 3.54, "step": 18000 }, { "epoch": 0.53, "learning_rate": 4.8353225825123584e-05, "loss": 3.4855, "step": 18100 }, { "epoch": 0.54, "learning_rate": 4.8295342725830915e-05, "loss": 3.4912, "step": 18200 }, { "epoch": 0.54, "learning_rate": 4.823803845753117e-05, "loss": 3.4644, "step": 18300 }, { "epoch": 0.54, "learning_rate": 4.8180155358238504e-05, "loss": 3.5219, "step": 18400 }, { "epoch": 0.55, "learning_rate": 4.8122272258945834e-05, "loss": 3.4587, "step": 18500 }, { "epoch": 0.55, "learning_rate": 4.8064389159653164e-05, "loss": 3.5399, "step": 18600 }, { "epoch": 0.55, "learning_rate": 4.8006506060360494e-05, "loss": 3.5016, "step": 18700 }, { "epoch": 0.55, "learning_rate": 4.794862296106783e-05, "loss": 3.4807, "step": 18800 }, { "epoch": 0.56, "learning_rate": 4.789073986177516e-05, "loss": 3.5076, "step": 18900 }, { "epoch": 0.56, "learning_rate": 4.783285676248249e-05, "loss": 3.4874, "step": 19000 }, { "epoch": 0.56, "learning_rate": 4.777555249418275e-05, "loss": 3.4773, "step": 19100 }, { "epoch": 0.57, "learning_rate": 4.771766939489008e-05, "loss": 3.5177, "step": 19200 }, { "epoch": 0.57, "learning_rate": 4.765978629559741e-05, "loss": 3.4338, "step": 19300 }, { "epoch": 0.57, "learning_rate": 4.760190319630475e-05, "loss": 3.4832, "step": 19400 }, { "epoch": 0.58, "learning_rate": 4.754402009701208e-05, "loss": 3.3966, "step": 19500 }, { "epoch": 0.58, "learning_rate": 4.748613699771941e-05, "loss": 3.5248, "step": 19600 }, { "epoch": 0.58, "learning_rate": 4.742825389842674e-05, "loss": 3.5104, "step": 19700 }, { "epoch": 0.58, "learning_rate": 4.7370370799134075e-05, "loss": 3.4758, "step": 19800 }, { "epoch": 0.59, "learning_rate": 4.73124876998414e-05, "loss": 3.468, "step": 19900 }, { "epoch": 0.59, "learning_rate": 4.7254604600548735e-05, "loss": 3.5482, "step": 20000 }, { "epoch": 0.59, "learning_rate": 4.7196721501256066e-05, "loss": 3.4968, "step": 20100 }, { "epoch": 0.6, "learning_rate": 4.7138838401963396e-05, "loss": 3.4743, "step": 20200 }, { "epoch": 0.6, "learning_rate": 4.7080955302670726e-05, "loss": 3.4734, "step": 20300 }, { "epoch": 0.6, "learning_rate": 4.702307220337806e-05, "loss": 3.4668, "step": 20400 }, { "epoch": 0.61, "learning_rate": 4.6965189104085386e-05, "loss": 3.4479, "step": 20500 }, { "epoch": 0.61, "learning_rate": 4.690730600479272e-05, "loss": 3.4609, "step": 20600 }, { "epoch": 0.61, "learning_rate": 4.684942290550006e-05, "loss": 3.5061, "step": 20700 }, { "epoch": 0.61, "learning_rate": 4.679153980620738e-05, "loss": 3.474, "step": 20800 }, { "epoch": 0.62, "learning_rate": 4.673365670691472e-05, "loss": 3.46, "step": 20900 }, { "epoch": 0.62, "learning_rate": 4.667577360762205e-05, "loss": 3.4596, "step": 21000 }, { "epoch": 0.62, "learning_rate": 4.661789050832938e-05, "loss": 3.5117, "step": 21100 }, { "epoch": 0.63, "learning_rate": 4.656000740903671e-05, "loss": 3.5576, "step": 21200 }, { "epoch": 0.63, "learning_rate": 4.650212430974405e-05, "loss": 3.5696, "step": 21300 }, { "epoch": 0.63, "learning_rate": 4.644424121045137e-05, "loss": 3.4833, "step": 21400 }, { "epoch": 0.63, "learning_rate": 4.638635811115871e-05, "loss": 3.4939, "step": 21500 }, { "epoch": 0.64, "learning_rate": 4.632905384285897e-05, "loss": 3.4656, "step": 21600 }, { "epoch": 0.64, "learning_rate": 4.62711707435663e-05, "loss": 3.4646, "step": 21700 }, { "epoch": 0.64, "learning_rate": 4.621328764427363e-05, "loss": 3.5053, "step": 21800 }, { "epoch": 0.65, "learning_rate": 4.615540454498096e-05, "loss": 3.426, "step": 21900 }, { "epoch": 0.65, "learning_rate": 4.6097521445688294e-05, "loss": 3.5011, "step": 22000 }, { "epoch": 0.65, "learning_rate": 4.603963834639562e-05, "loss": 3.5055, "step": 22100 }, { "epoch": 0.66, "learning_rate": 4.5981755247102954e-05, "loss": 3.5099, "step": 22200 }, { "epoch": 0.66, "learning_rate": 4.5923872147810284e-05, "loss": 3.4436, "step": 22300 }, { "epoch": 0.66, "learning_rate": 4.5865989048517614e-05, "loss": 3.4445, "step": 22400 }, { "epoch": 0.66, "learning_rate": 4.580810594922495e-05, "loss": 3.4688, "step": 22500 }, { "epoch": 0.67, "learning_rate": 4.575022284993228e-05, "loss": 3.4393, "step": 22600 }, { "epoch": 0.67, "learning_rate": 4.569233975063961e-05, "loss": 3.455, "step": 22700 }, { "epoch": 0.67, "learning_rate": 4.563445665134694e-05, "loss": 3.47, "step": 22800 }, { "epoch": 0.68, "learning_rate": 4.557657355205427e-05, "loss": 3.4016, "step": 22900 }, { "epoch": 0.68, "learning_rate": 4.55186904527616e-05, "loss": 3.4426, "step": 23000 }, { "epoch": 0.68, "learning_rate": 4.546080735346894e-05, "loss": 3.4976, "step": 23100 }, { "epoch": 0.68, "learning_rate": 4.540292425417627e-05, "loss": 3.4368, "step": 23200 }, { "epoch": 0.69, "learning_rate": 4.53450411548836e-05, "loss": 3.439, "step": 23300 }, { "epoch": 0.69, "learning_rate": 4.528715805559093e-05, "loss": 3.5062, "step": 23400 }, { "epoch": 0.69, "learning_rate": 4.522927495629826e-05, "loss": 3.4495, "step": 23500 }, { "epoch": 0.7, "learning_rate": 4.5171391857005596e-05, "loss": 3.4425, "step": 23600 }, { "epoch": 0.7, "learning_rate": 4.5113508757712926e-05, "loss": 3.4741, "step": 23700 }, { "epoch": 0.7, "learning_rate": 4.5055625658420256e-05, "loss": 3.4746, "step": 23800 }, { "epoch": 0.71, "learning_rate": 4.4997742559127586e-05, "loss": 3.3827, "step": 23900 }, { "epoch": 0.71, "learning_rate": 4.493985945983492e-05, "loss": 3.3663, "step": 24000 }, { "epoch": 0.71, "learning_rate": 4.4881976360542246e-05, "loss": 3.4349, "step": 24100 }, { "epoch": 0.71, "learning_rate": 4.482409326124958e-05, "loss": 3.4367, "step": 24200 }, { "epoch": 0.72, "learning_rate": 4.4766210161956913e-05, "loss": 3.3795, "step": 24300 }, { "epoch": 0.72, "learning_rate": 4.4708327062664244e-05, "loss": 3.4992, "step": 24400 }, { "epoch": 0.72, "learning_rate": 4.4650443963371574e-05, "loss": 3.4647, "step": 24500 }, { "epoch": 0.73, "learning_rate": 4.459256086407891e-05, "loss": 3.441, "step": 24600 }, { "epoch": 0.73, "learning_rate": 4.453467776478624e-05, "loss": 3.4718, "step": 24700 }, { "epoch": 0.73, "learning_rate": 4.447679466549357e-05, "loss": 3.4318, "step": 24800 }, { "epoch": 0.74, "learning_rate": 4.441891156620091e-05, "loss": 3.4683, "step": 24900 }, { "epoch": 0.74, "learning_rate": 4.436102846690823e-05, "loss": 3.4339, "step": 25000 }, { "epoch": 0.74, "learning_rate": 4.430314536761557e-05, "loss": 3.4285, "step": 25100 }, { "epoch": 0.74, "learning_rate": 4.42452622683229e-05, "loss": 3.3831, "step": 25200 }, { "epoch": 0.75, "learning_rate": 4.418737916903023e-05, "loss": 3.4561, "step": 25300 }, { "epoch": 0.75, "learning_rate": 4.412949606973756e-05, "loss": 3.4521, "step": 25400 }, { "epoch": 0.75, "learning_rate": 4.4071612970444895e-05, "loss": 3.4461, "step": 25500 }, { "epoch": 0.76, "learning_rate": 4.401372987115222e-05, "loss": 3.52, "step": 25600 }, { "epoch": 0.76, "learning_rate": 4.3955846771859555e-05, "loss": 3.4293, "step": 25700 }, { "epoch": 0.76, "learning_rate": 4.3897963672566885e-05, "loss": 3.3902, "step": 25800 }, { "epoch": 0.76, "learning_rate": 4.3840080573274215e-05, "loss": 3.4207, "step": 25900 }, { "epoch": 0.77, "learning_rate": 4.378219747398155e-05, "loss": 3.5207, "step": 26000 }, { "epoch": 0.77, "learning_rate": 4.372431437468888e-05, "loss": 3.3999, "step": 26100 }, { "epoch": 0.77, "learning_rate": 4.366643127539621e-05, "loss": 3.4113, "step": 26200 }, { "epoch": 0.78, "learning_rate": 4.360854817610354e-05, "loss": 3.4914, "step": 26300 }, { "epoch": 0.78, "learning_rate": 4.355066507681087e-05, "loss": 3.4771, "step": 26400 }, { "epoch": 0.78, "learning_rate": 4.349336080851113e-05, "loss": 3.4838, "step": 26500 }, { "epoch": 0.79, "learning_rate": 4.343547770921846e-05, "loss": 3.4307, "step": 26600 }, { "epoch": 0.79, "learning_rate": 4.33775946099258e-05, "loss": 3.539, "step": 26700 }, { "epoch": 0.79, "learning_rate": 4.331971151063313e-05, "loss": 3.4617, "step": 26800 }, { "epoch": 0.79, "learning_rate": 4.326182841134046e-05, "loss": 3.4023, "step": 26900 }, { "epoch": 0.8, "learning_rate": 4.320394531204779e-05, "loss": 3.4677, "step": 27000 }, { "epoch": 0.8, "learning_rate": 4.314606221275512e-05, "loss": 3.4065, "step": 27100 }, { "epoch": 0.8, "learning_rate": 4.308817911346245e-05, "loss": 3.4339, "step": 27200 }, { "epoch": 0.81, "learning_rate": 4.3030296014169787e-05, "loss": 3.4403, "step": 27300 }, { "epoch": 0.81, "learning_rate": 4.297241291487712e-05, "loss": 3.4626, "step": 27400 }, { "epoch": 0.81, "learning_rate": 4.291452981558445e-05, "loss": 3.4698, "step": 27500 }, { "epoch": 0.81, "learning_rate": 4.2856646716291784e-05, "loss": 3.4339, "step": 27600 }, { "epoch": 0.82, "learning_rate": 4.279876361699911e-05, "loss": 3.3525, "step": 27700 }, { "epoch": 0.82, "learning_rate": 4.2740880517706444e-05, "loss": 3.4385, "step": 27800 }, { "epoch": 0.82, "learning_rate": 4.2682997418413774e-05, "loss": 3.3924, "step": 27900 }, { "epoch": 0.83, "learning_rate": 4.2625114319121104e-05, "loss": 3.4542, "step": 28000 }, { "epoch": 0.83, "learning_rate": 4.2567810050821364e-05, "loss": 3.439, "step": 28100 }, { "epoch": 0.83, "learning_rate": 4.2509926951528694e-05, "loss": 3.4499, "step": 28200 }, { "epoch": 0.84, "learning_rate": 4.245204385223603e-05, "loss": 3.4655, "step": 28300 }, { "epoch": 0.84, "learning_rate": 4.2394160752943354e-05, "loss": 3.4605, "step": 28400 }, { "epoch": 0.84, "learning_rate": 4.233627765365069e-05, "loss": 3.413, "step": 28500 }, { "epoch": 0.84, "learning_rate": 4.227839455435802e-05, "loss": 3.4802, "step": 28600 }, { "epoch": 0.85, "learning_rate": 4.222051145506535e-05, "loss": 3.4359, "step": 28700 }, { "epoch": 0.85, "learning_rate": 4.216262835577268e-05, "loss": 3.3859, "step": 28800 }, { "epoch": 0.85, "learning_rate": 4.210474525648002e-05, "loss": 3.3875, "step": 28900 }, { "epoch": 0.86, "learning_rate": 4.204686215718734e-05, "loss": 3.3715, "step": 29000 }, { "epoch": 0.86, "learning_rate": 4.198897905789468e-05, "loss": 3.3754, "step": 29100 }, { "epoch": 0.86, "learning_rate": 4.193109595860201e-05, "loss": 3.3918, "step": 29200 }, { "epoch": 0.86, "learning_rate": 4.187321285930934e-05, "loss": 3.4793, "step": 29300 }, { "epoch": 0.87, "learning_rate": 4.1815329760016675e-05, "loss": 3.4157, "step": 29400 }, { "epoch": 0.87, "learning_rate": 4.1757446660724005e-05, "loss": 3.4226, "step": 29500 }, { "epoch": 0.87, "learning_rate": 4.1699563561431335e-05, "loss": 3.4078, "step": 29600 }, { "epoch": 0.88, "learning_rate": 4.1641680462138665e-05, "loss": 3.4142, "step": 29700 }, { "epoch": 0.88, "learning_rate": 4.1583797362846e-05, "loss": 3.4222, "step": 29800 }, { "epoch": 0.88, "learning_rate": 4.1525914263553326e-05, "loss": 3.4325, "step": 29900 }, { "epoch": 0.89, "learning_rate": 4.146803116426066e-05, "loss": 3.4555, "step": 30000 }, { "epoch": 0.89, "learning_rate": 4.141014806496799e-05, "loss": 3.432, "step": 30100 }, { "epoch": 0.89, "learning_rate": 4.135226496567532e-05, "loss": 3.3894, "step": 30200 }, { "epoch": 0.89, "learning_rate": 4.129438186638265e-05, "loss": 3.4094, "step": 30300 }, { "epoch": 0.9, "learning_rate": 4.123649876708999e-05, "loss": 3.4366, "step": 30400 }, { "epoch": 0.9, "learning_rate": 4.117861566779732e-05, "loss": 3.4109, "step": 30500 }, { "epoch": 0.9, "learning_rate": 4.112073256850465e-05, "loss": 3.4778, "step": 30600 }, { "epoch": 0.91, "learning_rate": 4.106284946921199e-05, "loss": 3.4626, "step": 30700 }, { "epoch": 0.91, "learning_rate": 4.100496636991931e-05, "loss": 3.3753, "step": 30800 }, { "epoch": 0.91, "learning_rate": 4.094708327062665e-05, "loss": 3.4158, "step": 30900 }, { "epoch": 0.92, "learning_rate": 4.088920017133398e-05, "loss": 3.4925, "step": 31000 }, { "epoch": 0.92, "learning_rate": 4.083131707204131e-05, "loss": 3.4383, "step": 31100 }, { "epoch": 0.92, "learning_rate": 4.077343397274864e-05, "loss": 3.3489, "step": 31200 }, { "epoch": 0.92, "learning_rate": 4.0715550873455974e-05, "loss": 3.3988, "step": 31300 }, { "epoch": 0.93, "learning_rate": 4.06576677741633e-05, "loss": 3.4295, "step": 31400 }, { "epoch": 0.93, "learning_rate": 4.0599784674870634e-05, "loss": 3.4186, "step": 31500 }, { "epoch": 0.93, "learning_rate": 4.0541901575577965e-05, "loss": 3.4509, "step": 31600 }, { "epoch": 0.94, "learning_rate": 4.0484018476285295e-05, "loss": 3.3621, "step": 31700 }, { "epoch": 0.94, "learning_rate": 4.042613537699263e-05, "loss": 3.4542, "step": 31800 }, { "epoch": 0.94, "learning_rate": 4.036825227769996e-05, "loss": 3.3612, "step": 31900 }, { "epoch": 0.94, "learning_rate": 4.031036917840729e-05, "loss": 3.3997, "step": 32000 }, { "epoch": 0.95, "learning_rate": 4.025248607911462e-05, "loss": 3.4172, "step": 32100 }, { "epoch": 0.95, "learning_rate": 4.019460297982195e-05, "loss": 3.4839, "step": 32200 }, { "epoch": 0.95, "learning_rate": 4.013671988052928e-05, "loss": 3.3381, "step": 32300 }, { "epoch": 0.96, "learning_rate": 4.007883678123662e-05, "loss": 3.4331, "step": 32400 }, { "epoch": 0.96, "learning_rate": 4.002095368194395e-05, "loss": 3.3947, "step": 32500 }, { "epoch": 0.96, "learning_rate": 3.996307058265128e-05, "loss": 3.3557, "step": 32600 }, { "epoch": 0.97, "learning_rate": 3.990518748335861e-05, "loss": 3.3993, "step": 32700 }, { "epoch": 0.97, "learning_rate": 3.984730438406594e-05, "loss": 3.3265, "step": 32800 }, { "epoch": 0.97, "learning_rate": 3.9789421284773276e-05, "loss": 3.3464, "step": 32900 }, { "epoch": 0.97, "learning_rate": 3.9731538185480606e-05, "loss": 3.354, "step": 33000 }, { "epoch": 0.98, "learning_rate": 3.9673655086187936e-05, "loss": 3.4181, "step": 33100 }, { "epoch": 0.98, "learning_rate": 3.9615771986895266e-05, "loss": 3.3932, "step": 33200 }, { "epoch": 0.98, "learning_rate": 3.9558467718595526e-05, "loss": 3.3718, "step": 33300 }, { "epoch": 0.99, "learning_rate": 3.950058461930286e-05, "loss": 3.3526, "step": 33400 }, { "epoch": 0.99, "learning_rate": 3.9442701520010186e-05, "loss": 3.4861, "step": 33500 }, { "epoch": 0.99, "learning_rate": 3.938481842071752e-05, "loss": 3.4428, "step": 33600 }, { "epoch": 0.99, "learning_rate": 3.932693532142485e-05, "loss": 3.4784, "step": 33700 }, { "epoch": 1.0, "learning_rate": 3.926905222213218e-05, "loss": 3.3259, "step": 33800 }, { "epoch": 1.0, "eval_gen_len": 19.3901, "eval_loss": 3.25349760055542, "eval_rouge1": 17.942, "eval_rouge2": 4.5143, "eval_rougeL": 14.2766, "eval_rougeLsum": 15.582, "eval_runtime": 746.4697, "eval_samples_per_second": 10.083, "eval_steps_per_second": 2.521, "step": 33875 }, { "epoch": 1.0, "learning_rate": 3.921116912283951e-05, "loss": 3.3341, "step": 33900 }, { "epoch": 1.0, "learning_rate": 3.915328602354685e-05, "loss": 3.2272, "step": 34000 }, { "epoch": 1.01, "learning_rate": 3.9095402924254174e-05, "loss": 3.1016, "step": 34100 }, { "epoch": 1.01, "learning_rate": 3.903751982496151e-05, "loss": 3.0869, "step": 34200 }, { "epoch": 1.01, "learning_rate": 3.897963672566884e-05, "loss": 3.0574, "step": 34300 }, { "epoch": 1.02, "learning_rate": 3.892175362637617e-05, "loss": 3.0656, "step": 34400 }, { "epoch": 1.02, "learning_rate": 3.88638705270835e-05, "loss": 3.0487, "step": 34500 }, { "epoch": 1.02, "learning_rate": 3.880598742779084e-05, "loss": 3.0965, "step": 34600 }, { "epoch": 1.02, "learning_rate": 3.874810432849817e-05, "loss": 3.1261, "step": 34700 }, { "epoch": 1.03, "learning_rate": 3.86902212292055e-05, "loss": 3.0195, "step": 34800 }, { "epoch": 1.03, "learning_rate": 3.8632338129912835e-05, "loss": 3.1102, "step": 34900 }, { "epoch": 1.03, "learning_rate": 3.857445503062016e-05, "loss": 3.0478, "step": 35000 }, { "epoch": 1.04, "learning_rate": 3.8516571931327495e-05, "loss": 3.0921, "step": 35100 }, { "epoch": 1.04, "learning_rate": 3.8458688832034825e-05, "loss": 3.0725, "step": 35200 }, { "epoch": 1.04, "learning_rate": 3.8400805732742155e-05, "loss": 3.0383, "step": 35300 }, { "epoch": 1.05, "learning_rate": 3.8342922633449485e-05, "loss": 3.0486, "step": 35400 }, { "epoch": 1.05, "learning_rate": 3.828503953415682e-05, "loss": 3.0608, "step": 35500 }, { "epoch": 1.05, "learning_rate": 3.8227156434864145e-05, "loss": 3.09, "step": 35600 }, { "epoch": 1.05, "learning_rate": 3.8169852166564405e-05, "loss": 3.0758, "step": 35700 }, { "epoch": 1.06, "learning_rate": 3.811196906727174e-05, "loss": 3.0034, "step": 35800 }, { "epoch": 1.06, "learning_rate": 3.805408596797907e-05, "loss": 3.1317, "step": 35900 }, { "epoch": 1.06, "learning_rate": 3.79962028686864e-05, "loss": 3.09, "step": 36000 }, { "epoch": 1.07, "learning_rate": 3.793831976939373e-05, "loss": 3.1168, "step": 36100 }, { "epoch": 1.07, "learning_rate": 3.788043667010107e-05, "loss": 3.115, "step": 36200 }, { "epoch": 1.07, "learning_rate": 3.782255357080839e-05, "loss": 3.0612, "step": 36300 }, { "epoch": 1.07, "learning_rate": 3.776467047151573e-05, "loss": 3.0553, "step": 36400 }, { "epoch": 1.08, "learning_rate": 3.770678737222306e-05, "loss": 3.0838, "step": 36500 }, { "epoch": 1.08, "learning_rate": 3.764890427293039e-05, "loss": 3.13, "step": 36600 }, { "epoch": 1.08, "learning_rate": 3.7591021173637726e-05, "loss": 3.0685, "step": 36700 }, { "epoch": 1.09, "learning_rate": 3.7533138074345056e-05, "loss": 3.0328, "step": 36800 }, { "epoch": 1.09, "learning_rate": 3.7475254975052386e-05, "loss": 3.0607, "step": 36900 }, { "epoch": 1.09, "learning_rate": 3.7417371875759717e-05, "loss": 3.0557, "step": 37000 }, { "epoch": 1.1, "learning_rate": 3.735948877646705e-05, "loss": 3.1025, "step": 37100 }, { "epoch": 1.1, "learning_rate": 3.730160567717438e-05, "loss": 3.0598, "step": 37200 }, { "epoch": 1.1, "learning_rate": 3.7243722577881714e-05, "loss": 3.1292, "step": 37300 }, { "epoch": 1.1, "learning_rate": 3.7185839478589044e-05, "loss": 3.0739, "step": 37400 }, { "epoch": 1.11, "learning_rate": 3.7127956379296374e-05, "loss": 3.1366, "step": 37500 }, { "epoch": 1.11, "learning_rate": 3.707007328000371e-05, "loss": 3.0558, "step": 37600 }, { "epoch": 1.11, "learning_rate": 3.7012190180711034e-05, "loss": 3.0975, "step": 37700 }, { "epoch": 1.12, "learning_rate": 3.695430708141837e-05, "loss": 3.0948, "step": 37800 }, { "epoch": 1.12, "learning_rate": 3.68964239821257e-05, "loss": 3.1102, "step": 37900 }, { "epoch": 1.12, "learning_rate": 3.683854088283303e-05, "loss": 3.1117, "step": 38000 }, { "epoch": 1.12, "learning_rate": 3.678065778354036e-05, "loss": 3.007, "step": 38100 }, { "epoch": 1.13, "learning_rate": 3.67227746842477e-05, "loss": 2.9965, "step": 38200 }, { "epoch": 1.13, "learning_rate": 3.666489158495502e-05, "loss": 3.0942, "step": 38300 }, { "epoch": 1.13, "learning_rate": 3.660700848566236e-05, "loss": 3.0544, "step": 38400 }, { "epoch": 1.14, "learning_rate": 3.654912538636969e-05, "loss": 3.071, "step": 38500 }, { "epoch": 1.14, "learning_rate": 3.649124228707702e-05, "loss": 3.0174, "step": 38600 }, { "epoch": 1.14, "learning_rate": 3.6433359187784355e-05, "loss": 3.0719, "step": 38700 }, { "epoch": 1.15, "learning_rate": 3.6375476088491685e-05, "loss": 3.0904, "step": 38800 }, { "epoch": 1.15, "learning_rate": 3.6317592989199016e-05, "loss": 3.0974, "step": 38900 }, { "epoch": 1.15, "learning_rate": 3.6259709889906346e-05, "loss": 3.0983, "step": 39000 }, { "epoch": 1.15, "learning_rate": 3.620182679061368e-05, "loss": 3.092, "step": 39100 }, { "epoch": 1.16, "learning_rate": 3.6143943691321006e-05, "loss": 3.0146, "step": 39200 }, { "epoch": 1.16, "learning_rate": 3.608606059202834e-05, "loss": 3.1821, "step": 39300 }, { "epoch": 1.16, "learning_rate": 3.602817749273567e-05, "loss": 3.1287, "step": 39400 }, { "epoch": 1.17, "learning_rate": 3.5970294393443e-05, "loss": 3.0698, "step": 39500 }, { "epoch": 1.17, "learning_rate": 3.591241129415033e-05, "loss": 3.0935, "step": 39600 }, { "epoch": 1.17, "learning_rate": 3.585510702585059e-05, "loss": 3.0959, "step": 39700 }, { "epoch": 1.17, "learning_rate": 3.579722392655793e-05, "loss": 3.0677, "step": 39800 }, { "epoch": 1.18, "learning_rate": 3.573934082726525e-05, "loss": 3.12, "step": 39900 }, { "epoch": 1.18, "learning_rate": 3.568145772797259e-05, "loss": 3.0438, "step": 40000 }, { "epoch": 1.18, "learning_rate": 3.562357462867992e-05, "loss": 3.1065, "step": 40100 }, { "epoch": 1.19, "learning_rate": 3.556569152938725e-05, "loss": 3.122, "step": 40200 }, { "epoch": 1.19, "learning_rate": 3.550780843009458e-05, "loss": 3.0528, "step": 40300 }, { "epoch": 1.19, "learning_rate": 3.544992533080192e-05, "loss": 3.0939, "step": 40400 }, { "epoch": 1.2, "learning_rate": 3.539204223150925e-05, "loss": 3.0964, "step": 40500 }, { "epoch": 1.2, "learning_rate": 3.533415913221658e-05, "loss": 3.0712, "step": 40600 }, { "epoch": 1.2, "learning_rate": 3.5276276032923914e-05, "loss": 3.0369, "step": 40700 }, { "epoch": 1.2, "learning_rate": 3.521839293363124e-05, "loss": 3.08, "step": 40800 }, { "epoch": 1.21, "learning_rate": 3.5160509834338574e-05, "loss": 3.1407, "step": 40900 }, { "epoch": 1.21, "learning_rate": 3.5102626735045904e-05, "loss": 3.0949, "step": 41000 }, { "epoch": 1.21, "learning_rate": 3.5044743635753234e-05, "loss": 3.157, "step": 41100 }, { "epoch": 1.22, "learning_rate": 3.4986860536460564e-05, "loss": 3.0251, "step": 41200 }, { "epoch": 1.22, "learning_rate": 3.49289774371679e-05, "loss": 3.1212, "step": 41300 }, { "epoch": 1.22, "learning_rate": 3.4871094337875225e-05, "loss": 3.1201, "step": 41400 }, { "epoch": 1.23, "learning_rate": 3.481321123858256e-05, "loss": 3.1373, "step": 41500 }, { "epoch": 1.23, "learning_rate": 3.475532813928989e-05, "loss": 3.1454, "step": 41600 }, { "epoch": 1.23, "learning_rate": 3.469802387099015e-05, "loss": 3.1846, "step": 41700 }, { "epoch": 1.23, "learning_rate": 3.464014077169748e-05, "loss": 3.075, "step": 41800 }, { "epoch": 1.24, "learning_rate": 3.458225767240481e-05, "loss": 3.1067, "step": 41900 }, { "epoch": 1.24, "learning_rate": 3.452437457311215e-05, "loss": 3.0698, "step": 42000 }, { "epoch": 1.24, "learning_rate": 3.446649147381947e-05, "loss": 3.197, "step": 42100 }, { "epoch": 1.25, "learning_rate": 3.440860837452681e-05, "loss": 3.0678, "step": 42200 }, { "epoch": 1.25, "learning_rate": 3.435072527523414e-05, "loss": 3.111, "step": 42300 }, { "epoch": 1.25, "learning_rate": 3.429284217594147e-05, "loss": 3.1098, "step": 42400 }, { "epoch": 1.25, "learning_rate": 3.4234959076648805e-05, "loss": 3.0416, "step": 42500 }, { "epoch": 1.26, "learning_rate": 3.4177075977356136e-05, "loss": 3.0468, "step": 42600 }, { "epoch": 1.26, "learning_rate": 3.4119192878063466e-05, "loss": 3.1422, "step": 42700 }, { "epoch": 1.26, "learning_rate": 3.4061309778770796e-05, "loss": 3.1238, "step": 42800 }, { "epoch": 1.27, "learning_rate": 3.4003426679478126e-05, "loss": 3.0876, "step": 42900 }, { "epoch": 1.27, "learning_rate": 3.3945543580185456e-05, "loss": 3.0583, "step": 43000 }, { "epoch": 1.27, "learning_rate": 3.388766048089279e-05, "loss": 3.1087, "step": 43100 }, { "epoch": 1.28, "learning_rate": 3.382977738160012e-05, "loss": 3.0599, "step": 43200 }, { "epoch": 1.28, "learning_rate": 3.377189428230745e-05, "loss": 3.0935, "step": 43300 }, { "epoch": 1.28, "learning_rate": 3.371401118301479e-05, "loss": 3.1727, "step": 43400 }, { "epoch": 1.28, "learning_rate": 3.365612808372211e-05, "loss": 3.0115, "step": 43500 }, { "epoch": 1.29, "learning_rate": 3.359824498442945e-05, "loss": 3.1088, "step": 43600 }, { "epoch": 1.29, "learning_rate": 3.354036188513678e-05, "loss": 3.0485, "step": 43700 }, { "epoch": 1.29, "learning_rate": 3.348247878584411e-05, "loss": 3.102, "step": 43800 }, { "epoch": 1.3, "learning_rate": 3.342459568655144e-05, "loss": 3.0692, "step": 43900 }, { "epoch": 1.3, "learning_rate": 3.336671258725878e-05, "loss": 3.1147, "step": 44000 }, { "epoch": 1.3, "learning_rate": 3.33088294879661e-05, "loss": 3.1878, "step": 44100 }, { "epoch": 1.3, "learning_rate": 3.325152521966636e-05, "loss": 3.0811, "step": 44200 }, { "epoch": 1.31, "learning_rate": 3.31936421203737e-05, "loss": 3.0288, "step": 44300 }, { "epoch": 1.31, "learning_rate": 3.313575902108103e-05, "loss": 3.0671, "step": 44400 }, { "epoch": 1.31, "learning_rate": 3.307787592178836e-05, "loss": 3.0503, "step": 44500 }, { "epoch": 1.32, "learning_rate": 3.301999282249569e-05, "loss": 3.0966, "step": 44600 }, { "epoch": 1.32, "learning_rate": 3.2962109723203024e-05, "loss": 3.0395, "step": 44700 }, { "epoch": 1.32, "learning_rate": 3.290422662391035e-05, "loss": 3.0602, "step": 44800 }, { "epoch": 1.33, "learning_rate": 3.2846343524617684e-05, "loss": 3.075, "step": 44900 }, { "epoch": 1.33, "learning_rate": 3.2788460425325014e-05, "loss": 3.1158, "step": 45000 }, { "epoch": 1.33, "learning_rate": 3.2730577326032345e-05, "loss": 3.0903, "step": 45100 }, { "epoch": 1.33, "learning_rate": 3.267269422673968e-05, "loss": 3.0556, "step": 45200 }, { "epoch": 1.34, "learning_rate": 3.261481112744701e-05, "loss": 3.1016, "step": 45300 }, { "epoch": 1.34, "learning_rate": 3.255692802815434e-05, "loss": 3.1105, "step": 45400 }, { "epoch": 1.34, "learning_rate": 3.249904492886167e-05, "loss": 3.1088, "step": 45500 }, { "epoch": 1.35, "learning_rate": 3.244116182956901e-05, "loss": 3.1597, "step": 45600 }, { "epoch": 1.35, "learning_rate": 3.238327873027633e-05, "loss": 3.0724, "step": 45700 }, { "epoch": 1.35, "learning_rate": 3.232539563098367e-05, "loss": 3.0631, "step": 45800 }, { "epoch": 1.35, "learning_rate": 3.2267512531691e-05, "loss": 3.0182, "step": 45900 }, { "epoch": 1.36, "learning_rate": 3.220962943239833e-05, "loss": 3.0879, "step": 46000 }, { "epoch": 1.36, "learning_rate": 3.215174633310566e-05, "loss": 3.1382, "step": 46100 }, { "epoch": 1.36, "learning_rate": 3.2093863233812996e-05, "loss": 3.1101, "step": 46200 }, { "epoch": 1.37, "learning_rate": 3.2035980134520326e-05, "loss": 3.0471, "step": 46300 }, { "epoch": 1.37, "learning_rate": 3.1978097035227656e-05, "loss": 3.0564, "step": 46400 }, { "epoch": 1.37, "learning_rate": 3.1920213935934986e-05, "loss": 3.1499, "step": 46500 }, { "epoch": 1.38, "learning_rate": 3.1862330836642316e-05, "loss": 3.092, "step": 46600 }, { "epoch": 1.38, "learning_rate": 3.180444773734965e-05, "loss": 3.0732, "step": 46700 }, { "epoch": 1.38, "learning_rate": 3.1746564638056983e-05, "loss": 3.0483, "step": 46800 }, { "epoch": 1.38, "learning_rate": 3.1688681538764314e-05, "loss": 3.0419, "step": 46900 }, { "epoch": 1.39, "learning_rate": 3.1630798439471644e-05, "loss": 3.0499, "step": 47000 }, { "epoch": 1.39, "learning_rate": 3.1572915340178974e-05, "loss": 3.1122, "step": 47100 }, { "epoch": 1.39, "learning_rate": 3.1515032240886304e-05, "loss": 3.1401, "step": 47200 }, { "epoch": 1.4, "learning_rate": 3.145714914159364e-05, "loss": 3.072, "step": 47300 }, { "epoch": 1.4, "learning_rate": 3.139926604230097e-05, "loss": 3.0722, "step": 47400 }, { "epoch": 1.4, "learning_rate": 3.13413829430083e-05, "loss": 3.0141, "step": 47500 }, { "epoch": 1.41, "learning_rate": 3.128349984371564e-05, "loss": 3.0253, "step": 47600 }, { "epoch": 1.41, "learning_rate": 3.122561674442296e-05, "loss": 3.037, "step": 47700 }, { "epoch": 1.41, "learning_rate": 3.11677336451303e-05, "loss": 3.0667, "step": 47800 }, { "epoch": 1.41, "learning_rate": 3.110985054583763e-05, "loss": 3.0826, "step": 47900 }, { "epoch": 1.42, "learning_rate": 3.105196744654496e-05, "loss": 3.0285, "step": 48000 }, { "epoch": 1.42, "learning_rate": 3.099408434725229e-05, "loss": 3.1176, "step": 48100 }, { "epoch": 1.42, "learning_rate": 3.0936201247959625e-05, "loss": 3.0716, "step": 48200 }, { "epoch": 1.43, "learning_rate": 3.0878896979659885e-05, "loss": 2.9875, "step": 48300 }, { "epoch": 1.43, "learning_rate": 3.082101388036721e-05, "loss": 3.0571, "step": 48400 }, { "epoch": 1.43, "learning_rate": 3.0763130781074545e-05, "loss": 3.0559, "step": 48500 }, { "epoch": 1.43, "learning_rate": 3.0705247681781875e-05, "loss": 3.072, "step": 48600 }, { "epoch": 1.44, "learning_rate": 3.0647364582489205e-05, "loss": 3.0742, "step": 48700 }, { "epoch": 1.44, "learning_rate": 3.0589481483196535e-05, "loss": 3.0804, "step": 48800 }, { "epoch": 1.44, "learning_rate": 3.053159838390387e-05, "loss": 3.1331, "step": 48900 }, { "epoch": 1.45, "learning_rate": 3.04737152846112e-05, "loss": 3.0802, "step": 49000 }, { "epoch": 1.45, "learning_rate": 3.0415832185318532e-05, "loss": 3.0288, "step": 49100 }, { "epoch": 1.45, "learning_rate": 3.0357949086025862e-05, "loss": 3.0775, "step": 49200 }, { "epoch": 1.46, "learning_rate": 3.0300065986733196e-05, "loss": 3.0955, "step": 49300 }, { "epoch": 1.46, "learning_rate": 3.024218288744053e-05, "loss": 3.1154, "step": 49400 }, { "epoch": 1.46, "learning_rate": 3.0184299788147856e-05, "loss": 3.0906, "step": 49500 }, { "epoch": 1.46, "learning_rate": 3.0126416688855193e-05, "loss": 3.0599, "step": 49600 }, { "epoch": 1.47, "learning_rate": 3.006853358956252e-05, "loss": 3.0214, "step": 49700 }, { "epoch": 1.47, "learning_rate": 3.0010650490269853e-05, "loss": 3.0938, "step": 49800 }, { "epoch": 1.47, "learning_rate": 2.9953346221970113e-05, "loss": 3.0698, "step": 49900 }, { "epoch": 1.48, "learning_rate": 2.9895463122677443e-05, "loss": 3.1009, "step": 50000 }, { "epoch": 1.48, "learning_rate": 2.9837580023384776e-05, "loss": 3.0987, "step": 50100 }, { "epoch": 1.48, "learning_rate": 2.9779696924092103e-05, "loss": 3.0924, "step": 50200 }, { "epoch": 1.48, "learning_rate": 2.972181382479944e-05, "loss": 3.0633, "step": 50300 }, { "epoch": 1.49, "learning_rate": 2.9663930725506767e-05, "loss": 3.0363, "step": 50400 }, { "epoch": 1.49, "learning_rate": 2.96060476262141e-05, "loss": 3.0711, "step": 50500 }, { "epoch": 1.49, "learning_rate": 2.954816452692143e-05, "loss": 3.1192, "step": 50600 }, { "epoch": 1.5, "learning_rate": 2.9490281427628764e-05, "loss": 3.0765, "step": 50700 }, { "epoch": 1.5, "learning_rate": 2.943239832833609e-05, "loss": 3.1171, "step": 50800 }, { "epoch": 1.5, "learning_rate": 2.9374515229043427e-05, "loss": 3.1118, "step": 50900 }, { "epoch": 1.51, "learning_rate": 2.9316632129750754e-05, "loss": 2.9891, "step": 51000 }, { "epoch": 1.51, "learning_rate": 2.9258749030458087e-05, "loss": 3.1239, "step": 51100 }, { "epoch": 1.51, "learning_rate": 2.920086593116542e-05, "loss": 3.0719, "step": 51200 }, { "epoch": 1.51, "learning_rate": 2.914298283187275e-05, "loss": 3.0945, "step": 51300 }, { "epoch": 1.52, "learning_rate": 2.9085099732580084e-05, "loss": 3.1234, "step": 51400 }, { "epoch": 1.52, "learning_rate": 2.9027216633287415e-05, "loss": 3.1663, "step": 51500 }, { "epoch": 1.52, "learning_rate": 2.8969333533994748e-05, "loss": 3.0325, "step": 51600 }, { "epoch": 1.53, "learning_rate": 2.8911450434702075e-05, "loss": 3.0971, "step": 51700 }, { "epoch": 1.53, "learning_rate": 2.8853567335409408e-05, "loss": 3.0374, "step": 51800 }, { "epoch": 1.53, "learning_rate": 2.879568423611674e-05, "loss": 3.0485, "step": 51900 }, { "epoch": 1.54, "learning_rate": 2.8737801136824072e-05, "loss": 3.0894, "step": 52000 }, { "epoch": 1.54, "learning_rate": 2.8679918037531402e-05, "loss": 3.1166, "step": 52100 }, { "epoch": 1.54, "learning_rate": 2.8622034938238735e-05, "loss": 3.0091, "step": 52200 }, { "epoch": 1.54, "learning_rate": 2.856415183894607e-05, "loss": 3.0695, "step": 52300 }, { "epoch": 1.55, "learning_rate": 2.8506268739653396e-05, "loss": 3.0978, "step": 52400 }, { "epoch": 1.55, "learning_rate": 2.844838564036073e-05, "loss": 3.0559, "step": 52500 }, { "epoch": 1.55, "learning_rate": 2.839050254106806e-05, "loss": 3.0492, "step": 52600 }, { "epoch": 1.56, "learning_rate": 2.8332619441775393e-05, "loss": 3.0304, "step": 52700 }, { "epoch": 1.56, "learning_rate": 2.8274736342482723e-05, "loss": 3.1342, "step": 52800 }, { "epoch": 1.56, "learning_rate": 2.8216853243190056e-05, "loss": 3.076, "step": 52900 }, { "epoch": 1.56, "learning_rate": 2.8158970143897383e-05, "loss": 2.9751, "step": 53000 }, { "epoch": 1.57, "learning_rate": 2.8101087044604717e-05, "loss": 3.0754, "step": 53100 }, { "epoch": 1.57, "learning_rate": 2.8043203945312047e-05, "loss": 3.0723, "step": 53200 }, { "epoch": 1.57, "learning_rate": 2.798532084601938e-05, "loss": 3.0367, "step": 53300 }, { "epoch": 1.58, "learning_rate": 2.7927437746726714e-05, "loss": 3.0745, "step": 53400 }, { "epoch": 1.58, "learning_rate": 2.7869554647434044e-05, "loss": 3.0561, "step": 53500 }, { "epoch": 1.58, "learning_rate": 2.7811671548141377e-05, "loss": 3.1498, "step": 53600 }, { "epoch": 1.59, "learning_rate": 2.7753788448848704e-05, "loss": 3.0278, "step": 53700 }, { "epoch": 1.59, "learning_rate": 2.769590534955604e-05, "loss": 3.0985, "step": 53800 }, { "epoch": 1.59, "learning_rate": 2.7638022250263367e-05, "loss": 3.0816, "step": 53900 }, { "epoch": 1.59, "learning_rate": 2.75801391509707e-05, "loss": 3.0399, "step": 54000 }, { "epoch": 1.6, "learning_rate": 2.752283488267096e-05, "loss": 3.0389, "step": 54100 }, { "epoch": 1.6, "learning_rate": 2.746495178337829e-05, "loss": 3.0626, "step": 54200 }, { "epoch": 1.6, "learning_rate": 2.7407068684085624e-05, "loss": 3.0901, "step": 54300 }, { "epoch": 1.61, "learning_rate": 2.734918558479295e-05, "loss": 3.0882, "step": 54400 }, { "epoch": 1.61, "learning_rate": 2.7291302485500288e-05, "loss": 3.1266, "step": 54500 }, { "epoch": 1.61, "learning_rate": 2.7233419386207614e-05, "loss": 3.0028, "step": 54600 }, { "epoch": 1.61, "learning_rate": 2.7175536286914948e-05, "loss": 3.0538, "step": 54700 }, { "epoch": 1.62, "learning_rate": 2.7117653187622278e-05, "loss": 3.0825, "step": 54800 }, { "epoch": 1.62, "learning_rate": 2.705977008832961e-05, "loss": 3.0424, "step": 54900 }, { "epoch": 1.62, "learning_rate": 2.7001886989036938e-05, "loss": 3.1286, "step": 55000 }, { "epoch": 1.63, "learning_rate": 2.6944003889744275e-05, "loss": 3.1161, "step": 55100 }, { "epoch": 1.63, "learning_rate": 2.688612079045161e-05, "loss": 3.0712, "step": 55200 }, { "epoch": 1.63, "learning_rate": 2.6828237691158935e-05, "loss": 3.0373, "step": 55300 }, { "epoch": 1.64, "learning_rate": 2.677035459186627e-05, "loss": 3.0883, "step": 55400 }, { "epoch": 1.64, "learning_rate": 2.67124714925736e-05, "loss": 3.041, "step": 55500 }, { "epoch": 1.64, "learning_rate": 2.6654588393280932e-05, "loss": 3.0623, "step": 55600 }, { "epoch": 1.64, "learning_rate": 2.6596705293988262e-05, "loss": 3.1202, "step": 55700 }, { "epoch": 1.65, "learning_rate": 2.6538822194695596e-05, "loss": 3.0211, "step": 55800 }, { "epoch": 1.65, "learning_rate": 2.6480939095402923e-05, "loss": 3.1307, "step": 55900 }, { "epoch": 1.65, "learning_rate": 2.6423055996110256e-05, "loss": 3.0234, "step": 56000 }, { "epoch": 1.66, "learning_rate": 2.6365172896817586e-05, "loss": 3.013, "step": 56100 }, { "epoch": 1.66, "learning_rate": 2.6307868628517846e-05, "loss": 3.0643, "step": 56200 }, { "epoch": 1.66, "learning_rate": 2.624998552922518e-05, "loss": 3.046, "step": 56300 }, { "epoch": 1.66, "learning_rate": 2.619210242993251e-05, "loss": 2.9987, "step": 56400 }, { "epoch": 1.67, "learning_rate": 2.6134219330639843e-05, "loss": 3.1, "step": 56500 }, { "epoch": 1.67, "learning_rate": 2.6076915062340102e-05, "loss": 3.0732, "step": 56600 }, { "epoch": 1.67, "learning_rate": 2.601903196304743e-05, "loss": 2.996, "step": 56700 }, { "epoch": 1.68, "learning_rate": 2.5961148863754763e-05, "loss": 2.9735, "step": 56800 }, { "epoch": 1.68, "learning_rate": 2.5903265764462093e-05, "loss": 3.1598, "step": 56900 }, { "epoch": 1.68, "learning_rate": 2.5845382665169426e-05, "loss": 3.0743, "step": 57000 }, { "epoch": 1.69, "learning_rate": 2.5787499565876756e-05, "loss": 3.0798, "step": 57100 }, { "epoch": 1.69, "learning_rate": 2.572961646658409e-05, "loss": 3.0895, "step": 57200 }, { "epoch": 1.69, "learning_rate": 2.5671733367291416e-05, "loss": 3.051, "step": 57300 }, { "epoch": 1.69, "learning_rate": 2.561385026799875e-05, "loss": 3.0544, "step": 57400 }, { "epoch": 1.7, "learning_rate": 2.555596716870608e-05, "loss": 3.0848, "step": 57500 }, { "epoch": 1.7, "learning_rate": 2.5498084069413413e-05, "loss": 3.0787, "step": 57600 }, { "epoch": 1.7, "learning_rate": 2.5440779801113673e-05, "loss": 3.1021, "step": 57700 }, { "epoch": 1.71, "learning_rate": 2.5382896701821003e-05, "loss": 3.0226, "step": 57800 }, { "epoch": 1.71, "learning_rate": 2.5325013602528337e-05, "loss": 3.0637, "step": 57900 }, { "epoch": 1.71, "learning_rate": 2.5267130503235663e-05, "loss": 3.1298, "step": 58000 }, { "epoch": 1.72, "learning_rate": 2.5209247403942997e-05, "loss": 3.0767, "step": 58100 }, { "epoch": 1.72, "learning_rate": 2.5151364304650327e-05, "loss": 2.9812, "step": 58200 }, { "epoch": 1.72, "learning_rate": 2.509348120535766e-05, "loss": 3.0861, "step": 58300 }, { "epoch": 1.72, "learning_rate": 2.5035598106064994e-05, "loss": 3.1261, "step": 58400 }, { "epoch": 1.73, "learning_rate": 2.4977715006772324e-05, "loss": 3.0393, "step": 58500 }, { "epoch": 1.73, "learning_rate": 2.4919831907479654e-05, "loss": 3.0622, "step": 58600 }, { "epoch": 1.73, "learning_rate": 2.4861948808186984e-05, "loss": 3.0922, "step": 58700 }, { "epoch": 1.74, "learning_rate": 2.480406570889432e-05, "loss": 3.0503, "step": 58800 }, { "epoch": 1.74, "learning_rate": 2.474618260960165e-05, "loss": 3.0453, "step": 58900 }, { "epoch": 1.74, "learning_rate": 2.468829951030898e-05, "loss": 3.0043, "step": 59000 }, { "epoch": 1.74, "learning_rate": 2.4630416411016315e-05, "loss": 3.0512, "step": 59100 }, { "epoch": 1.75, "learning_rate": 2.4572533311723645e-05, "loss": 3.0278, "step": 59200 }, { "epoch": 1.75, "learning_rate": 2.4514650212430975e-05, "loss": 3.0435, "step": 59300 }, { "epoch": 1.75, "learning_rate": 2.445676711313831e-05, "loss": 3.0182, "step": 59400 }, { "epoch": 1.76, "learning_rate": 2.439888401384564e-05, "loss": 3.013, "step": 59500 }, { "epoch": 1.76, "learning_rate": 2.434100091455297e-05, "loss": 3.1001, "step": 59600 }, { "epoch": 1.76, "learning_rate": 2.4283117815260302e-05, "loss": 3.0871, "step": 59700 }, { "epoch": 1.77, "learning_rate": 2.4225234715967632e-05, "loss": 3.0644, "step": 59800 }, { "epoch": 1.77, "learning_rate": 2.4167351616674962e-05, "loss": 3.0736, "step": 59900 }, { "epoch": 1.77, "learning_rate": 2.4109468517382296e-05, "loss": 3.0558, "step": 60000 }, { "epoch": 1.77, "learning_rate": 2.405158541808963e-05, "loss": 3.0976, "step": 60100 }, { "epoch": 1.78, "learning_rate": 2.399370231879696e-05, "loss": 3.0441, "step": 60200 }, { "epoch": 1.78, "learning_rate": 2.393581921950429e-05, "loss": 3.0623, "step": 60300 }, { "epoch": 1.78, "learning_rate": 2.3877936120211623e-05, "loss": 3.0432, "step": 60400 }, { "epoch": 1.79, "learning_rate": 2.3820053020918953e-05, "loss": 3.031, "step": 60500 }, { "epoch": 1.79, "learning_rate": 2.3762169921626283e-05, "loss": 3.0308, "step": 60600 }, { "epoch": 1.79, "learning_rate": 2.3704286822333617e-05, "loss": 3.0621, "step": 60700 }, { "epoch": 1.79, "learning_rate": 2.3646403723040947e-05, "loss": 2.9949, "step": 60800 }, { "epoch": 1.8, "learning_rate": 2.3588520623748277e-05, "loss": 3.0625, "step": 60900 }, { "epoch": 1.8, "learning_rate": 2.353063752445561e-05, "loss": 3.0175, "step": 61000 }, { "epoch": 1.8, "learning_rate": 2.3472754425162944e-05, "loss": 2.9987, "step": 61100 }, { "epoch": 1.81, "learning_rate": 2.3414871325870274e-05, "loss": 2.998, "step": 61200 }, { "epoch": 1.81, "learning_rate": 2.3356988226577607e-05, "loss": 3.0248, "step": 61300 }, { "epoch": 1.81, "learning_rate": 2.3299105127284938e-05, "loss": 3.0506, "step": 61400 }, { "epoch": 1.82, "learning_rate": 2.3241222027992268e-05, "loss": 3.1104, "step": 61500 }, { "epoch": 1.82, "learning_rate": 2.31833389286996e-05, "loss": 3.0363, "step": 61600 }, { "epoch": 1.82, "learning_rate": 2.312545582940693e-05, "loss": 3.0894, "step": 61700 }, { "epoch": 1.82, "learning_rate": 2.306757273011426e-05, "loss": 3.0058, "step": 61800 }, { "epoch": 1.83, "learning_rate": 2.300968963082159e-05, "loss": 2.9939, "step": 61900 }, { "epoch": 1.83, "learning_rate": 2.2951806531528925e-05, "loss": 2.9725, "step": 62000 }, { "epoch": 1.83, "learning_rate": 2.2893923432236255e-05, "loss": 3.0223, "step": 62100 }, { "epoch": 1.84, "learning_rate": 2.283604033294359e-05, "loss": 3.0144, "step": 62200 }, { "epoch": 1.84, "learning_rate": 2.2778157233650922e-05, "loss": 3.0021, "step": 62300 }, { "epoch": 1.84, "learning_rate": 2.2720274134358252e-05, "loss": 2.9874, "step": 62400 }, { "epoch": 1.85, "learning_rate": 2.2662391035065582e-05, "loss": 3.0061, "step": 62500 }, { "epoch": 1.85, "learning_rate": 2.2604507935772916e-05, "loss": 2.9589, "step": 62600 }, { "epoch": 1.85, "learning_rate": 2.2546624836480246e-05, "loss": 3.0054, "step": 62700 }, { "epoch": 1.85, "learning_rate": 2.2488741737187576e-05, "loss": 3.0209, "step": 62800 }, { "epoch": 1.86, "learning_rate": 2.243085863789491e-05, "loss": 3.0615, "step": 62900 }, { "epoch": 1.86, "learning_rate": 2.237297553860224e-05, "loss": 2.9817, "step": 63000 }, { "epoch": 1.86, "learning_rate": 2.231509243930957e-05, "loss": 3.0172, "step": 63100 }, { "epoch": 1.87, "learning_rate": 2.2257209340016903e-05, "loss": 3.0459, "step": 63200 }, { "epoch": 1.87, "learning_rate": 2.2199326240724233e-05, "loss": 3.0939, "step": 63300 }, { "epoch": 1.87, "learning_rate": 2.2141443141431567e-05, "loss": 3.0101, "step": 63400 }, { "epoch": 1.87, "learning_rate": 2.2083560042138897e-05, "loss": 3.0242, "step": 63500 }, { "epoch": 1.88, "learning_rate": 2.202567694284623e-05, "loss": 3.0988, "step": 63600 }, { "epoch": 1.88, "learning_rate": 2.196779384355356e-05, "loss": 3.089, "step": 63700 }, { "epoch": 1.88, "learning_rate": 2.190991074426089e-05, "loss": 3.1095, "step": 63800 }, { "epoch": 1.89, "learning_rate": 2.185260647596115e-05, "loss": 3.0332, "step": 63900 }, { "epoch": 1.89, "learning_rate": 2.1794723376668483e-05, "loss": 3.0994, "step": 64000 }, { "epoch": 1.89, "learning_rate": 2.1736840277375814e-05, "loss": 3.0328, "step": 64100 }, { "epoch": 1.9, "learning_rate": 2.1678957178083144e-05, "loss": 3.0795, "step": 64200 }, { "epoch": 1.9, "learning_rate": 2.1621074078790477e-05, "loss": 3.0247, "step": 64300 }, { "epoch": 1.9, "learning_rate": 2.1563190979497807e-05, "loss": 3.0352, "step": 64400 }, { "epoch": 1.9, "learning_rate": 2.1505307880205137e-05, "loss": 3.0503, "step": 64500 }, { "epoch": 1.91, "learning_rate": 2.144742478091247e-05, "loss": 3.0047, "step": 64600 }, { "epoch": 1.91, "learning_rate": 2.13895416816198e-05, "loss": 3.0388, "step": 64700 }, { "epoch": 1.91, "learning_rate": 2.133165858232713e-05, "loss": 3.0377, "step": 64800 }, { "epoch": 1.92, "learning_rate": 2.1273775483034465e-05, "loss": 3.0398, "step": 64900 }, { "epoch": 1.92, "learning_rate": 2.1215892383741795e-05, "loss": 3.0546, "step": 65000 }, { "epoch": 1.92, "learning_rate": 2.1158009284449125e-05, "loss": 3.019, "step": 65100 }, { "epoch": 1.92, "learning_rate": 2.110012618515646e-05, "loss": 3.0192, "step": 65200 }, { "epoch": 1.93, "learning_rate": 2.1042243085863792e-05, "loss": 2.9797, "step": 65300 }, { "epoch": 1.93, "learning_rate": 2.0984359986571122e-05, "loss": 3.1248, "step": 65400 }, { "epoch": 1.93, "learning_rate": 2.0926476887278455e-05, "loss": 3.0729, "step": 65500 }, { "epoch": 1.94, "learning_rate": 2.0868593787985785e-05, "loss": 3.0108, "step": 65600 }, { "epoch": 1.94, "learning_rate": 2.0810710688693116e-05, "loss": 2.9739, "step": 65700 }, { "epoch": 1.94, "learning_rate": 2.075282758940045e-05, "loss": 3.058, "step": 65800 }, { "epoch": 1.95, "learning_rate": 2.069494449010778e-05, "loss": 3.0111, "step": 65900 }, { "epoch": 1.95, "learning_rate": 2.063706139081511e-05, "loss": 3.075, "step": 66000 }, { "epoch": 1.95, "learning_rate": 2.0579178291522443e-05, "loss": 3.0493, "step": 66100 }, { "epoch": 1.95, "learning_rate": 2.0521295192229773e-05, "loss": 2.9718, "step": 66200 }, { "epoch": 1.96, "learning_rate": 2.0463412092937106e-05, "loss": 3.0471, "step": 66300 }, { "epoch": 1.96, "learning_rate": 2.0405528993644436e-05, "loss": 3.0502, "step": 66400 }, { "epoch": 1.96, "learning_rate": 2.0348803556337622e-05, "loss": 3.0482, "step": 66500 }, { "epoch": 1.97, "learning_rate": 2.0290920457044955e-05, "loss": 3.0391, "step": 66600 }, { "epoch": 1.97, "learning_rate": 2.0233037357752285e-05, "loss": 2.9687, "step": 66700 }, { "epoch": 1.97, "learning_rate": 2.0175154258459616e-05, "loss": 3.0298, "step": 66800 }, { "epoch": 1.97, "learning_rate": 2.011727115916695e-05, "loss": 3.0962, "step": 66900 }, { "epoch": 1.98, "learning_rate": 2.005938805987428e-05, "loss": 3.0824, "step": 67000 }, { "epoch": 1.98, "learning_rate": 2.000150496058161e-05, "loss": 3.0595, "step": 67100 }, { "epoch": 1.98, "learning_rate": 1.9943621861288943e-05, "loss": 2.9599, "step": 67200 }, { "epoch": 1.99, "learning_rate": 1.9885738761996273e-05, "loss": 2.9756, "step": 67300 }, { "epoch": 1.99, "learning_rate": 1.9827855662703603e-05, "loss": 2.9989, "step": 67400 }, { "epoch": 1.99, "learning_rate": 1.9769972563410936e-05, "loss": 3.0489, "step": 67500 }, { "epoch": 2.0, "learning_rate": 1.971208946411827e-05, "loss": 3.0045, "step": 67600 }, { "epoch": 2.0, "learning_rate": 1.96542063648256e-05, "loss": 2.9764, "step": 67700 }, { "epoch": 2.0, "eval_gen_len": 19.9174, "eval_loss": 3.12784481048584, "eval_rouge1": 18.6558, "eval_rouge2": 5.1844, "eval_rougeL": 15.0939, "eval_rougeLsum": 16.3367, "eval_runtime": 747.2706, "eval_samples_per_second": 10.073, "eval_steps_per_second": 2.518, "step": 67750 }, { "epoch": 2.0, "learning_rate": 1.959632326553293e-05, "loss": 2.7806, "step": 67800 }, { "epoch": 2.0, "learning_rate": 1.9538440166240264e-05, "loss": 2.6776, "step": 67900 }, { "epoch": 2.01, "learning_rate": 1.9480557066947594e-05, "loss": 2.6714, "step": 68000 }, { "epoch": 2.01, "learning_rate": 1.9422673967654924e-05, "loss": 2.6082, "step": 68100 }, { "epoch": 2.01, "learning_rate": 1.9364790868362257e-05, "loss": 2.6635, "step": 68200 }, { "epoch": 2.02, "learning_rate": 1.9306907769069587e-05, "loss": 2.6254, "step": 68300 }, { "epoch": 2.02, "learning_rate": 1.9249024669776918e-05, "loss": 2.6649, "step": 68400 }, { "epoch": 2.02, "learning_rate": 1.919114157048425e-05, "loss": 2.6291, "step": 68500 }, { "epoch": 2.03, "learning_rate": 1.913325847119158e-05, "loss": 2.6979, "step": 68600 }, { "epoch": 2.03, "learning_rate": 1.907537537189891e-05, "loss": 2.6635, "step": 68700 }, { "epoch": 2.03, "learning_rate": 1.9017492272606248e-05, "loss": 2.6288, "step": 68800 }, { "epoch": 2.03, "learning_rate": 1.8959609173313578e-05, "loss": 2.6401, "step": 68900 }, { "epoch": 2.04, "learning_rate": 1.890172607402091e-05, "loss": 2.6152, "step": 69000 }, { "epoch": 2.04, "learning_rate": 1.8843842974728242e-05, "loss": 2.6912, "step": 69100 }, { "epoch": 2.04, "learning_rate": 1.8785959875435572e-05, "loss": 2.6413, "step": 69200 }, { "epoch": 2.05, "learning_rate": 1.8728076776142902e-05, "loss": 2.6168, "step": 69300 }, { "epoch": 2.05, "learning_rate": 1.8670193676850235e-05, "loss": 2.661, "step": 69400 }, { "epoch": 2.05, "learning_rate": 1.8612310577557566e-05, "loss": 2.6739, "step": 69500 }, { "epoch": 2.05, "learning_rate": 1.8554427478264896e-05, "loss": 2.6745, "step": 69600 }, { "epoch": 2.06, "learning_rate": 1.849654437897223e-05, "loss": 2.7476, "step": 69700 }, { "epoch": 2.06, "learning_rate": 1.843924011067249e-05, "loss": 2.6416, "step": 69800 }, { "epoch": 2.06, "learning_rate": 1.838135701137982e-05, "loss": 2.6431, "step": 69900 }, { "epoch": 2.07, "learning_rate": 1.832347391208715e-05, "loss": 2.6663, "step": 70000 }, { "epoch": 2.07, "learning_rate": 1.8265590812794482e-05, "loss": 2.5952, "step": 70100 }, { "epoch": 2.07, "learning_rate": 1.8207707713501812e-05, "loss": 2.6954, "step": 70200 }, { "epoch": 2.08, "learning_rate": 1.8149824614209143e-05, "loss": 2.6513, "step": 70300 }, { "epoch": 2.08, "learning_rate": 1.8091941514916476e-05, "loss": 2.6924, "step": 70400 }, { "epoch": 2.08, "learning_rate": 1.8034058415623806e-05, "loss": 2.7011, "step": 70500 }, { "epoch": 2.08, "learning_rate": 1.797617531633114e-05, "loss": 2.734, "step": 70600 }, { "epoch": 2.09, "learning_rate": 1.791829221703847e-05, "loss": 2.6728, "step": 70700 }, { "epoch": 2.09, "learning_rate": 1.7860409117745803e-05, "loss": 2.6891, "step": 70800 }, { "epoch": 2.09, "learning_rate": 1.7802526018453133e-05, "loss": 2.6044, "step": 70900 }, { "epoch": 2.1, "learning_rate": 1.7744642919160463e-05, "loss": 2.7256, "step": 71000 }, { "epoch": 2.1, "learning_rate": 1.7686759819867797e-05, "loss": 2.5696, "step": 71100 }, { "epoch": 2.1, "learning_rate": 1.7628876720575127e-05, "loss": 2.6064, "step": 71200 }, { "epoch": 2.1, "learning_rate": 1.7570993621282457e-05, "loss": 2.6711, "step": 71300 }, { "epoch": 2.11, "learning_rate": 1.751311052198979e-05, "loss": 2.6586, "step": 71400 }, { "epoch": 2.11, "learning_rate": 1.745522742269712e-05, "loss": 2.6367, "step": 71500 }, { "epoch": 2.11, "learning_rate": 1.739734432340445e-05, "loss": 2.5933, "step": 71600 }, { "epoch": 2.12, "learning_rate": 1.7339461224111784e-05, "loss": 2.7044, "step": 71700 }, { "epoch": 2.12, "learning_rate": 1.7281578124819118e-05, "loss": 2.6311, "step": 71800 }, { "epoch": 2.12, "learning_rate": 1.7223695025526448e-05, "loss": 2.6212, "step": 71900 }, { "epoch": 2.13, "learning_rate": 1.7165811926233778e-05, "loss": 2.6314, "step": 72000 }, { "epoch": 2.13, "learning_rate": 1.710792882694111e-05, "loss": 2.7652, "step": 72100 }, { "epoch": 2.13, "learning_rate": 1.705004572764844e-05, "loss": 2.6811, "step": 72200 }, { "epoch": 2.13, "learning_rate": 1.6992741459348698e-05, "loss": 2.6055, "step": 72300 }, { "epoch": 2.14, "learning_rate": 1.693485836005603e-05, "loss": 2.6991, "step": 72400 }, { "epoch": 2.14, "learning_rate": 1.6876975260763365e-05, "loss": 2.6355, "step": 72500 }, { "epoch": 2.14, "learning_rate": 1.6819092161470695e-05, "loss": 2.6464, "step": 72600 }, { "epoch": 2.15, "learning_rate": 1.6761209062178025e-05, "loss": 2.6495, "step": 72700 }, { "epoch": 2.15, "learning_rate": 1.670332596288536e-05, "loss": 2.6867, "step": 72800 }, { "epoch": 2.15, "learning_rate": 1.664544286359269e-05, "loss": 2.6519, "step": 72900 }, { "epoch": 2.15, "learning_rate": 1.658755976430002e-05, "loss": 2.6794, "step": 73000 }, { "epoch": 2.16, "learning_rate": 1.6529676665007352e-05, "loss": 2.6281, "step": 73100 }, { "epoch": 2.16, "learning_rate": 1.6471793565714682e-05, "loss": 2.6679, "step": 73200 }, { "epoch": 2.16, "learning_rate": 1.6413910466422012e-05, "loss": 2.6788, "step": 73300 }, { "epoch": 2.17, "learning_rate": 1.6356027367129346e-05, "loss": 2.6201, "step": 73400 }, { "epoch": 2.17, "learning_rate": 1.629814426783668e-05, "loss": 2.6763, "step": 73500 }, { "epoch": 2.17, "learning_rate": 1.624026116854401e-05, "loss": 2.6572, "step": 73600 }, { "epoch": 2.18, "learning_rate": 1.6182378069251343e-05, "loss": 2.6733, "step": 73700 }, { "epoch": 2.18, "learning_rate": 1.6124494969958673e-05, "loss": 2.6219, "step": 73800 }, { "epoch": 2.18, "learning_rate": 1.6066611870666003e-05, "loss": 2.5916, "step": 73900 }, { "epoch": 2.18, "learning_rate": 1.6008728771373337e-05, "loss": 2.689, "step": 74000 }, { "epoch": 2.19, "learning_rate": 1.5950845672080667e-05, "loss": 2.6972, "step": 74100 }, { "epoch": 2.19, "learning_rate": 1.5892962572787997e-05, "loss": 2.6629, "step": 74200 }, { "epoch": 2.19, "learning_rate": 1.583507947349533e-05, "loss": 2.6508, "step": 74300 }, { "epoch": 2.2, "learning_rate": 1.577719637420266e-05, "loss": 2.7088, "step": 74400 }, { "epoch": 2.2, "learning_rate": 1.571931327490999e-05, "loss": 2.644, "step": 74500 }, { "epoch": 2.2, "learning_rate": 1.5661430175617324e-05, "loss": 2.6248, "step": 74600 }, { "epoch": 2.21, "learning_rate": 1.5603547076324657e-05, "loss": 2.5826, "step": 74700 }, { "epoch": 2.21, "learning_rate": 1.5545663977031988e-05, "loss": 2.6165, "step": 74800 }, { "epoch": 2.21, "learning_rate": 1.5487780877739318e-05, "loss": 2.6644, "step": 74900 }, { "epoch": 2.21, "learning_rate": 1.542989777844665e-05, "loss": 2.7307, "step": 75000 }, { "epoch": 2.22, "learning_rate": 1.537201467915398e-05, "loss": 2.664, "step": 75100 }, { "epoch": 2.22, "learning_rate": 1.531413157986131e-05, "loss": 2.6484, "step": 75200 }, { "epoch": 2.22, "learning_rate": 1.5256248480568645e-05, "loss": 2.6445, "step": 75300 }, { "epoch": 2.23, "learning_rate": 1.5198365381275975e-05, "loss": 2.6681, "step": 75400 }, { "epoch": 2.23, "learning_rate": 1.5140482281983307e-05, "loss": 2.6783, "step": 75500 }, { "epoch": 2.23, "learning_rate": 1.5082599182690638e-05, "loss": 2.6748, "step": 75600 }, { "epoch": 2.23, "learning_rate": 1.5024716083397969e-05, "loss": 2.6884, "step": 75700 }, { "epoch": 2.24, "learning_rate": 1.4966832984105302e-05, "loss": 2.6373, "step": 75800 }, { "epoch": 2.24, "learning_rate": 1.4908949884812634e-05, "loss": 2.6803, "step": 75900 }, { "epoch": 2.24, "learning_rate": 1.4851066785519966e-05, "loss": 2.6763, "step": 76000 }, { "epoch": 2.25, "learning_rate": 1.4793183686227296e-05, "loss": 2.6724, "step": 76100 }, { "epoch": 2.25, "learning_rate": 1.4735300586934628e-05, "loss": 2.6505, "step": 76200 }, { "epoch": 2.25, "learning_rate": 1.467741748764196e-05, "loss": 2.6467, "step": 76300 }, { "epoch": 2.26, "learning_rate": 1.4620113219342215e-05, "loss": 2.6609, "step": 76400 }, { "epoch": 2.26, "learning_rate": 1.4562230120049549e-05, "loss": 2.6014, "step": 76500 }, { "epoch": 2.26, "learning_rate": 1.450434702075688e-05, "loss": 2.6619, "step": 76600 }, { "epoch": 2.26, "learning_rate": 1.4446463921464213e-05, "loss": 2.6916, "step": 76700 }, { "epoch": 2.27, "learning_rate": 1.4388580822171543e-05, "loss": 2.7405, "step": 76800 }, { "epoch": 2.27, "learning_rate": 1.4330697722878874e-05, "loss": 2.6346, "step": 76900 }, { "epoch": 2.27, "learning_rate": 1.4272814623586206e-05, "loss": 2.6796, "step": 77000 }, { "epoch": 2.28, "learning_rate": 1.4214931524293536e-05, "loss": 2.6257, "step": 77100 }, { "epoch": 2.28, "learning_rate": 1.4157048425000868e-05, "loss": 2.622, "step": 77200 }, { "epoch": 2.28, "learning_rate": 1.40991653257082e-05, "loss": 2.6802, "step": 77300 }, { "epoch": 2.28, "learning_rate": 1.404128222641553e-05, "loss": 2.6289, "step": 77400 }, { "epoch": 2.29, "learning_rate": 1.3983399127122862e-05, "loss": 2.6751, "step": 77500 }, { "epoch": 2.29, "learning_rate": 1.3925516027830195e-05, "loss": 2.6744, "step": 77600 }, { "epoch": 2.29, "learning_rate": 1.3867632928537527e-05, "loss": 2.6766, "step": 77700 }, { "epoch": 2.3, "learning_rate": 1.3809749829244859e-05, "loss": 2.7145, "step": 77800 }, { "epoch": 2.3, "learning_rate": 1.3751866729952189e-05, "loss": 2.727, "step": 77900 }, { "epoch": 2.3, "learning_rate": 1.369398363065952e-05, "loss": 2.647, "step": 78000 }, { "epoch": 2.31, "learning_rate": 1.3636100531366853e-05, "loss": 2.7302, "step": 78100 }, { "epoch": 2.31, "learning_rate": 1.3578217432074183e-05, "loss": 2.7183, "step": 78200 }, { "epoch": 2.31, "learning_rate": 1.3520334332781515e-05, "loss": 2.5994, "step": 78300 }, { "epoch": 2.31, "learning_rate": 1.3462451233488846e-05, "loss": 2.632, "step": 78400 }, { "epoch": 2.32, "learning_rate": 1.3404568134196176e-05, "loss": 2.6969, "step": 78500 }, { "epoch": 2.32, "learning_rate": 1.3346685034903508e-05, "loss": 2.6856, "step": 78600 }, { "epoch": 2.32, "learning_rate": 1.3289380766603768e-05, "loss": 2.6106, "step": 78700 }, { "epoch": 2.33, "learning_rate": 1.32314976673111e-05, "loss": 2.642, "step": 78800 }, { "epoch": 2.33, "learning_rate": 1.317361456801843e-05, "loss": 2.6484, "step": 78900 }, { "epoch": 2.33, "learning_rate": 1.3115731468725761e-05, "loss": 2.6598, "step": 79000 }, { "epoch": 2.34, "learning_rate": 1.3057848369433093e-05, "loss": 2.5861, "step": 79100 }, { "epoch": 2.34, "learning_rate": 1.2999965270140423e-05, "loss": 2.6432, "step": 79200 }, { "epoch": 2.34, "learning_rate": 1.2942082170847755e-05, "loss": 2.6069, "step": 79300 }, { "epoch": 2.34, "learning_rate": 1.2884199071555089e-05, "loss": 2.7582, "step": 79400 }, { "epoch": 2.35, "learning_rate": 1.282631597226242e-05, "loss": 2.7005, "step": 79500 }, { "epoch": 2.35, "learning_rate": 1.2768432872969752e-05, "loss": 2.5888, "step": 79600 }, { "epoch": 2.35, "learning_rate": 1.2711128604670008e-05, "loss": 2.7044, "step": 79700 }, { "epoch": 2.36, "learning_rate": 1.265324550537734e-05, "loss": 2.6257, "step": 79800 }, { "epoch": 2.36, "learning_rate": 1.259536240608467e-05, "loss": 2.6742, "step": 79900 }, { "epoch": 2.36, "learning_rate": 1.2537479306792002e-05, "loss": 2.675, "step": 80000 }, { "epoch": 2.36, "learning_rate": 1.2479596207499335e-05, "loss": 2.6145, "step": 80100 }, { "epoch": 2.37, "learning_rate": 1.2421713108206666e-05, "loss": 2.68, "step": 80200 }, { "epoch": 2.37, "learning_rate": 1.2363830008913999e-05, "loss": 2.6717, "step": 80300 }, { "epoch": 2.37, "learning_rate": 1.230594690962133e-05, "loss": 2.6563, "step": 80400 }, { "epoch": 2.38, "learning_rate": 1.2248063810328661e-05, "loss": 2.7272, "step": 80500 }, { "epoch": 2.38, "learning_rate": 1.2190180711035993e-05, "loss": 2.6457, "step": 80600 }, { "epoch": 2.38, "learning_rate": 1.2132297611743323e-05, "loss": 2.7066, "step": 80700 }, { "epoch": 2.39, "learning_rate": 1.2074414512450655e-05, "loss": 2.6104, "step": 80800 }, { "epoch": 2.39, "learning_rate": 1.2016531413157988e-05, "loss": 2.6494, "step": 80900 }, { "epoch": 2.39, "learning_rate": 1.1958648313865318e-05, "loss": 2.656, "step": 81000 }, { "epoch": 2.39, "learning_rate": 1.190076521457265e-05, "loss": 2.7509, "step": 81100 }, { "epoch": 2.4, "learning_rate": 1.184288211527998e-05, "loss": 2.6835, "step": 81200 }, { "epoch": 2.4, "learning_rate": 1.1784999015987312e-05, "loss": 2.6426, "step": 81300 }, { "epoch": 2.4, "learning_rate": 1.1727115916694644e-05, "loss": 2.6237, "step": 81400 }, { "epoch": 2.41, "learning_rate": 1.1669232817401976e-05, "loss": 2.6958, "step": 81500 }, { "epoch": 2.41, "learning_rate": 1.1611349718109307e-05, "loss": 2.6218, "step": 81600 }, { "epoch": 2.41, "learning_rate": 1.1553466618816639e-05, "loss": 2.5956, "step": 81700 }, { "epoch": 2.41, "learning_rate": 1.149558351952397e-05, "loss": 2.6071, "step": 81800 }, { "epoch": 2.42, "learning_rate": 1.1437700420231301e-05, "loss": 2.6936, "step": 81900 }, { "epoch": 2.42, "learning_rate": 1.1379817320938633e-05, "loss": 2.7179, "step": 82000 }, { "epoch": 2.42, "learning_rate": 1.1321934221645965e-05, "loss": 2.7001, "step": 82100 }, { "epoch": 2.43, "learning_rate": 1.1264051122353296e-05, "loss": 2.5915, "step": 82200 }, { "epoch": 2.43, "learning_rate": 1.1206168023060626e-05, "loss": 2.6884, "step": 82300 }, { "epoch": 2.43, "learning_rate": 1.1148284923767958e-05, "loss": 2.647, "step": 82400 }, { "epoch": 2.44, "learning_rate": 1.109040182447529e-05, "loss": 2.6499, "step": 82500 }, { "epoch": 2.44, "learning_rate": 1.1032518725182622e-05, "loss": 2.654, "step": 82600 }, { "epoch": 2.44, "learning_rate": 1.0974635625889954e-05, "loss": 2.6469, "step": 82700 }, { "epoch": 2.44, "learning_rate": 1.0916752526597285e-05, "loss": 2.6167, "step": 82800 }, { "epoch": 2.45, "learning_rate": 1.0858869427304616e-05, "loss": 2.6403, "step": 82900 }, { "epoch": 2.45, "learning_rate": 1.0800986328011947e-05, "loss": 2.632, "step": 83000 }, { "epoch": 2.45, "learning_rate": 1.074310322871928e-05, "loss": 2.6953, "step": 83100 }, { "epoch": 2.46, "learning_rate": 1.0685220129426611e-05, "loss": 2.7353, "step": 83200 }, { "epoch": 2.46, "learning_rate": 1.0627337030133943e-05, "loss": 2.6768, "step": 83300 }, { "epoch": 2.46, "learning_rate": 1.0569453930841273e-05, "loss": 2.6685, "step": 83400 }, { "epoch": 2.46, "learning_rate": 1.0511570831548605e-05, "loss": 2.7646, "step": 83500 }, { "epoch": 2.47, "learning_rate": 1.0453687732255936e-05, "loss": 2.5976, "step": 83600 }, { "epoch": 2.47, "learning_rate": 1.0395804632963268e-05, "loss": 2.6466, "step": 83700 }, { "epoch": 2.47, "learning_rate": 1.03379215336706e-05, "loss": 2.7244, "step": 83800 }, { "epoch": 2.48, "learning_rate": 1.0280038434377932e-05, "loss": 2.6485, "step": 83900 }, { "epoch": 2.48, "learning_rate": 1.0222155335085262e-05, "loss": 2.6187, "step": 84000 }, { "epoch": 2.48, "learning_rate": 1.016485106678552e-05, "loss": 2.5851, "step": 84100 }, { "epoch": 2.49, "learning_rate": 1.0106967967492852e-05, "loss": 2.5945, "step": 84200 }, { "epoch": 2.49, "learning_rate": 1.0049084868200183e-05, "loss": 2.6874, "step": 84300 }, { "epoch": 2.49, "learning_rate": 9.991780599900441e-06, "loss": 2.6016, "step": 84400 }, { "epoch": 2.49, "learning_rate": 9.933897500607773e-06, "loss": 2.7371, "step": 84500 }, { "epoch": 2.5, "learning_rate": 9.876014401315105e-06, "loss": 2.6355, "step": 84600 }, { "epoch": 2.5, "learning_rate": 9.818131302022437e-06, "loss": 2.6629, "step": 84700 }, { "epoch": 2.5, "learning_rate": 9.760248202729767e-06, "loss": 2.6944, "step": 84800 }, { "epoch": 2.51, "learning_rate": 9.702365103437098e-06, "loss": 2.6812, "step": 84900 }, { "epoch": 2.51, "learning_rate": 9.64448200414443e-06, "loss": 2.6876, "step": 85000 }, { "epoch": 2.51, "learning_rate": 9.586598904851762e-06, "loss": 2.6602, "step": 85100 }, { "epoch": 2.52, "learning_rate": 9.528715805559094e-06, "loss": 2.6384, "step": 85200 }, { "epoch": 2.52, "learning_rate": 9.470832706266426e-06, "loss": 2.6552, "step": 85300 }, { "epoch": 2.52, "learning_rate": 9.412949606973756e-06, "loss": 2.6656, "step": 85400 }, { "epoch": 2.52, "learning_rate": 9.355066507681088e-06, "loss": 2.6523, "step": 85500 }, { "epoch": 2.53, "learning_rate": 9.29718340838842e-06, "loss": 2.741, "step": 85600 }, { "epoch": 2.53, "learning_rate": 9.239300309095751e-06, "loss": 2.6532, "step": 85700 }, { "epoch": 2.53, "learning_rate": 9.181417209803083e-06, "loss": 2.6828, "step": 85800 }, { "epoch": 2.54, "learning_rate": 9.123534110510413e-06, "loss": 2.7142, "step": 85900 }, { "epoch": 2.54, "learning_rate": 9.065651011217745e-06, "loss": 2.6404, "step": 86000 }, { "epoch": 2.54, "learning_rate": 9.007767911925077e-06, "loss": 2.661, "step": 86100 }, { "epoch": 2.54, "learning_rate": 8.949884812632408e-06, "loss": 2.676, "step": 86200 }, { "epoch": 2.55, "learning_rate": 8.89200171333974e-06, "loss": 2.6487, "step": 86300 }, { "epoch": 2.55, "learning_rate": 8.83411861404707e-06, "loss": 2.6235, "step": 86400 }, { "epoch": 2.55, "learning_rate": 8.776235514754402e-06, "loss": 2.6581, "step": 86500 }, { "epoch": 2.56, "learning_rate": 8.718352415461734e-06, "loss": 2.6467, "step": 86600 }, { "epoch": 2.56, "learning_rate": 8.660469316169064e-06, "loss": 2.6874, "step": 86700 }, { "epoch": 2.56, "learning_rate": 8.602586216876397e-06, "loss": 2.5841, "step": 86800 }, { "epoch": 2.57, "learning_rate": 8.54470311758373e-06, "loss": 2.6598, "step": 86900 }, { "epoch": 2.57, "learning_rate": 8.48682001829106e-06, "loss": 2.6999, "step": 87000 }, { "epoch": 2.57, "learning_rate": 8.428936918998391e-06, "loss": 2.6375, "step": 87100 }, { "epoch": 2.57, "learning_rate": 8.371053819705723e-06, "loss": 2.6541, "step": 87200 }, { "epoch": 2.58, "learning_rate": 8.313170720413053e-06, "loss": 2.6914, "step": 87300 }, { "epoch": 2.58, "learning_rate": 8.255287621120387e-06, "loss": 2.6972, "step": 87400 }, { "epoch": 2.58, "learning_rate": 8.197404521827717e-06, "loss": 2.6217, "step": 87500 }, { "epoch": 2.59, "learning_rate": 8.139521422535048e-06, "loss": 2.6914, "step": 87600 }, { "epoch": 2.59, "learning_rate": 8.08163832324238e-06, "loss": 2.6, "step": 87700 }, { "epoch": 2.59, "learning_rate": 8.02375522394971e-06, "loss": 2.6203, "step": 87800 }, { "epoch": 2.59, "learning_rate": 7.965872124657044e-06, "loss": 2.6175, "step": 87900 }, { "epoch": 2.6, "learning_rate": 7.907989025364376e-06, "loss": 2.6322, "step": 88000 }, { "epoch": 2.6, "learning_rate": 7.850105926071706e-06, "loss": 2.6456, "step": 88100 }, { "epoch": 2.6, "learning_rate": 7.792222826779038e-06, "loss": 2.6561, "step": 88200 }, { "epoch": 2.61, "learning_rate": 7.73433972748637e-06, "loss": 2.5974, "step": 88300 }, { "epoch": 2.61, "learning_rate": 7.6764566281937e-06, "loss": 2.6547, "step": 88400 }, { "epoch": 2.61, "learning_rate": 7.618573528901032e-06, "loss": 2.6182, "step": 88500 }, { "epoch": 2.62, "learning_rate": 7.560690429608364e-06, "loss": 2.5736, "step": 88600 }, { "epoch": 2.62, "learning_rate": 7.502807330315695e-06, "loss": 2.6473, "step": 88700 }, { "epoch": 2.62, "learning_rate": 7.445503062015953e-06, "loss": 2.6361, "step": 88800 }, { "epoch": 2.62, "learning_rate": 7.387619962723284e-06, "loss": 2.6453, "step": 88900 }, { "epoch": 2.63, "learning_rate": 7.329736863430615e-06, "loss": 2.6865, "step": 89000 }, { "epoch": 2.63, "learning_rate": 7.271853764137946e-06, "loss": 2.5792, "step": 89100 }, { "epoch": 2.63, "learning_rate": 7.213970664845279e-06, "loss": 2.6216, "step": 89200 }, { "epoch": 2.64, "learning_rate": 7.156666396545538e-06, "loss": 2.6165, "step": 89300 }, { "epoch": 2.64, "learning_rate": 7.0987832972528685e-06, "loss": 2.6615, "step": 89400 }, { "epoch": 2.64, "learning_rate": 7.0409001979601995e-06, "loss": 2.6344, "step": 89500 }, { "epoch": 2.65, "learning_rate": 6.983017098667531e-06, "loss": 2.6097, "step": 89600 }, { "epoch": 2.65, "learning_rate": 6.925133999374862e-06, "loss": 2.6498, "step": 89700 }, { "epoch": 2.65, "learning_rate": 6.867250900082195e-06, "loss": 2.7191, "step": 89800 }, { "epoch": 2.65, "learning_rate": 6.809367800789526e-06, "loss": 2.6154, "step": 89900 }, { "epoch": 2.66, "learning_rate": 6.751484701496858e-06, "loss": 2.5856, "step": 90000 }, { "epoch": 2.66, "learning_rate": 6.6936016022041886e-06, "loss": 2.6058, "step": 90100 }, { "epoch": 2.66, "learning_rate": 6.6357185029115195e-06, "loss": 2.7033, "step": 90200 }, { "epoch": 2.67, "learning_rate": 6.577835403618851e-06, "loss": 2.6258, "step": 90300 }, { "epoch": 2.67, "learning_rate": 6.519952304326184e-06, "loss": 2.6289, "step": 90400 }, { "epoch": 2.67, "learning_rate": 6.462069205033515e-06, "loss": 2.6527, "step": 90500 }, { "epoch": 2.67, "learning_rate": 6.404186105740846e-06, "loss": 2.5685, "step": 90600 }, { "epoch": 2.68, "learning_rate": 6.346303006448178e-06, "loss": 2.6636, "step": 90700 }, { "epoch": 2.68, "learning_rate": 6.288419907155509e-06, "loss": 2.6443, "step": 90800 }, { "epoch": 2.68, "learning_rate": 6.23053680786284e-06, "loss": 2.6078, "step": 90900 }, { "epoch": 2.69, "learning_rate": 6.172653708570171e-06, "loss": 2.6654, "step": 91000 }, { "epoch": 2.69, "learning_rate": 6.114770609277504e-06, "loss": 2.6124, "step": 91100 }, { "epoch": 2.69, "learning_rate": 6.056887509984835e-06, "loss": 2.6233, "step": 91200 }, { "epoch": 2.7, "learning_rate": 5.999004410692166e-06, "loss": 2.614, "step": 91300 }, { "epoch": 2.7, "learning_rate": 5.941121311399498e-06, "loss": 2.6472, "step": 91400 }, { "epoch": 2.7, "learning_rate": 5.8832382121068295e-06, "loss": 2.6337, "step": 91500 }, { "epoch": 2.7, "learning_rate": 5.82535511281416e-06, "loss": 2.5766, "step": 91600 }, { "epoch": 2.71, "learning_rate": 5.767472013521492e-06, "loss": 2.6287, "step": 91700 }, { "epoch": 2.71, "learning_rate": 5.709588914228824e-06, "loss": 2.5166, "step": 91800 }, { "epoch": 2.71, "learning_rate": 5.651705814936156e-06, "loss": 2.6885, "step": 91900 }, { "epoch": 2.72, "learning_rate": 5.593822715643487e-06, "loss": 2.6536, "step": 92000 }, { "epoch": 2.72, "learning_rate": 5.535939616350818e-06, "loss": 2.6554, "step": 92100 }, { "epoch": 2.72, "learning_rate": 5.47805651705815e-06, "loss": 2.5428, "step": 92200 }, { "epoch": 2.72, "learning_rate": 5.420173417765481e-06, "loss": 2.6454, "step": 92300 }, { "epoch": 2.73, "learning_rate": 5.362290318472812e-06, "loss": 2.6419, "step": 92400 }, { "epoch": 2.73, "learning_rate": 5.304407219180144e-06, "loss": 2.6316, "step": 92500 }, { "epoch": 2.73, "learning_rate": 5.247102950880403e-06, "loss": 2.6203, "step": 92600 }, { "epoch": 2.74, "learning_rate": 5.189219851587734e-06, "loss": 2.604, "step": 92700 }, { "epoch": 2.74, "learning_rate": 5.131336752295065e-06, "loss": 2.6726, "step": 92800 }, { "epoch": 2.74, "learning_rate": 5.073453653002397e-06, "loss": 2.629, "step": 92900 }, { "epoch": 2.75, "learning_rate": 5.015570553709728e-06, "loss": 2.6562, "step": 93000 }, { "epoch": 2.75, "learning_rate": 4.957687454417059e-06, "loss": 2.637, "step": 93100 }, { "epoch": 2.75, "learning_rate": 4.899804355124391e-06, "loss": 2.6057, "step": 93200 }, { "epoch": 2.75, "learning_rate": 4.841921255831723e-06, "loss": 2.7188, "step": 93300 }, { "epoch": 2.76, "learning_rate": 4.784038156539054e-06, "loss": 2.6679, "step": 93400 }, { "epoch": 2.76, "learning_rate": 4.7261550572463855e-06, "loss": 2.658, "step": 93500 }, { "epoch": 2.76, "learning_rate": 4.668271957953716e-06, "loss": 2.7056, "step": 93600 }, { "epoch": 2.77, "learning_rate": 4.610388858661048e-06, "loss": 2.568, "step": 93700 }, { "epoch": 2.77, "learning_rate": 4.55250575936838e-06, "loss": 2.6135, "step": 93800 }, { "epoch": 2.77, "learning_rate": 4.494622660075711e-06, "loss": 2.6412, "step": 93900 }, { "epoch": 2.77, "learning_rate": 4.436739560783043e-06, "loss": 2.6006, "step": 94000 }, { "epoch": 2.78, "learning_rate": 4.3788564614903745e-06, "loss": 2.6539, "step": 94100 }, { "epoch": 2.78, "learning_rate": 4.3209733621977055e-06, "loss": 2.5896, "step": 94200 }, { "epoch": 2.78, "learning_rate": 4.263090262905037e-06, "loss": 2.6026, "step": 94300 }, { "epoch": 2.79, "learning_rate": 4.205207163612369e-06, "loss": 2.6195, "step": 94400 }, { "epoch": 2.79, "learning_rate": 4.1473240643197e-06, "loss": 2.6127, "step": 94500 }, { "epoch": 2.79, "learning_rate": 4.089440965027032e-06, "loss": 2.6983, "step": 94600 }, { "epoch": 2.8, "learning_rate": 4.031557865734363e-06, "loss": 2.5699, "step": 94700 }, { "epoch": 2.8, "learning_rate": 3.9736747664416946e-06, "loss": 2.6127, "step": 94800 }, { "epoch": 2.8, "learning_rate": 3.915791667149026e-06, "loss": 2.6966, "step": 94900 }, { "epoch": 2.8, "learning_rate": 3.857908567856357e-06, "loss": 2.6416, "step": 95000 }, { "epoch": 2.81, "learning_rate": 3.8000254685636887e-06, "loss": 2.6861, "step": 95100 }, { "epoch": 2.81, "learning_rate": 3.7421423692710205e-06, "loss": 2.5394, "step": 95200 }, { "epoch": 2.81, "learning_rate": 3.684259269978352e-06, "loss": 2.5937, "step": 95300 }, { "epoch": 2.82, "learning_rate": 3.626376170685683e-06, "loss": 2.6151, "step": 95400 }, { "epoch": 2.82, "learning_rate": 3.568493071393015e-06, "loss": 2.6567, "step": 95500 }, { "epoch": 2.82, "learning_rate": 3.5106099721003464e-06, "loss": 2.5833, "step": 95600 }, { "epoch": 2.83, "learning_rate": 3.452726872807678e-06, "loss": 2.6624, "step": 95700 }, { "epoch": 2.83, "learning_rate": 3.394843773515009e-06, "loss": 2.6202, "step": 95800 }, { "epoch": 2.83, "learning_rate": 3.3369606742223405e-06, "loss": 2.5676, "step": 95900 }, { "epoch": 2.83, "learning_rate": 3.2790775749296723e-06, "loss": 2.6249, "step": 96000 }, { "epoch": 2.84, "learning_rate": 3.2211944756370037e-06, "loss": 2.6144, "step": 96100 }, { "epoch": 2.84, "learning_rate": 3.163311376344335e-06, "loss": 2.6163, "step": 96200 }, { "epoch": 2.84, "learning_rate": 3.1054282770516664e-06, "loss": 2.5738, "step": 96300 }, { "epoch": 2.85, "learning_rate": 3.047545177758998e-06, "loss": 2.6832, "step": 96400 }, { "epoch": 2.85, "learning_rate": 2.9896620784663296e-06, "loss": 2.5983, "step": 96500 }, { "epoch": 2.85, "learning_rate": 2.932357810166588e-06, "loss": 2.6802, "step": 96600 }, { "epoch": 2.85, "learning_rate": 2.875053541866846e-06, "loss": 2.6392, "step": 96700 }, { "epoch": 2.86, "learning_rate": 2.817170442574177e-06, "loss": 2.627, "step": 96800 }, { "epoch": 2.86, "learning_rate": 2.7592873432815088e-06, "loss": 2.6258, "step": 96900 }, { "epoch": 2.86, "learning_rate": 2.70140424398884e-06, "loss": 2.5845, "step": 97000 }, { "epoch": 2.87, "learning_rate": 2.643521144696172e-06, "loss": 2.6863, "step": 97100 }, { "epoch": 2.87, "learning_rate": 2.5856380454035033e-06, "loss": 2.5749, "step": 97200 }, { "epoch": 2.87, "learning_rate": 2.5277549461108347e-06, "loss": 2.5582, "step": 97300 }, { "epoch": 2.88, "learning_rate": 2.469871846818166e-06, "loss": 2.6249, "step": 97400 }, { "epoch": 2.88, "learning_rate": 2.4119887475254974e-06, "loss": 2.6195, "step": 97500 }, { "epoch": 2.88, "learning_rate": 2.3541056482328292e-06, "loss": 2.5851, "step": 97600 }, { "epoch": 2.88, "learning_rate": 2.2962225489401606e-06, "loss": 2.6251, "step": 97700 }, { "epoch": 2.89, "learning_rate": 2.238339449647492e-06, "loss": 2.6062, "step": 97800 }, { "epoch": 2.89, "learning_rate": 2.1804563503548233e-06, "loss": 2.6002, "step": 97900 }, { "epoch": 2.89, "learning_rate": 2.122573251062155e-06, "loss": 2.5926, "step": 98000 }, { "epoch": 2.9, "learning_rate": 2.0646901517694865e-06, "loss": 2.5646, "step": 98100 }, { "epoch": 2.9, "learning_rate": 2.006807052476818e-06, "loss": 2.6043, "step": 98200 }, { "epoch": 2.9, "learning_rate": 1.9489239531841493e-06, "loss": 2.6333, "step": 98300 }, { "epoch": 2.9, "learning_rate": 1.891040853891481e-06, "loss": 2.6832, "step": 98400 }, { "epoch": 2.91, "learning_rate": 1.8331577545988122e-06, "loss": 2.5962, "step": 98500 }, { "epoch": 2.91, "learning_rate": 1.7752746553061438e-06, "loss": 2.5622, "step": 98600 }, { "epoch": 2.91, "learning_rate": 1.7173915560134754e-06, "loss": 2.5946, "step": 98700 }, { "epoch": 2.92, "learning_rate": 1.659508456720807e-06, "loss": 2.5805, "step": 98800 }, { "epoch": 2.92, "learning_rate": 1.6016253574281381e-06, "loss": 2.6398, "step": 98900 }, { "epoch": 2.92, "learning_rate": 1.5437422581354697e-06, "loss": 2.5873, "step": 99000 }, { "epoch": 2.93, "learning_rate": 1.485859158842801e-06, "loss": 2.5955, "step": 99100 }, { "epoch": 2.93, "learning_rate": 1.4279760595501327e-06, "loss": 2.6957, "step": 99200 }, { "epoch": 2.93, "learning_rate": 1.370092960257464e-06, "loss": 2.5805, "step": 99300 }, { "epoch": 2.93, "learning_rate": 1.3122098609647956e-06, "loss": 2.5943, "step": 99400 }, { "epoch": 2.94, "learning_rate": 1.254326761672127e-06, "loss": 2.652, "step": 99500 }, { "epoch": 2.94, "learning_rate": 1.1964436623794584e-06, "loss": 2.656, "step": 99600 }, { "epoch": 2.94, "learning_rate": 1.13856056308679e-06, "loss": 2.5938, "step": 99700 }, { "epoch": 2.95, "learning_rate": 1.0806774637941215e-06, "loss": 2.5805, "step": 99800 }, { "epoch": 2.95, "learning_rate": 1.0227943645014529e-06, "loss": 2.6259, "step": 99900 }, { "epoch": 2.95, "learning_rate": 9.649112652087845e-07, "loss": 2.6734, "step": 100000 }, { "epoch": 2.95, "learning_rate": 9.070281659161157e-07, "loss": 2.6134, "step": 100100 }, { "epoch": 2.96, "learning_rate": 8.491450666234473e-07, "loss": 2.6146, "step": 100200 }, { "epoch": 2.96, "learning_rate": 7.912619673307788e-07, "loss": 2.6645, "step": 100300 }, { "epoch": 2.96, "learning_rate": 7.333788680381103e-07, "loss": 2.6002, "step": 100400 }, { "epoch": 2.97, "learning_rate": 6.754957687454418e-07, "loss": 2.6512, "step": 100500 }, { "epoch": 2.97, "learning_rate": 6.176126694527732e-07, "loss": 2.6164, "step": 100600 }, { "epoch": 2.97, "learning_rate": 5.597295701601047e-07, "loss": 2.5653, "step": 100700 }, { "epoch": 2.98, "learning_rate": 5.024253018603628e-07, "loss": 2.5852, "step": 100800 }, { "epoch": 2.98, "learning_rate": 4.4454220256769427e-07, "loss": 2.6347, "step": 100900 }, { "epoch": 2.98, "learning_rate": 3.866591032750258e-07, "loss": 2.589, "step": 101000 }, { "epoch": 2.98, "learning_rate": 3.287760039823572e-07, "loss": 2.5523, "step": 101100 }, { "epoch": 2.99, "learning_rate": 2.708929046896887e-07, "loss": 2.6118, "step": 101200 }, { "epoch": 2.99, "learning_rate": 2.130098053970202e-07, "loss": 2.6409, "step": 101300 }, { "epoch": 2.99, "learning_rate": 1.5570553709727834e-07, "loss": 2.5427, "step": 101400 }, { "epoch": 3.0, "learning_rate": 9.782243780460982e-08, "loss": 2.6227, "step": 101500 }, { "epoch": 3.0, "learning_rate": 3.993933851194129e-08, "loss": 2.5889, "step": 101600 }, { "epoch": 3.0, "eval_gen_len": 19.8855, "eval_loss": 3.09702730178833, "eval_rouge1": 19.1763, "eval_rouge2": 5.4517, "eval_rougeL": 15.5342, "eval_rougeLsum": 16.7186, "eval_runtime": 748.2031, "eval_samples_per_second": 10.06, "eval_steps_per_second": 2.515, "step": 101625 }, { "epoch": 3.0, "step": 101625, "total_flos": 7.175645735046021e+17, "train_loss": 3.0545432459982558, "train_runtime": 34500.7445, "train_samples_per_second": 11.782, "train_steps_per_second": 2.946 } ], "max_steps": 101625, "num_train_epochs": 3, "total_flos": 7.175645735046021e+17, "trial_name": null, "trial_params": null }