{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 421875, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.9964515555555556e-05, "loss": 0.0351, "step": 500 }, { "epoch": 0.01, "learning_rate": 2.9928959999999998e-05, "loss": 0.0074, "step": 1000 }, { "epoch": 0.01, "learning_rate": 2.9893404444444444e-05, "loss": 0.0063, "step": 1500 }, { "epoch": 0.01, "learning_rate": 2.985784888888889e-05, "loss": 0.0055, "step": 2000 }, { "epoch": 0.02, "learning_rate": 2.9822293333333335e-05, "loss": 0.0051, "step": 2500 }, { "epoch": 0.02, "learning_rate": 2.9786737777777777e-05, "loss": 0.0047, "step": 3000 }, { "epoch": 0.02, "learning_rate": 2.9751182222222222e-05, "loss": 0.0043, "step": 3500 }, { "epoch": 0.03, "learning_rate": 2.9715626666666668e-05, "loss": 0.0042, "step": 4000 }, { "epoch": 0.03, "learning_rate": 2.9680071111111114e-05, "loss": 0.0039, "step": 4500 }, { "epoch": 0.04, "learning_rate": 2.9644515555555556e-05, "loss": 0.0038, "step": 5000 }, { "epoch": 0.04, "learning_rate": 2.960896e-05, "loss": 0.0036, "step": 5500 }, { "epoch": 0.04, "learning_rate": 2.9573404444444447e-05, "loss": 0.0034, "step": 6000 }, { "epoch": 0.05, "learning_rate": 2.9537848888888892e-05, "loss": 0.0033, "step": 6500 }, { "epoch": 0.05, "learning_rate": 2.9502293333333334e-05, "loss": 0.0033, "step": 7000 }, { "epoch": 0.05, "learning_rate": 2.9466737777777777e-05, "loss": 0.0032, "step": 7500 }, { "epoch": 0.06, "learning_rate": 2.9431182222222222e-05, "loss": 0.0031, "step": 8000 }, { "epoch": 0.06, "learning_rate": 2.9395626666666668e-05, "loss": 0.0032, "step": 8500 }, { "epoch": 0.06, "learning_rate": 2.936007111111111e-05, "loss": 0.0029, "step": 9000 }, { "epoch": 0.07, "learning_rate": 2.9324515555555555e-05, "loss": 0.0029, "step": 9500 }, { "epoch": 0.07, "learning_rate": 2.928896e-05, "loss": 0.0029, "step": 10000 }, { "epoch": 0.07, "learning_rate": 2.9253475555555556e-05, "loss": 0.0028, "step": 10500 }, { "epoch": 0.08, "learning_rate": 2.9217919999999998e-05, "loss": 0.0028, "step": 11000 }, { "epoch": 0.08, "learning_rate": 2.9182364444444444e-05, "loss": 0.0027, "step": 11500 }, { "epoch": 0.09, "learning_rate": 2.914680888888889e-05, "loss": 0.0027, "step": 12000 }, { "epoch": 0.09, "learning_rate": 2.9111324444444445e-05, "loss": 0.0028, "step": 12500 }, { "epoch": 0.09, "learning_rate": 2.907576888888889e-05, "loss": 0.0027, "step": 13000 }, { "epoch": 0.1, "learning_rate": 2.9040213333333336e-05, "loss": 0.0026, "step": 13500 }, { "epoch": 0.1, "learning_rate": 2.900465777777778e-05, "loss": 0.0026, "step": 14000 }, { "epoch": 0.1, "learning_rate": 2.8969173333333333e-05, "loss": 0.0026, "step": 14500 }, { "epoch": 0.11, "learning_rate": 2.893361777777778e-05, "loss": 0.0025, "step": 15000 }, { "epoch": 0.11, "learning_rate": 2.8898062222222224e-05, "loss": 0.0025, "step": 15500 }, { "epoch": 0.11, "learning_rate": 2.886250666666667e-05, "loss": 0.0025, "step": 16000 }, { "epoch": 0.12, "learning_rate": 2.882702222222222e-05, "loss": 0.0025, "step": 16500 }, { "epoch": 0.12, "learning_rate": 2.8791466666666667e-05, "loss": 0.0024, "step": 17000 }, { "epoch": 0.12, "learning_rate": 2.8755911111111113e-05, "loss": 0.0024, "step": 17500 }, { "epoch": 0.13, "learning_rate": 2.8720355555555558e-05, "loss": 0.0024, "step": 18000 }, { "epoch": 0.13, "learning_rate": 2.868487111111111e-05, "loss": 0.0024, "step": 18500 }, { "epoch": 0.14, "learning_rate": 2.8649315555555555e-05, "loss": 0.0024, "step": 19000 }, { "epoch": 0.14, "learning_rate": 2.861376e-05, "loss": 0.0024, "step": 19500 }, { "epoch": 0.14, "learning_rate": 2.8578204444444443e-05, "loss": 0.0023, "step": 20000 }, { "epoch": 0.15, "learning_rate": 2.8542720000000002e-05, "loss": 0.0023, "step": 20500 }, { "epoch": 0.15, "learning_rate": 2.8507164444444447e-05, "loss": 0.0023, "step": 21000 }, { "epoch": 0.15, "learning_rate": 2.847160888888889e-05, "loss": 0.0023, "step": 21500 }, { "epoch": 0.16, "learning_rate": 2.843605333333333e-05, "loss": 0.0023, "step": 22000 }, { "epoch": 0.16, "learning_rate": 2.8400497777777777e-05, "loss": 0.0022, "step": 22500 }, { "epoch": 0.16, "learning_rate": 2.8365013333333336e-05, "loss": 0.0022, "step": 23000 }, { "epoch": 0.17, "learning_rate": 2.832945777777778e-05, "loss": 0.0022, "step": 23500 }, { "epoch": 0.17, "learning_rate": 2.8293902222222223e-05, "loss": 0.0022, "step": 24000 }, { "epoch": 0.17, "learning_rate": 2.825834666666667e-05, "loss": 0.0021, "step": 24500 }, { "epoch": 0.18, "learning_rate": 2.8222862222222224e-05, "loss": 0.0023, "step": 25000 }, { "epoch": 0.18, "learning_rate": 2.8187306666666666e-05, "loss": 0.0022, "step": 25500 }, { "epoch": 0.18, "learning_rate": 2.8151751111111112e-05, "loss": 0.0021, "step": 26000 }, { "epoch": 0.19, "learning_rate": 2.8116195555555557e-05, "loss": 0.0021, "step": 26500 }, { "epoch": 0.19, "learning_rate": 2.8080711111111113e-05, "loss": 0.0021, "step": 27000 }, { "epoch": 0.2, "learning_rate": 2.8045155555555555e-05, "loss": 0.0022, "step": 27500 }, { "epoch": 0.2, "learning_rate": 2.80096e-05, "loss": 0.0022, "step": 28000 }, { "epoch": 0.2, "learning_rate": 2.7974044444444446e-05, "loss": 0.0022, "step": 28500 }, { "epoch": 0.21, "learning_rate": 2.793856e-05, "loss": 0.0021, "step": 29000 }, { "epoch": 0.21, "learning_rate": 2.7903004444444443e-05, "loss": 0.0021, "step": 29500 }, { "epoch": 0.21, "learning_rate": 2.786744888888889e-05, "loss": 0.0021, "step": 30000 }, { "epoch": 0.22, "learning_rate": 2.7831893333333334e-05, "loss": 0.0021, "step": 30500 }, { "epoch": 0.22, "learning_rate": 2.779633777777778e-05, "loss": 0.0021, "step": 31000 }, { "epoch": 0.22, "learning_rate": 2.776085333333333e-05, "loss": 0.0021, "step": 31500 }, { "epoch": 0.23, "learning_rate": 2.7725297777777777e-05, "loss": 0.0021, "step": 32000 }, { "epoch": 0.23, "learning_rate": 2.7689742222222223e-05, "loss": 0.002, "step": 32500 }, { "epoch": 0.23, "learning_rate": 2.7654186666666665e-05, "loss": 0.002, "step": 33000 }, { "epoch": 0.24, "learning_rate": 2.7618702222222223e-05, "loss": 0.0021, "step": 33500 }, { "epoch": 0.24, "learning_rate": 2.758314666666667e-05, "loss": 0.0021, "step": 34000 }, { "epoch": 0.25, "learning_rate": 2.7547591111111115e-05, "loss": 0.0021, "step": 34500 }, { "epoch": 0.25, "learning_rate": 2.7512035555555553e-05, "loss": 0.0021, "step": 35000 }, { "epoch": 0.25, "learning_rate": 2.7476551111111112e-05, "loss": 0.002, "step": 35500 }, { "epoch": 0.26, "learning_rate": 2.7440995555555557e-05, "loss": 0.002, "step": 36000 }, { "epoch": 0.26, "learning_rate": 2.7405440000000003e-05, "loss": 0.0021, "step": 36500 }, { "epoch": 0.26, "learning_rate": 2.7369884444444445e-05, "loss": 0.002, "step": 37000 }, { "epoch": 0.27, "learning_rate": 2.733432888888889e-05, "loss": 0.002, "step": 37500 }, { "epoch": 0.27, "learning_rate": 2.7298844444444446e-05, "loss": 0.002, "step": 38000 }, { "epoch": 0.27, "learning_rate": 2.726328888888889e-05, "loss": 0.002, "step": 38500 }, { "epoch": 0.28, "learning_rate": 2.7227733333333334e-05, "loss": 0.002, "step": 39000 }, { "epoch": 0.28, "learning_rate": 2.719217777777778e-05, "loss": 0.002, "step": 39500 }, { "epoch": 0.28, "learning_rate": 2.7156693333333334e-05, "loss": 0.002, "step": 40000 }, { "epoch": 0.29, "learning_rate": 2.7121137777777777e-05, "loss": 0.002, "step": 40500 }, { "epoch": 0.29, "learning_rate": 2.7085582222222222e-05, "loss": 0.002, "step": 41000 }, { "epoch": 0.3, "learning_rate": 2.7050026666666668e-05, "loss": 0.002, "step": 41500 }, { "epoch": 0.3, "learning_rate": 2.7014542222222223e-05, "loss": 0.002, "step": 42000 }, { "epoch": 0.3, "learning_rate": 2.6978986666666665e-05, "loss": 0.0019, "step": 42500 }, { "epoch": 0.31, "learning_rate": 2.694343111111111e-05, "loss": 0.0019, "step": 43000 }, { "epoch": 0.31, "learning_rate": 2.6907875555555556e-05, "loss": 0.0019, "step": 43500 }, { "epoch": 0.31, "learning_rate": 2.687232e-05, "loss": 0.002, "step": 44000 }, { "epoch": 0.32, "learning_rate": 2.6836835555555557e-05, "loss": 0.002, "step": 44500 }, { "epoch": 0.32, "learning_rate": 2.680128e-05, "loss": 0.0019, "step": 45000 }, { "epoch": 0.32, "learning_rate": 2.6765724444444445e-05, "loss": 0.0019, "step": 45500 }, { "epoch": 0.33, "learning_rate": 2.673016888888889e-05, "loss": 0.002, "step": 46000 }, { "epoch": 0.33, "learning_rate": 2.6694613333333332e-05, "loss": 0.0019, "step": 46500 }, { "epoch": 0.33, "learning_rate": 2.6659057777777778e-05, "loss": 0.0019, "step": 47000 }, { "epoch": 0.34, "learning_rate": 2.6623573333333336e-05, "loss": 0.002, "step": 47500 }, { "epoch": 0.34, "learning_rate": 2.658801777777778e-05, "loss": 0.0019, "step": 48000 }, { "epoch": 0.34, "learning_rate": 2.655246222222222e-05, "loss": 0.0019, "step": 48500 }, { "epoch": 0.35, "learning_rate": 2.6516906666666666e-05, "loss": 0.0019, "step": 49000 }, { "epoch": 0.35, "learning_rate": 2.6481351111111112e-05, "loss": 0.0019, "step": 49500 }, { "epoch": 0.36, "learning_rate": 2.6445866666666667e-05, "loss": 0.0019, "step": 50000 }, { "epoch": 0.36, "learning_rate": 2.6410311111111112e-05, "loss": 0.0019, "step": 50500 }, { "epoch": 0.36, "learning_rate": 2.6374755555555558e-05, "loss": 0.0019, "step": 51000 }, { "epoch": 0.37, "learning_rate": 2.6339200000000004e-05, "loss": 0.0019, "step": 51500 }, { "epoch": 0.37, "learning_rate": 2.6303715555555555e-05, "loss": 0.0019, "step": 52000 }, { "epoch": 0.37, "learning_rate": 2.626816e-05, "loss": 0.0019, "step": 52500 }, { "epoch": 0.38, "learning_rate": 2.6232604444444446e-05, "loss": 0.0019, "step": 53000 }, { "epoch": 0.38, "learning_rate": 2.6197048888888892e-05, "loss": 0.0018, "step": 53500 }, { "epoch": 0.38, "learning_rate": 2.6161564444444444e-05, "loss": 0.0019, "step": 54000 }, { "epoch": 0.39, "learning_rate": 2.612600888888889e-05, "loss": 0.0019, "step": 54500 }, { "epoch": 0.39, "learning_rate": 2.6090453333333335e-05, "loss": 0.0019, "step": 55000 }, { "epoch": 0.39, "learning_rate": 2.6054897777777777e-05, "loss": 0.0019, "step": 55500 }, { "epoch": 0.4, "learning_rate": 2.6019413333333332e-05, "loss": 0.0018, "step": 56000 }, { "epoch": 0.4, "learning_rate": 2.5983857777777778e-05, "loss": 0.0018, "step": 56500 }, { "epoch": 0.41, "learning_rate": 2.5948302222222223e-05, "loss": 0.0019, "step": 57000 }, { "epoch": 0.41, "learning_rate": 2.5912746666666666e-05, "loss": 0.0019, "step": 57500 }, { "epoch": 0.41, "learning_rate": 2.587719111111111e-05, "loss": 0.0018, "step": 58000 }, { "epoch": 0.42, "learning_rate": 2.5841706666666666e-05, "loss": 0.0018, "step": 58500 }, { "epoch": 0.42, "learning_rate": 2.5806151111111112e-05, "loss": 0.0018, "step": 59000 }, { "epoch": 0.42, "learning_rate": 2.5770595555555554e-05, "loss": 0.0018, "step": 59500 }, { "epoch": 0.43, "learning_rate": 2.573504e-05, "loss": 0.0018, "step": 60000 }, { "epoch": 0.43, "learning_rate": 2.5699484444444445e-05, "loss": 0.0019, "step": 60500 }, { "epoch": 0.43, "learning_rate": 2.5664e-05, "loss": 0.0018, "step": 61000 }, { "epoch": 0.44, "learning_rate": 2.5628444444444446e-05, "loss": 0.0018, "step": 61500 }, { "epoch": 0.44, "learning_rate": 2.5592888888888888e-05, "loss": 0.0018, "step": 62000 }, { "epoch": 0.44, "learning_rate": 2.5557333333333333e-05, "loss": 0.0019, "step": 62500 }, { "epoch": 0.45, "learning_rate": 2.552184888888889e-05, "loss": 0.0018, "step": 63000 }, { "epoch": 0.45, "learning_rate": 2.5486293333333334e-05, "loss": 0.0018, "step": 63500 }, { "epoch": 0.46, "learning_rate": 2.545073777777778e-05, "loss": 0.0018, "step": 64000 }, { "epoch": 0.46, "learning_rate": 2.5415182222222225e-05, "loss": 0.0018, "step": 64500 }, { "epoch": 0.46, "learning_rate": 2.5379697777777777e-05, "loss": 0.0018, "step": 65000 }, { "epoch": 0.47, "learning_rate": 2.5344142222222223e-05, "loss": 0.0018, "step": 65500 }, { "epoch": 0.47, "learning_rate": 2.5308586666666668e-05, "loss": 0.0018, "step": 66000 }, { "epoch": 0.47, "learning_rate": 2.5273031111111114e-05, "loss": 0.0018, "step": 66500 }, { "epoch": 0.48, "learning_rate": 2.5237475555555556e-05, "loss": 0.0018, "step": 67000 }, { "epoch": 0.48, "learning_rate": 2.520192e-05, "loss": 0.0018, "step": 67500 }, { "epoch": 0.48, "learning_rate": 2.5166364444444447e-05, "loss": 0.0018, "step": 68000 }, { "epoch": 0.49, "learning_rate": 2.513088e-05, "loss": 0.0018, "step": 68500 }, { "epoch": 0.49, "learning_rate": 2.5095324444444444e-05, "loss": 0.0018, "step": 69000 }, { "epoch": 0.49, "learning_rate": 2.505976888888889e-05, "loss": 0.0018, "step": 69500 }, { "epoch": 0.5, "learning_rate": 2.5024213333333335e-05, "loss": 0.0018, "step": 70000 }, { "epoch": 0.5, "learning_rate": 2.4988728888888887e-05, "loss": 0.0017, "step": 70500 }, { "epoch": 0.5, "learning_rate": 2.4953173333333333e-05, "loss": 0.0018, "step": 71000 }, { "epoch": 0.51, "learning_rate": 2.491761777777778e-05, "loss": 0.0018, "step": 71500 }, { "epoch": 0.51, "learning_rate": 2.4882062222222224e-05, "loss": 0.0018, "step": 72000 }, { "epoch": 0.52, "learning_rate": 2.4846506666666666e-05, "loss": 0.0018, "step": 72500 }, { "epoch": 0.52, "learning_rate": 2.481102222222222e-05, "loss": 0.0017, "step": 73000 }, { "epoch": 0.52, "learning_rate": 2.4775466666666667e-05, "loss": 0.0018, "step": 73500 }, { "epoch": 0.53, "learning_rate": 2.4739911111111112e-05, "loss": 0.0018, "step": 74000 }, { "epoch": 0.53, "learning_rate": 2.4704355555555554e-05, "loss": 0.0017, "step": 74500 }, { "epoch": 0.53, "learning_rate": 2.46688e-05, "loss": 0.0018, "step": 75000 }, { "epoch": 0.54, "learning_rate": 2.463331555555556e-05, "loss": 0.0017, "step": 75500 }, { "epoch": 0.54, "learning_rate": 2.459776e-05, "loss": 0.0018, "step": 76000 }, { "epoch": 0.54, "learning_rate": 2.4562204444444443e-05, "loss": 0.0017, "step": 76500 }, { "epoch": 0.55, "learning_rate": 2.452664888888889e-05, "loss": 0.0017, "step": 77000 }, { "epoch": 0.55, "learning_rate": 2.4491093333333334e-05, "loss": 0.0018, "step": 77500 }, { "epoch": 0.55, "learning_rate": 2.445560888888889e-05, "loss": 0.0017, "step": 78000 }, { "epoch": 0.56, "learning_rate": 2.4420053333333335e-05, "loss": 0.0018, "step": 78500 }, { "epoch": 0.56, "learning_rate": 2.438449777777778e-05, "loss": 0.0017, "step": 79000 }, { "epoch": 0.57, "learning_rate": 2.4348942222222222e-05, "loss": 0.0018, "step": 79500 }, { "epoch": 0.57, "learning_rate": 2.4313386666666665e-05, "loss": 0.0017, "step": 80000 }, { "epoch": 0.57, "learning_rate": 2.4277902222222223e-05, "loss": 0.0018, "step": 80500 }, { "epoch": 0.58, "learning_rate": 2.424234666666667e-05, "loss": 0.0017, "step": 81000 }, { "epoch": 0.58, "learning_rate": 2.4206791111111114e-05, "loss": 0.0018, "step": 81500 }, { "epoch": 0.58, "learning_rate": 2.4171235555555556e-05, "loss": 0.0017, "step": 82000 }, { "epoch": 0.59, "learning_rate": 2.4135680000000002e-05, "loss": 0.0017, "step": 82500 }, { "epoch": 0.59, "learning_rate": 2.4100195555555557e-05, "loss": 0.0017, "step": 83000 }, { "epoch": 0.59, "learning_rate": 2.4064640000000003e-05, "loss": 0.0017, "step": 83500 }, { "epoch": 0.6, "learning_rate": 2.4029084444444445e-05, "loss": 0.0017, "step": 84000 }, { "epoch": 0.6, "learning_rate": 2.399352888888889e-05, "loss": 0.0016, "step": 84500 }, { "epoch": 0.6, "learning_rate": 2.3958044444444446e-05, "loss": 0.0017, "step": 85000 }, { "epoch": 0.61, "learning_rate": 2.3922488888888888e-05, "loss": 0.0017, "step": 85500 }, { "epoch": 0.61, "learning_rate": 2.3886933333333333e-05, "loss": 0.0017, "step": 86000 }, { "epoch": 0.62, "learning_rate": 2.385137777777778e-05, "loss": 0.0017, "step": 86500 }, { "epoch": 0.62, "learning_rate": 2.3815822222222224e-05, "loss": 0.0017, "step": 87000 }, { "epoch": 0.62, "learning_rate": 2.3780337777777776e-05, "loss": 0.0017, "step": 87500 }, { "epoch": 0.63, "learning_rate": 2.3744782222222222e-05, "loss": 0.0016, "step": 88000 }, { "epoch": 0.63, "learning_rate": 2.3709226666666667e-05, "loss": 0.0017, "step": 88500 }, { "epoch": 0.63, "learning_rate": 2.3673671111111113e-05, "loss": 0.0017, "step": 89000 }, { "epoch": 0.64, "learning_rate": 2.3638186666666665e-05, "loss": 0.0016, "step": 89500 }, { "epoch": 0.64, "learning_rate": 2.360263111111111e-05, "loss": 0.0017, "step": 90000 }, { "epoch": 0.64, "learning_rate": 2.3567075555555556e-05, "loss": 0.0017, "step": 90500 }, { "epoch": 0.65, "learning_rate": 2.353152e-05, "loss": 0.0017, "step": 91000 }, { "epoch": 0.65, "learning_rate": 2.3495964444444443e-05, "loss": 0.0016, "step": 91500 }, { "epoch": 0.65, "learning_rate": 2.346040888888889e-05, "loss": 0.0017, "step": 92000 }, { "epoch": 0.66, "learning_rate": 2.3424924444444448e-05, "loss": 0.0017, "step": 92500 }, { "epoch": 0.66, "learning_rate": 2.3389368888888886e-05, "loss": 0.0017, "step": 93000 }, { "epoch": 0.66, "learning_rate": 2.3353813333333332e-05, "loss": 0.0017, "step": 93500 }, { "epoch": 0.67, "learning_rate": 2.3318257777777777e-05, "loss": 0.0016, "step": 94000 }, { "epoch": 0.67, "learning_rate": 2.3282702222222223e-05, "loss": 0.0017, "step": 94500 }, { "epoch": 0.68, "learning_rate": 2.3247146666666665e-05, "loss": 0.0016, "step": 95000 }, { "epoch": 0.68, "learning_rate": 2.3211662222222224e-05, "loss": 0.0016, "step": 95500 }, { "epoch": 0.68, "learning_rate": 2.317610666666667e-05, "loss": 0.0017, "step": 96000 }, { "epoch": 0.69, "learning_rate": 2.314055111111111e-05, "loss": 0.0017, "step": 96500 }, { "epoch": 0.69, "learning_rate": 2.3104995555555554e-05, "loss": 0.0017, "step": 97000 }, { "epoch": 0.69, "learning_rate": 2.306944e-05, "loss": 0.0017, "step": 97500 }, { "epoch": 0.7, "learning_rate": 2.3033955555555558e-05, "loss": 0.0017, "step": 98000 }, { "epoch": 0.7, "learning_rate": 2.2998400000000003e-05, "loss": 0.0016, "step": 98500 }, { "epoch": 0.7, "learning_rate": 2.2962844444444445e-05, "loss": 0.0016, "step": 99000 }, { "epoch": 0.71, "learning_rate": 2.292728888888889e-05, "loss": 0.0016, "step": 99500 }, { "epoch": 0.71, "learning_rate": 2.2891733333333333e-05, "loss": 0.0016, "step": 100000 }, { "epoch": 0.71, "learning_rate": 2.2856177777777775e-05, "loss": 0.0016, "step": 100500 }, { "epoch": 0.72, "learning_rate": 2.282062222222222e-05, "loss": 0.0017, "step": 101000 }, { "epoch": 0.72, "learning_rate": 2.2785066666666666e-05, "loss": 0.0016, "step": 101500 }, { "epoch": 0.73, "learning_rate": 2.2749582222222225e-05, "loss": 0.0016, "step": 102000 }, { "epoch": 0.73, "learning_rate": 2.2714026666666667e-05, "loss": 0.0016, "step": 102500 }, { "epoch": 0.73, "learning_rate": 2.2678471111111113e-05, "loss": 0.0016, "step": 103000 }, { "epoch": 0.74, "learning_rate": 2.2642986666666668e-05, "loss": 0.0016, "step": 103500 }, { "epoch": 0.74, "learning_rate": 2.2607431111111113e-05, "loss": 0.0016, "step": 104000 }, { "epoch": 0.74, "learning_rate": 2.2571875555555556e-05, "loss": 0.0017, "step": 104500 }, { "epoch": 0.75, "learning_rate": 2.253632e-05, "loss": 0.0016, "step": 105000 }, { "epoch": 0.75, "learning_rate": 2.2500764444444447e-05, "loss": 0.0016, "step": 105500 }, { "epoch": 0.75, "learning_rate": 2.2465208888888892e-05, "loss": 0.0016, "step": 106000 }, { "epoch": 0.76, "learning_rate": 2.2429653333333334e-05, "loss": 0.0016, "step": 106500 }, { "epoch": 0.76, "learning_rate": 2.239409777777778e-05, "loss": 0.0016, "step": 107000 }, { "epoch": 0.76, "learning_rate": 2.2358613333333335e-05, "loss": 0.0016, "step": 107500 }, { "epoch": 0.77, "learning_rate": 2.2323057777777777e-05, "loss": 0.0016, "step": 108000 }, { "epoch": 0.77, "learning_rate": 2.2287573333333332e-05, "loss": 0.0016, "step": 108500 }, { "epoch": 0.78, "learning_rate": 2.2252017777777778e-05, "loss": 0.0016, "step": 109000 }, { "epoch": 0.78, "learning_rate": 2.2216462222222224e-05, "loss": 0.0016, "step": 109500 }, { "epoch": 0.78, "learning_rate": 2.2180906666666666e-05, "loss": 0.0016, "step": 110000 }, { "epoch": 0.79, "learning_rate": 2.214535111111111e-05, "loss": 0.0016, "step": 110500 }, { "epoch": 0.79, "learning_rate": 2.2109866666666666e-05, "loss": 0.0016, "step": 111000 }, { "epoch": 0.79, "learning_rate": 2.2074311111111112e-05, "loss": 0.0016, "step": 111500 }, { "epoch": 0.8, "learning_rate": 2.2038755555555554e-05, "loss": 0.0016, "step": 112000 }, { "epoch": 0.8, "learning_rate": 2.20032e-05, "loss": 0.0016, "step": 112500 }, { "epoch": 0.8, "learning_rate": 2.1967644444444445e-05, "loss": 0.0016, "step": 113000 }, { "epoch": 0.81, "learning_rate": 2.1932160000000004e-05, "loss": 0.0016, "step": 113500 }, { "epoch": 0.81, "learning_rate": 2.1896604444444443e-05, "loss": 0.0015, "step": 114000 }, { "epoch": 0.81, "learning_rate": 2.1861048888888888e-05, "loss": 0.0016, "step": 114500 }, { "epoch": 0.82, "learning_rate": 2.1825493333333334e-05, "loss": 0.0016, "step": 115000 }, { "epoch": 0.82, "learning_rate": 2.178993777777778e-05, "loss": 0.0015, "step": 115500 }, { "epoch": 0.82, "learning_rate": 2.1754453333333334e-05, "loss": 0.0016, "step": 116000 }, { "epoch": 0.83, "learning_rate": 2.171889777777778e-05, "loss": 0.0016, "step": 116500 }, { "epoch": 0.83, "learning_rate": 2.1683342222222225e-05, "loss": 0.0016, "step": 117000 }, { "epoch": 0.84, "learning_rate": 2.1647786666666664e-05, "loss": 0.0016, "step": 117500 }, { "epoch": 0.84, "learning_rate": 2.1612302222222223e-05, "loss": 0.0016, "step": 118000 }, { "epoch": 0.84, "learning_rate": 2.157674666666667e-05, "loss": 0.0015, "step": 118500 }, { "epoch": 0.85, "learning_rate": 2.1541191111111114e-05, "loss": 0.0015, "step": 119000 }, { "epoch": 0.85, "learning_rate": 2.1505635555555556e-05, "loss": 0.0016, "step": 119500 }, { "epoch": 0.85, "learning_rate": 2.147008e-05, "loss": 0.0016, "step": 120000 }, { "epoch": 0.86, "learning_rate": 2.1434524444444447e-05, "loss": 0.0016, "step": 120500 }, { "epoch": 0.86, "learning_rate": 2.1399040000000002e-05, "loss": 0.0016, "step": 121000 }, { "epoch": 0.86, "learning_rate": 2.1363484444444445e-05, "loss": 0.0016, "step": 121500 }, { "epoch": 0.87, "learning_rate": 2.132792888888889e-05, "loss": 0.0015, "step": 122000 }, { "epoch": 0.87, "learning_rate": 2.1292373333333336e-05, "loss": 0.0015, "step": 122500 }, { "epoch": 0.87, "learning_rate": 2.1256888888888887e-05, "loss": 0.0016, "step": 123000 }, { "epoch": 0.88, "learning_rate": 2.1221333333333333e-05, "loss": 0.0016, "step": 123500 }, { "epoch": 0.88, "learning_rate": 2.118577777777778e-05, "loss": 0.0015, "step": 124000 }, { "epoch": 0.89, "learning_rate": 2.1150222222222224e-05, "loss": 0.0016, "step": 124500 }, { "epoch": 0.89, "learning_rate": 2.1114666666666666e-05, "loss": 0.0015, "step": 125000 }, { "epoch": 0.89, "learning_rate": 2.1079111111111112e-05, "loss": 0.0015, "step": 125500 }, { "epoch": 0.9, "learning_rate": 2.1043555555555557e-05, "loss": 0.0015, "step": 126000 }, { "epoch": 0.9, "learning_rate": 2.1008000000000003e-05, "loss": 0.0015, "step": 126500 }, { "epoch": 0.9, "learning_rate": 2.0972515555555555e-05, "loss": 0.0016, "step": 127000 }, { "epoch": 0.91, "learning_rate": 2.093703111111111e-05, "loss": 0.0016, "step": 127500 }, { "epoch": 0.91, "learning_rate": 2.0901475555555555e-05, "loss": 0.0016, "step": 128000 }, { "epoch": 0.91, "learning_rate": 2.086592e-05, "loss": 0.0015, "step": 128500 }, { "epoch": 0.92, "learning_rate": 2.0830364444444443e-05, "loss": 0.0015, "step": 129000 }, { "epoch": 0.92, "learning_rate": 2.079480888888889e-05, "loss": 0.0015, "step": 129500 }, { "epoch": 0.92, "learning_rate": 2.0759253333333334e-05, "loss": 0.0015, "step": 130000 }, { "epoch": 0.93, "learning_rate": 2.072369777777778e-05, "loss": 0.0015, "step": 130500 }, { "epoch": 0.93, "learning_rate": 2.068821333333333e-05, "loss": 0.0015, "step": 131000 }, { "epoch": 0.94, "learning_rate": 2.0652657777777777e-05, "loss": 0.0015, "step": 131500 }, { "epoch": 0.94, "learning_rate": 2.0617102222222223e-05, "loss": 0.0016, "step": 132000 }, { "epoch": 0.94, "learning_rate": 2.0581546666666665e-05, "loss": 0.0015, "step": 132500 }, { "epoch": 0.95, "learning_rate": 2.054599111111111e-05, "loss": 0.0015, "step": 133000 }, { "epoch": 0.95, "learning_rate": 2.051050666666667e-05, "loss": 0.0015, "step": 133500 }, { "epoch": 0.95, "learning_rate": 2.0474951111111114e-05, "loss": 0.0016, "step": 134000 }, { "epoch": 0.96, "learning_rate": 2.0439395555555553e-05, "loss": 0.0016, "step": 134500 }, { "epoch": 0.96, "learning_rate": 2.040384e-05, "loss": 0.0015, "step": 135000 }, { "epoch": 0.96, "learning_rate": 2.0368284444444444e-05, "loss": 0.0015, "step": 135500 }, { "epoch": 0.97, "learning_rate": 2.033272888888889e-05, "loss": 0.0015, "step": 136000 }, { "epoch": 0.97, "learning_rate": 2.0297173333333332e-05, "loss": 0.0015, "step": 136500 }, { "epoch": 0.97, "learning_rate": 2.0261617777777778e-05, "loss": 0.0015, "step": 137000 }, { "epoch": 0.98, "learning_rate": 2.0226133333333336e-05, "loss": 0.0015, "step": 137500 }, { "epoch": 0.98, "learning_rate": 2.0190577777777778e-05, "loss": 0.0015, "step": 138000 }, { "epoch": 0.98, "learning_rate": 2.015502222222222e-05, "loss": 0.0015, "step": 138500 }, { "epoch": 0.99, "learning_rate": 2.0119466666666666e-05, "loss": 0.0015, "step": 139000 }, { "epoch": 0.99, "learning_rate": 2.0084053333333334e-05, "loss": 0.0015, "step": 139500 }, { "epoch": 1.0, "learning_rate": 2.0048497777777776e-05, "loss": 0.0015, "step": 140000 }, { "epoch": 1.0, "learning_rate": 2.0012942222222222e-05, "loss": 0.0015, "step": 140500 }, { "epoch": 1.0, "learning_rate": 1.9977386666666668e-05, "loss": 0.0015, "step": 141000 }, { "epoch": 1.01, "learning_rate": 1.9941831111111113e-05, "loss": 0.0015, "step": 141500 }, { "epoch": 1.01, "learning_rate": 1.9906275555555555e-05, "loss": 0.0014, "step": 142000 }, { "epoch": 1.01, "learning_rate": 1.987072e-05, "loss": 0.0015, "step": 142500 }, { "epoch": 1.02, "learning_rate": 1.9835164444444446e-05, "loss": 0.0015, "step": 143000 }, { "epoch": 1.02, "learning_rate": 1.979968e-05, "loss": 0.0015, "step": 143500 }, { "epoch": 1.02, "learning_rate": 1.9764124444444444e-05, "loss": 0.0015, "step": 144000 }, { "epoch": 1.03, "learning_rate": 1.972856888888889e-05, "loss": 0.0015, "step": 144500 }, { "epoch": 1.03, "learning_rate": 1.9693013333333335e-05, "loss": 0.0014, "step": 145000 }, { "epoch": 1.03, "learning_rate": 1.965745777777778e-05, "loss": 0.0015, "step": 145500 }, { "epoch": 1.04, "learning_rate": 1.9621902222222222e-05, "loss": 0.0015, "step": 146000 }, { "epoch": 1.04, "learning_rate": 1.9586346666666668e-05, "loss": 0.0014, "step": 146500 }, { "epoch": 1.05, "learning_rate": 1.9550791111111113e-05, "loss": 0.0014, "step": 147000 }, { "epoch": 1.05, "learning_rate": 1.9515306666666665e-05, "loss": 0.0015, "step": 147500 }, { "epoch": 1.05, "learning_rate": 1.947982222222222e-05, "loss": 0.0014, "step": 148000 }, { "epoch": 1.06, "learning_rate": 1.9444266666666666e-05, "loss": 0.0015, "step": 148500 }, { "epoch": 1.06, "learning_rate": 1.940871111111111e-05, "loss": 0.0015, "step": 149000 }, { "epoch": 1.06, "learning_rate": 1.9373155555555554e-05, "loss": 0.0014, "step": 149500 }, { "epoch": 1.07, "learning_rate": 1.93376e-05, "loss": 0.0014, "step": 150000 }, { "epoch": 1.07, "learning_rate": 1.9302115555555558e-05, "loss": 0.0015, "step": 150500 }, { "epoch": 1.07, "learning_rate": 1.9266560000000003e-05, "loss": 0.0015, "step": 151000 }, { "epoch": 1.08, "learning_rate": 1.9231004444444442e-05, "loss": 0.0014, "step": 151500 }, { "epoch": 1.08, "learning_rate": 1.9195448888888888e-05, "loss": 0.0014, "step": 152000 }, { "epoch": 1.08, "learning_rate": 1.9159893333333333e-05, "loss": 0.0015, "step": 152500 }, { "epoch": 1.09, "learning_rate": 1.912433777777778e-05, "loss": 0.0015, "step": 153000 }, { "epoch": 1.09, "learning_rate": 1.908878222222222e-05, "loss": 0.0014, "step": 153500 }, { "epoch": 1.1, "learning_rate": 1.9053226666666667e-05, "loss": 0.0014, "step": 154000 }, { "epoch": 1.1, "learning_rate": 1.9017813333333335e-05, "loss": 0.0015, "step": 154500 }, { "epoch": 1.1, "learning_rate": 1.8982257777777777e-05, "loss": 0.0015, "step": 155000 }, { "epoch": 1.11, "learning_rate": 1.8946702222222223e-05, "loss": 0.0015, "step": 155500 }, { "epoch": 1.11, "learning_rate": 1.8911146666666668e-05, "loss": 0.0015, "step": 156000 }, { "epoch": 1.11, "learning_rate": 1.8875591111111114e-05, "loss": 0.0015, "step": 156500 }, { "epoch": 1.12, "learning_rate": 1.8840106666666665e-05, "loss": 0.0014, "step": 157000 }, { "epoch": 1.12, "learning_rate": 1.880455111111111e-05, "loss": 0.0015, "step": 157500 }, { "epoch": 1.12, "learning_rate": 1.8768995555555556e-05, "loss": 0.0014, "step": 158000 }, { "epoch": 1.13, "learning_rate": 1.8733440000000002e-05, "loss": 0.0014, "step": 158500 }, { "epoch": 1.13, "learning_rate": 1.8697884444444444e-05, "loss": 0.0015, "step": 159000 }, { "epoch": 1.13, "learning_rate": 1.866232888888889e-05, "loss": 0.0015, "step": 159500 }, { "epoch": 1.14, "learning_rate": 1.8626773333333335e-05, "loss": 0.0015, "step": 160000 }, { "epoch": 1.14, "learning_rate": 1.859121777777778e-05, "loss": 0.0014, "step": 160500 }, { "epoch": 1.14, "learning_rate": 1.8555733333333333e-05, "loss": 0.0015, "step": 161000 }, { "epoch": 1.15, "learning_rate": 1.8520177777777778e-05, "loss": 0.0015, "step": 161500 }, { "epoch": 1.15, "learning_rate": 1.8484622222222224e-05, "loss": 0.0014, "step": 162000 }, { "epoch": 1.16, "learning_rate": 1.8449066666666666e-05, "loss": 0.0014, "step": 162500 }, { "epoch": 1.16, "learning_rate": 1.841358222222222e-05, "loss": 0.0014, "step": 163000 }, { "epoch": 1.16, "learning_rate": 1.8378026666666667e-05, "loss": 0.0014, "step": 163500 }, { "epoch": 1.17, "learning_rate": 1.8342471111111112e-05, "loss": 0.0014, "step": 164000 }, { "epoch": 1.17, "learning_rate": 1.8306915555555554e-05, "loss": 0.0014, "step": 164500 }, { "epoch": 1.17, "learning_rate": 1.827136e-05, "loss": 0.0014, "step": 165000 }, { "epoch": 1.18, "learning_rate": 1.8235804444444445e-05, "loss": 0.0014, "step": 165500 }, { "epoch": 1.18, "learning_rate": 1.820032e-05, "loss": 0.0014, "step": 166000 }, { "epoch": 1.18, "learning_rate": 1.8164764444444443e-05, "loss": 0.0014, "step": 166500 }, { "epoch": 1.19, "learning_rate": 1.8129208888888888e-05, "loss": 0.0014, "step": 167000 }, { "epoch": 1.19, "learning_rate": 1.8093653333333334e-05, "loss": 0.0014, "step": 167500 }, { "epoch": 1.19, "learning_rate": 1.805809777777778e-05, "loss": 0.0014, "step": 168000 }, { "epoch": 1.2, "learning_rate": 1.802254222222222e-05, "loss": 0.0014, "step": 168500 }, { "epoch": 1.2, "learning_rate": 1.7987057777777777e-05, "loss": 0.0014, "step": 169000 }, { "epoch": 1.21, "learning_rate": 1.7951502222222222e-05, "loss": 0.0014, "step": 169500 }, { "epoch": 1.21, "learning_rate": 1.7915946666666664e-05, "loss": 0.0014, "step": 170000 }, { "epoch": 1.21, "learning_rate": 1.788039111111111e-05, "loss": 0.0014, "step": 170500 }, { "epoch": 1.22, "learning_rate": 1.7844835555555555e-05, "loss": 0.0015, "step": 171000 }, { "epoch": 1.22, "learning_rate": 1.780928e-05, "loss": 0.0014, "step": 171500 }, { "epoch": 1.22, "learning_rate": 1.7773724444444443e-05, "loss": 0.0014, "step": 172000 }, { "epoch": 1.23, "learning_rate": 1.773816888888889e-05, "loss": 0.0015, "step": 172500 }, { "epoch": 1.23, "learning_rate": 1.7702684444444444e-05, "loss": 0.0014, "step": 173000 }, { "epoch": 1.23, "learning_rate": 1.766712888888889e-05, "loss": 0.0014, "step": 173500 }, { "epoch": 1.24, "learning_rate": 1.763157333333333e-05, "loss": 0.0014, "step": 174000 }, { "epoch": 1.24, "learning_rate": 1.7596017777777777e-05, "loss": 0.0014, "step": 174500 }, { "epoch": 1.24, "learning_rate": 1.7560533333333336e-05, "loss": 0.0014, "step": 175000 }, { "epoch": 1.25, "learning_rate": 1.752497777777778e-05, "loss": 0.0014, "step": 175500 }, { "epoch": 1.25, "learning_rate": 1.748942222222222e-05, "loss": 0.0014, "step": 176000 }, { "epoch": 1.26, "learning_rate": 1.745393777777778e-05, "loss": 0.0014, "step": 176500 }, { "epoch": 1.26, "learning_rate": 1.7418382222222224e-05, "loss": 0.0014, "step": 177000 }, { "epoch": 1.26, "learning_rate": 1.7382826666666666e-05, "loss": 0.0014, "step": 177500 }, { "epoch": 1.27, "learning_rate": 1.7347271111111112e-05, "loss": 0.0014, "step": 178000 }, { "epoch": 1.27, "learning_rate": 1.7311715555555557e-05, "loss": 0.0014, "step": 178500 }, { "epoch": 1.27, "learning_rate": 1.7276231111111113e-05, "loss": 0.0014, "step": 179000 }, { "epoch": 1.28, "learning_rate": 1.7240675555555555e-05, "loss": 0.0014, "step": 179500 }, { "epoch": 1.28, "learning_rate": 1.720512e-05, "loss": 0.0014, "step": 180000 }, { "epoch": 1.28, "learning_rate": 1.7169564444444446e-05, "loss": 0.0014, "step": 180500 }, { "epoch": 1.29, "learning_rate": 1.713400888888889e-05, "loss": 0.0014, "step": 181000 }, { "epoch": 1.29, "learning_rate": 1.7098453333333334e-05, "loss": 0.0014, "step": 181500 }, { "epoch": 1.29, "learning_rate": 1.706296888888889e-05, "loss": 0.0014, "step": 182000 }, { "epoch": 1.3, "learning_rate": 1.7027413333333334e-05, "loss": 0.0014, "step": 182500 }, { "epoch": 1.3, "learning_rate": 1.699185777777778e-05, "loss": 0.0014, "step": 183000 }, { "epoch": 1.3, "learning_rate": 1.6956302222222222e-05, "loss": 0.0014, "step": 183500 }, { "epoch": 1.31, "learning_rate": 1.6920746666666668e-05, "loss": 0.0014, "step": 184000 }, { "epoch": 1.31, "learning_rate": 1.6885191111111113e-05, "loss": 0.0014, "step": 184500 }, { "epoch": 1.32, "learning_rate": 1.6849706666666665e-05, "loss": 0.0014, "step": 185000 }, { "epoch": 1.32, "learning_rate": 1.681415111111111e-05, "loss": 0.0014, "step": 185500 }, { "epoch": 1.32, "learning_rate": 1.6778595555555556e-05, "loss": 0.0014, "step": 186000 }, { "epoch": 1.33, "learning_rate": 1.674304e-05, "loss": 0.0014, "step": 186500 }, { "epoch": 1.33, "learning_rate": 1.6707484444444444e-05, "loss": 0.0014, "step": 187000 }, { "epoch": 1.33, "learning_rate": 1.667192888888889e-05, "loss": 0.0013, "step": 187500 }, { "epoch": 1.34, "learning_rate": 1.6636373333333335e-05, "loss": 0.0014, "step": 188000 }, { "epoch": 1.34, "learning_rate": 1.660081777777778e-05, "loss": 0.0014, "step": 188500 }, { "epoch": 1.34, "learning_rate": 1.6565333333333332e-05, "loss": 0.0014, "step": 189000 }, { "epoch": 1.35, "learning_rate": 1.6529777777777778e-05, "loss": 0.0014, "step": 189500 }, { "epoch": 1.35, "learning_rate": 1.6494222222222223e-05, "loss": 0.0014, "step": 190000 }, { "epoch": 1.35, "learning_rate": 1.645866666666667e-05, "loss": 0.0014, "step": 190500 }, { "epoch": 1.36, "learning_rate": 1.642318222222222e-05, "loss": 0.0014, "step": 191000 }, { "epoch": 1.36, "learning_rate": 1.6387626666666666e-05, "loss": 0.0014, "step": 191500 }, { "epoch": 1.37, "learning_rate": 1.635207111111111e-05, "loss": 0.0014, "step": 192000 }, { "epoch": 1.37, "learning_rate": 1.6316515555555557e-05, "loss": 0.0014, "step": 192500 }, { "epoch": 1.37, "learning_rate": 1.628103111111111e-05, "loss": 0.0014, "step": 193000 }, { "epoch": 1.38, "learning_rate": 1.6245475555555555e-05, "loss": 0.0014, "step": 193500 }, { "epoch": 1.38, "learning_rate": 1.620992e-05, "loss": 0.0014, "step": 194000 }, { "epoch": 1.38, "learning_rate": 1.6174364444444442e-05, "loss": 0.0014, "step": 194500 }, { "epoch": 1.39, "learning_rate": 1.613888e-05, "loss": 0.0013, "step": 195000 }, { "epoch": 1.39, "learning_rate": 1.6103395555555556e-05, "loss": 0.0014, "step": 195500 }, { "epoch": 1.39, "learning_rate": 1.606784e-05, "loss": 0.0014, "step": 196000 }, { "epoch": 1.4, "learning_rate": 1.6032284444444444e-05, "loss": 0.0014, "step": 196500 }, { "epoch": 1.4, "learning_rate": 1.599672888888889e-05, "loss": 0.0013, "step": 197000 }, { "epoch": 1.4, "learning_rate": 1.5961173333333335e-05, "loss": 0.0013, "step": 197500 }, { "epoch": 1.41, "learning_rate": 1.592561777777778e-05, "loss": 0.0013, "step": 198000 }, { "epoch": 1.41, "learning_rate": 1.5890062222222223e-05, "loss": 0.0014, "step": 198500 }, { "epoch": 1.42, "learning_rate": 1.5854506666666668e-05, "loss": 0.0014, "step": 199000 }, { "epoch": 1.42, "learning_rate": 1.5819022222222223e-05, "loss": 0.0014, "step": 199500 }, { "epoch": 1.42, "learning_rate": 1.5783466666666665e-05, "loss": 0.0014, "step": 200000 }, { "epoch": 1.43, "learning_rate": 1.574791111111111e-05, "loss": 0.0013, "step": 200500 }, { "epoch": 1.43, "learning_rate": 1.5712355555555557e-05, "loss": 0.0014, "step": 201000 }, { "epoch": 1.43, "learning_rate": 1.5676871111111112e-05, "loss": 0.0013, "step": 201500 }, { "epoch": 1.44, "learning_rate": 1.5641315555555554e-05, "loss": 0.0014, "step": 202000 }, { "epoch": 1.44, "learning_rate": 1.560576e-05, "loss": 0.0014, "step": 202500 }, { "epoch": 1.44, "learning_rate": 1.5570204444444445e-05, "loss": 0.0014, "step": 203000 }, { "epoch": 1.45, "learning_rate": 1.553464888888889e-05, "loss": 0.0014, "step": 203500 }, { "epoch": 1.45, "learning_rate": 1.5499093333333333e-05, "loss": 0.0013, "step": 204000 }, { "epoch": 1.45, "learning_rate": 1.5463537777777778e-05, "loss": 0.0014, "step": 204500 }, { "epoch": 1.46, "learning_rate": 1.5427982222222224e-05, "loss": 0.0014, "step": 205000 }, { "epoch": 1.46, "learning_rate": 1.539249777777778e-05, "loss": 0.0013, "step": 205500 }, { "epoch": 1.46, "learning_rate": 1.535694222222222e-05, "loss": 0.0014, "step": 206000 }, { "epoch": 1.47, "learning_rate": 1.5321386666666667e-05, "loss": 0.0014, "step": 206500 }, { "epoch": 1.47, "learning_rate": 1.5285902222222222e-05, "loss": 0.0014, "step": 207000 }, { "epoch": 1.48, "learning_rate": 1.5250346666666666e-05, "loss": 0.0013, "step": 207500 }, { "epoch": 1.48, "learning_rate": 1.521479111111111e-05, "loss": 0.0013, "step": 208000 }, { "epoch": 1.48, "learning_rate": 1.5179235555555555e-05, "loss": 0.0014, "step": 208500 }, { "epoch": 1.49, "learning_rate": 1.5143679999999999e-05, "loss": 0.0013, "step": 209000 }, { "epoch": 1.49, "learning_rate": 1.5108124444444445e-05, "loss": 0.0014, "step": 209500 }, { "epoch": 1.49, "learning_rate": 1.5072640000000001e-05, "loss": 0.0014, "step": 210000 }, { "epoch": 1.5, "learning_rate": 1.5037084444444444e-05, "loss": 0.0013, "step": 210500 }, { "epoch": 1.5, "learning_rate": 1.5001528888888887e-05, "loss": 0.0014, "step": 211000 }, { "epoch": 1.5, "learning_rate": 1.4965973333333335e-05, "loss": 0.0014, "step": 211500 }, { "epoch": 1.51, "learning_rate": 1.4930417777777779e-05, "loss": 0.0013, "step": 212000 }, { "epoch": 1.51, "learning_rate": 1.4894862222222222e-05, "loss": 0.0013, "step": 212500 }, { "epoch": 1.51, "learning_rate": 1.4859306666666666e-05, "loss": 0.0013, "step": 213000 }, { "epoch": 1.52, "learning_rate": 1.4823751111111112e-05, "loss": 0.0014, "step": 213500 }, { "epoch": 1.52, "learning_rate": 1.4788266666666667e-05, "loss": 0.0013, "step": 214000 }, { "epoch": 1.53, "learning_rate": 1.4752711111111112e-05, "loss": 0.0013, "step": 214500 }, { "epoch": 1.53, "learning_rate": 1.4717155555555556e-05, "loss": 0.0014, "step": 215000 }, { "epoch": 1.53, "learning_rate": 1.4681600000000002e-05, "loss": 0.0013, "step": 215500 }, { "epoch": 1.54, "learning_rate": 1.4646115555555555e-05, "loss": 0.0014, "step": 216000 }, { "epoch": 1.54, "learning_rate": 1.461056e-05, "loss": 0.0013, "step": 216500 }, { "epoch": 1.54, "learning_rate": 1.4575004444444445e-05, "loss": 0.0014, "step": 217000 }, { "epoch": 1.55, "learning_rate": 1.4539448888888889e-05, "loss": 0.0013, "step": 217500 }, { "epoch": 1.55, "learning_rate": 1.4503964444444446e-05, "loss": 0.0013, "step": 218000 }, { "epoch": 1.55, "learning_rate": 1.446840888888889e-05, "loss": 0.0013, "step": 218500 }, { "epoch": 1.56, "learning_rate": 1.4432853333333333e-05, "loss": 0.0013, "step": 219000 }, { "epoch": 1.56, "learning_rate": 1.4397297777777777e-05, "loss": 0.0013, "step": 219500 }, { "epoch": 1.56, "learning_rate": 1.4361813333333334e-05, "loss": 0.0013, "step": 220000 }, { "epoch": 1.57, "learning_rate": 1.4326257777777778e-05, "loss": 0.0014, "step": 220500 }, { "epoch": 1.57, "learning_rate": 1.4290702222222223e-05, "loss": 0.0014, "step": 221000 }, { "epoch": 1.58, "learning_rate": 1.4255217777777779e-05, "loss": 0.0013, "step": 221500 }, { "epoch": 1.58, "learning_rate": 1.4219662222222224e-05, "loss": 0.0013, "step": 222000 }, { "epoch": 1.58, "learning_rate": 1.4184106666666666e-05, "loss": 0.0014, "step": 222500 }, { "epoch": 1.59, "learning_rate": 1.4148551111111112e-05, "loss": 0.0013, "step": 223000 }, { "epoch": 1.59, "learning_rate": 1.4113066666666667e-05, "loss": 0.0014, "step": 223500 }, { "epoch": 1.59, "learning_rate": 1.4077511111111111e-05, "loss": 0.0013, "step": 224000 }, { "epoch": 1.6, "learning_rate": 1.4041955555555556e-05, "loss": 0.0013, "step": 224500 }, { "epoch": 1.6, "learning_rate": 1.40064e-05, "loss": 0.0013, "step": 225000 }, { "epoch": 1.6, "learning_rate": 1.3970844444444446e-05, "loss": 0.0013, "step": 225500 }, { "epoch": 1.61, "learning_rate": 1.3935288888888888e-05, "loss": 0.0013, "step": 226000 }, { "epoch": 1.61, "learning_rate": 1.3899733333333333e-05, "loss": 0.0013, "step": 226500 }, { "epoch": 1.61, "learning_rate": 1.3864177777777777e-05, "loss": 0.0013, "step": 227000 }, { "epoch": 1.62, "learning_rate": 1.3828622222222223e-05, "loss": 0.0013, "step": 227500 }, { "epoch": 1.62, "learning_rate": 1.3793137777777778e-05, "loss": 0.0013, "step": 228000 }, { "epoch": 1.62, "learning_rate": 1.3757582222222224e-05, "loss": 0.0013, "step": 228500 }, { "epoch": 1.63, "learning_rate": 1.3722097777777777e-05, "loss": 0.0013, "step": 229000 }, { "epoch": 1.63, "learning_rate": 1.3686542222222223e-05, "loss": 0.0013, "step": 229500 }, { "epoch": 1.64, "learning_rate": 1.3650986666666667e-05, "loss": 0.0013, "step": 230000 }, { "epoch": 1.64, "learning_rate": 1.3615431111111112e-05, "loss": 0.0013, "step": 230500 }, { "epoch": 1.64, "learning_rate": 1.3579875555555556e-05, "loss": 0.0013, "step": 231000 }, { "epoch": 1.65, "learning_rate": 1.354432e-05, "loss": 0.0014, "step": 231500 }, { "epoch": 1.65, "learning_rate": 1.3508764444444445e-05, "loss": 0.0013, "step": 232000 }, { "epoch": 1.65, "learning_rate": 1.347320888888889e-05, "loss": 0.0013, "step": 232500 }, { "epoch": 1.66, "learning_rate": 1.3437724444444444e-05, "loss": 0.0013, "step": 233000 }, { "epoch": 1.66, "learning_rate": 1.3402168888888888e-05, "loss": 0.0013, "step": 233500 }, { "epoch": 1.66, "learning_rate": 1.3366613333333334e-05, "loss": 0.0013, "step": 234000 }, { "epoch": 1.67, "learning_rate": 1.3331057777777778e-05, "loss": 0.0013, "step": 234500 }, { "epoch": 1.67, "learning_rate": 1.3295573333333335e-05, "loss": 0.0013, "step": 235000 }, { "epoch": 1.67, "learning_rate": 1.3260017777777778e-05, "loss": 0.0013, "step": 235500 }, { "epoch": 1.68, "learning_rate": 1.3224462222222224e-05, "loss": 0.0013, "step": 236000 }, { "epoch": 1.68, "learning_rate": 1.3188906666666666e-05, "loss": 0.0013, "step": 236500 }, { "epoch": 1.69, "learning_rate": 1.3153422222222223e-05, "loss": 0.0013, "step": 237000 }, { "epoch": 1.69, "learning_rate": 1.3117866666666667e-05, "loss": 0.0013, "step": 237500 }, { "epoch": 1.69, "learning_rate": 1.3082311111111112e-05, "loss": 0.0013, "step": 238000 }, { "epoch": 1.7, "learning_rate": 1.3046755555555556e-05, "loss": 0.0013, "step": 238500 }, { "epoch": 1.7, "learning_rate": 1.3011271111111111e-05, "loss": 0.0013, "step": 239000 }, { "epoch": 1.7, "learning_rate": 1.2975715555555555e-05, "loss": 0.0013, "step": 239500 }, { "epoch": 1.71, "learning_rate": 1.294023111111111e-05, "loss": 0.0013, "step": 240000 }, { "epoch": 1.71, "learning_rate": 1.2904675555555556e-05, "loss": 0.0013, "step": 240500 }, { "epoch": 1.71, "learning_rate": 1.286912e-05, "loss": 0.0013, "step": 241000 }, { "epoch": 1.72, "learning_rate": 1.2833564444444445e-05, "loss": 0.0013, "step": 241500 }, { "epoch": 1.72, "learning_rate": 1.279800888888889e-05, "loss": 0.0013, "step": 242000 }, { "epoch": 1.72, "learning_rate": 1.2762453333333335e-05, "loss": 0.0013, "step": 242500 }, { "epoch": 1.73, "learning_rate": 1.2726897777777777e-05, "loss": 0.0013, "step": 243000 }, { "epoch": 1.73, "learning_rate": 1.2691342222222222e-05, "loss": 0.0013, "step": 243500 }, { "epoch": 1.74, "learning_rate": 1.2655857777777778e-05, "loss": 0.0013, "step": 244000 }, { "epoch": 1.74, "learning_rate": 1.2620302222222223e-05, "loss": 0.0013, "step": 244500 }, { "epoch": 1.74, "learning_rate": 1.2584746666666667e-05, "loss": 0.0013, "step": 245000 }, { "epoch": 1.75, "learning_rate": 1.2549191111111113e-05, "loss": 0.0013, "step": 245500 }, { "epoch": 1.75, "learning_rate": 1.2513706666666668e-05, "loss": 0.0013, "step": 246000 }, { "epoch": 1.75, "learning_rate": 1.247815111111111e-05, "loss": 0.0013, "step": 246500 }, { "epoch": 1.76, "learning_rate": 1.2442595555555556e-05, "loss": 0.0013, "step": 247000 }, { "epoch": 1.76, "learning_rate": 1.240704e-05, "loss": 0.0013, "step": 247500 }, { "epoch": 1.76, "learning_rate": 1.2371555555555556e-05, "loss": 0.0013, "step": 248000 }, { "epoch": 1.77, "learning_rate": 1.2336e-05, "loss": 0.0013, "step": 248500 }, { "epoch": 1.77, "learning_rate": 1.2300444444444446e-05, "loss": 0.0013, "step": 249000 }, { "epoch": 1.77, "learning_rate": 1.226488888888889e-05, "loss": 0.0013, "step": 249500 }, { "epoch": 1.78, "learning_rate": 1.2229404444444445e-05, "loss": 0.0013, "step": 250000 }, { "epoch": 1.78, "learning_rate": 1.2193848888888889e-05, "loss": 0.0013, "step": 250500 }, { "epoch": 1.78, "learning_rate": 1.2158293333333334e-05, "loss": 0.0013, "step": 251000 }, { "epoch": 1.79, "learning_rate": 1.2122737777777778e-05, "loss": 0.0013, "step": 251500 }, { "epoch": 1.79, "learning_rate": 1.2087253333333335e-05, "loss": 0.0013, "step": 252000 }, { "epoch": 1.8, "learning_rate": 1.2051697777777779e-05, "loss": 0.0013, "step": 252500 }, { "epoch": 1.8, "learning_rate": 1.2016142222222221e-05, "loss": 0.0013, "step": 253000 }, { "epoch": 1.8, "learning_rate": 1.1980586666666666e-05, "loss": 0.0013, "step": 253500 }, { "epoch": 1.81, "learning_rate": 1.1945102222222222e-05, "loss": 0.0013, "step": 254000 }, { "epoch": 1.81, "learning_rate": 1.1909546666666667e-05, "loss": 0.0013, "step": 254500 }, { "epoch": 1.81, "learning_rate": 1.1873991111111111e-05, "loss": 0.0013, "step": 255000 }, { "epoch": 1.82, "learning_rate": 1.1838506666666668e-05, "loss": 0.0013, "step": 255500 }, { "epoch": 1.82, "learning_rate": 1.1802951111111112e-05, "loss": 0.0013, "step": 256000 }, { "epoch": 1.82, "learning_rate": 1.1767395555555556e-05, "loss": 0.0013, "step": 256500 }, { "epoch": 1.83, "learning_rate": 1.173184e-05, "loss": 0.0013, "step": 257000 }, { "epoch": 1.83, "learning_rate": 1.1696284444444445e-05, "loss": 0.0013, "step": 257500 }, { "epoch": 1.83, "learning_rate": 1.1660728888888889e-05, "loss": 0.0013, "step": 258000 }, { "epoch": 1.84, "learning_rate": 1.1625173333333334e-05, "loss": 0.0013, "step": 258500 }, { "epoch": 1.84, "learning_rate": 1.1589617777777778e-05, "loss": 0.0013, "step": 259000 }, { "epoch": 1.85, "learning_rate": 1.1554133333333335e-05, "loss": 0.0013, "step": 259500 }, { "epoch": 1.85, "learning_rate": 1.1518577777777777e-05, "loss": 0.0013, "step": 260000 }, { "epoch": 1.85, "learning_rate": 1.1483022222222221e-05, "loss": 0.0013, "step": 260500 }, { "epoch": 1.86, "learning_rate": 1.1447466666666667e-05, "loss": 0.0013, "step": 261000 }, { "epoch": 1.86, "learning_rate": 1.141191111111111e-05, "loss": 0.0012, "step": 261500 }, { "epoch": 1.86, "learning_rate": 1.1376355555555556e-05, "loss": 0.0013, "step": 262000 }, { "epoch": 1.87, "learning_rate": 1.13408e-05, "loss": 0.0012, "step": 262500 }, { "epoch": 1.87, "learning_rate": 1.1305244444444445e-05, "loss": 0.0013, "step": 263000 }, { "epoch": 1.87, "learning_rate": 1.1269759999999999e-05, "loss": 0.0013, "step": 263500 }, { "epoch": 1.88, "learning_rate": 1.1234204444444445e-05, "loss": 0.0013, "step": 264000 }, { "epoch": 1.88, "learning_rate": 1.1198648888888888e-05, "loss": 0.0013, "step": 264500 }, { "epoch": 1.88, "learning_rate": 1.1163093333333334e-05, "loss": 0.0013, "step": 265000 }, { "epoch": 1.89, "learning_rate": 1.1127608888888889e-05, "loss": 0.0012, "step": 265500 }, { "epoch": 1.89, "learning_rate": 1.1092053333333335e-05, "loss": 0.0013, "step": 266000 }, { "epoch": 1.9, "learning_rate": 1.1056497777777778e-05, "loss": 0.0013, "step": 266500 }, { "epoch": 1.9, "learning_rate": 1.1020942222222222e-05, "loss": 0.0012, "step": 267000 }, { "epoch": 1.9, "learning_rate": 1.0985457777777778e-05, "loss": 0.0013, "step": 267500 }, { "epoch": 1.91, "learning_rate": 1.0949902222222223e-05, "loss": 0.0013, "step": 268000 }, { "epoch": 1.91, "learning_rate": 1.0914346666666667e-05, "loss": 0.0013, "step": 268500 }, { "epoch": 1.91, "learning_rate": 1.087879111111111e-05, "loss": 0.0013, "step": 269000 }, { "epoch": 1.92, "learning_rate": 1.0843306666666668e-05, "loss": 0.0013, "step": 269500 }, { "epoch": 1.92, "learning_rate": 1.0807751111111112e-05, "loss": 0.0013, "step": 270000 }, { "epoch": 1.92, "learning_rate": 1.0772195555555555e-05, "loss": 0.0012, "step": 270500 }, { "epoch": 1.93, "learning_rate": 1.073664e-05, "loss": 0.0013, "step": 271000 }, { "epoch": 1.93, "learning_rate": 1.0701155555555556e-05, "loss": 0.0013, "step": 271500 }, { "epoch": 1.93, "learning_rate": 1.06656e-05, "loss": 0.0013, "step": 272000 }, { "epoch": 1.94, "learning_rate": 1.0630044444444446e-05, "loss": 0.0012, "step": 272500 }, { "epoch": 1.94, "learning_rate": 1.059448888888889e-05, "loss": 0.0013, "step": 273000 }, { "epoch": 1.94, "learning_rate": 1.0559004444444445e-05, "loss": 0.0013, "step": 273500 }, { "epoch": 1.95, "learning_rate": 1.0523448888888888e-05, "loss": 0.0013, "step": 274000 }, { "epoch": 1.95, "learning_rate": 1.0487893333333334e-05, "loss": 0.0013, "step": 274500 }, { "epoch": 1.96, "learning_rate": 1.0452337777777778e-05, "loss": 0.0013, "step": 275000 }, { "epoch": 1.96, "learning_rate": 1.0416853333333333e-05, "loss": 0.0013, "step": 275500 }, { "epoch": 1.96, "learning_rate": 1.0381297777777779e-05, "loss": 0.0013, "step": 276000 }, { "epoch": 1.97, "learning_rate": 1.0345742222222222e-05, "loss": 0.0013, "step": 276500 }, { "epoch": 1.97, "learning_rate": 1.0310186666666668e-05, "loss": 0.0013, "step": 277000 }, { "epoch": 1.97, "learning_rate": 1.0274702222222222e-05, "loss": 0.0013, "step": 277500 }, { "epoch": 1.98, "learning_rate": 1.0239146666666667e-05, "loss": 0.0012, "step": 278000 }, { "epoch": 1.98, "learning_rate": 1.0203591111111111e-05, "loss": 0.0012, "step": 278500 }, { "epoch": 1.98, "learning_rate": 1.0168035555555556e-05, "loss": 0.0013, "step": 279000 }, { "epoch": 1.99, "learning_rate": 1.0132551111111112e-05, "loss": 0.0013, "step": 279500 }, { "epoch": 1.99, "learning_rate": 1.0096995555555557e-05, "loss": 0.0013, "step": 280000 }, { "epoch": 1.99, "learning_rate": 1.006144e-05, "loss": 0.0013, "step": 280500 }, { "epoch": 2.0, "learning_rate": 1.0025884444444445e-05, "loss": 0.0013, "step": 281000 }, { "epoch": 2.0, "learning_rate": 9.9904e-06, "loss": 0.0013, "step": 281500 }, { "epoch": 2.01, "learning_rate": 9.954844444444444e-06, "loss": 0.0012, "step": 282000 }, { "epoch": 2.01, "learning_rate": 9.91928888888889e-06, "loss": 0.0012, "step": 282500 }, { "epoch": 2.01, "learning_rate": 9.883733333333333e-06, "loss": 0.0013, "step": 283000 }, { "epoch": 2.02, "learning_rate": 9.84824888888889e-06, "loss": 0.0013, "step": 283500 }, { "epoch": 2.02, "learning_rate": 9.812693333333332e-06, "loss": 0.0012, "step": 284000 }, { "epoch": 2.02, "learning_rate": 9.777137777777778e-06, "loss": 0.0012, "step": 284500 }, { "epoch": 2.03, "learning_rate": 9.741582222222222e-06, "loss": 0.0013, "step": 285000 }, { "epoch": 2.03, "learning_rate": 9.706097777777779e-06, "loss": 0.0013, "step": 285500 }, { "epoch": 2.03, "learning_rate": 9.670542222222223e-06, "loss": 0.0013, "step": 286000 }, { "epoch": 2.04, "learning_rate": 9.634986666666668e-06, "loss": 0.0012, "step": 286500 }, { "epoch": 2.04, "learning_rate": 9.599431111111112e-06, "loss": 0.0012, "step": 287000 }, { "epoch": 2.04, "learning_rate": 9.563946666666667e-06, "loss": 0.0012, "step": 287500 }, { "epoch": 2.05, "learning_rate": 9.528391111111111e-06, "loss": 0.0012, "step": 288000 }, { "epoch": 2.05, "learning_rate": 9.492835555555557e-06, "loss": 0.0012, "step": 288500 }, { "epoch": 2.06, "learning_rate": 9.45728e-06, "loss": 0.0012, "step": 289000 }, { "epoch": 2.06, "learning_rate": 9.421795555555556e-06, "loss": 0.0012, "step": 289500 }, { "epoch": 2.06, "learning_rate": 9.386240000000001e-06, "loss": 0.0012, "step": 290000 }, { "epoch": 2.07, "learning_rate": 9.350684444444443e-06, "loss": 0.0012, "step": 290500 }, { "epoch": 2.07, "learning_rate": 9.315128888888889e-06, "loss": 0.0012, "step": 291000 }, { "epoch": 2.07, "learning_rate": 9.279644444444444e-06, "loss": 0.0012, "step": 291500 }, { "epoch": 2.08, "learning_rate": 9.24408888888889e-06, "loss": 0.0012, "step": 292000 }, { "epoch": 2.08, "learning_rate": 9.208533333333333e-06, "loss": 0.0012, "step": 292500 }, { "epoch": 2.08, "learning_rate": 9.172977777777779e-06, "loss": 0.0012, "step": 293000 }, { "epoch": 2.09, "learning_rate": 9.137493333333334e-06, "loss": 0.0012, "step": 293500 }, { "epoch": 2.09, "learning_rate": 9.101937777777778e-06, "loss": 0.0012, "step": 294000 }, { "epoch": 2.09, "learning_rate": 9.066382222222222e-06, "loss": 0.0012, "step": 294500 }, { "epoch": 2.1, "learning_rate": 9.030826666666667e-06, "loss": 0.0012, "step": 295000 }, { "epoch": 2.1, "learning_rate": 8.995342222222223e-06, "loss": 0.0012, "step": 295500 }, { "epoch": 2.1, "learning_rate": 8.959786666666666e-06, "loss": 0.0012, "step": 296000 }, { "epoch": 2.11, "learning_rate": 8.924231111111112e-06, "loss": 0.0012, "step": 296500 }, { "epoch": 2.11, "learning_rate": 8.888675555555556e-06, "loss": 0.0013, "step": 297000 }, { "epoch": 2.12, "learning_rate": 8.85312e-06, "loss": 0.0012, "step": 297500 }, { "epoch": 2.12, "learning_rate": 8.817635555555555e-06, "loss": 0.0012, "step": 298000 }, { "epoch": 2.12, "learning_rate": 8.78208e-06, "loss": 0.0012, "step": 298500 }, { "epoch": 2.13, "learning_rate": 8.746524444444444e-06, "loss": 0.0012, "step": 299000 }, { "epoch": 2.13, "learning_rate": 8.71096888888889e-06, "loss": 0.0012, "step": 299500 }, { "epoch": 2.13, "learning_rate": 8.675484444444445e-06, "loss": 0.0013, "step": 300000 }, { "epoch": 2.14, "learning_rate": 8.639928888888889e-06, "loss": 0.0012, "step": 300500 }, { "epoch": 2.14, "learning_rate": 8.604373333333333e-06, "loss": 0.0012, "step": 301000 }, { "epoch": 2.14, "learning_rate": 8.568817777777778e-06, "loss": 0.0012, "step": 301500 }, { "epoch": 2.15, "learning_rate": 8.533333333333334e-06, "loss": 0.0013, "step": 302000 }, { "epoch": 2.15, "learning_rate": 8.497777777777777e-06, "loss": 0.0012, "step": 302500 }, { "epoch": 2.15, "learning_rate": 8.462222222222223e-06, "loss": 0.0012, "step": 303000 }, { "epoch": 2.16, "learning_rate": 8.426666666666667e-06, "loss": 0.0012, "step": 303500 }, { "epoch": 2.16, "learning_rate": 8.391182222222222e-06, "loss": 0.0012, "step": 304000 }, { "epoch": 2.17, "learning_rate": 8.355626666666666e-06, "loss": 0.0012, "step": 304500 }, { "epoch": 2.17, "learning_rate": 8.320071111111111e-06, "loss": 0.0013, "step": 305000 }, { "epoch": 2.17, "learning_rate": 8.284515555555555e-06, "loss": 0.0012, "step": 305500 }, { "epoch": 2.18, "learning_rate": 8.249031111111112e-06, "loss": 0.0012, "step": 306000 }, { "epoch": 2.18, "learning_rate": 8.213475555555556e-06, "loss": 0.0012, "step": 306500 }, { "epoch": 2.18, "learning_rate": 8.177920000000001e-06, "loss": 0.0013, "step": 307000 }, { "epoch": 2.19, "learning_rate": 8.142364444444444e-06, "loss": 0.0012, "step": 307500 }, { "epoch": 2.19, "learning_rate": 8.10688e-06, "loss": 0.0012, "step": 308000 }, { "epoch": 2.19, "learning_rate": 8.071324444444444e-06, "loss": 0.0012, "step": 308500 }, { "epoch": 2.2, "learning_rate": 8.03576888888889e-06, "loss": 0.0013, "step": 309000 }, { "epoch": 2.2, "learning_rate": 8.000213333333334e-06, "loss": 0.0012, "step": 309500 }, { "epoch": 2.2, "learning_rate": 7.964728888888889e-06, "loss": 0.0012, "step": 310000 }, { "epoch": 2.21, "learning_rate": 7.929244444444444e-06, "loss": 0.0013, "step": 310500 }, { "epoch": 2.21, "learning_rate": 7.893688888888888e-06, "loss": 0.0012, "step": 311000 }, { "epoch": 2.22, "learning_rate": 7.858133333333334e-06, "loss": 0.0013, "step": 311500 }, { "epoch": 2.22, "learning_rate": 7.822577777777777e-06, "loss": 0.0012, "step": 312000 }, { "epoch": 2.22, "learning_rate": 7.787022222222223e-06, "loss": 0.0012, "step": 312500 }, { "epoch": 2.23, "learning_rate": 7.751466666666667e-06, "loss": 0.0013, "step": 313000 }, { "epoch": 2.23, "learning_rate": 7.715911111111112e-06, "loss": 0.0012, "step": 313500 }, { "epoch": 2.23, "learning_rate": 7.680355555555556e-06, "loss": 0.0012, "step": 314000 }, { "epoch": 2.24, "learning_rate": 7.644871111111111e-06, "loss": 0.0012, "step": 314500 }, { "epoch": 2.24, "learning_rate": 7.609315555555555e-06, "loss": 0.0012, "step": 315000 }, { "epoch": 2.24, "learning_rate": 7.57376e-06, "loss": 0.0012, "step": 315500 }, { "epoch": 2.25, "learning_rate": 7.538204444444445e-06, "loss": 0.0012, "step": 316000 }, { "epoch": 2.25, "learning_rate": 7.502720000000001e-06, "loss": 0.0012, "step": 316500 }, { "epoch": 2.25, "learning_rate": 7.467235555555556e-06, "loss": 0.0012, "step": 317000 }, { "epoch": 2.26, "learning_rate": 7.431680000000001e-06, "loss": 0.0012, "step": 317500 }, { "epoch": 2.26, "learning_rate": 7.3961244444444445e-06, "loss": 0.0012, "step": 318000 }, { "epoch": 2.26, "learning_rate": 7.360568888888889e-06, "loss": 0.0012, "step": 318500 }, { "epoch": 2.27, "learning_rate": 7.325013333333334e-06, "loss": 0.0012, "step": 319000 }, { "epoch": 2.27, "learning_rate": 7.2894577777777786e-06, "loss": 0.0012, "step": 319500 }, { "epoch": 2.28, "learning_rate": 7.253902222222222e-06, "loss": 0.0012, "step": 320000 }, { "epoch": 2.28, "learning_rate": 7.218346666666667e-06, "loss": 0.0012, "step": 320500 }, { "epoch": 2.28, "learning_rate": 7.182862222222222e-06, "loss": 0.0012, "step": 321000 }, { "epoch": 2.29, "learning_rate": 7.147306666666666e-06, "loss": 0.0012, "step": 321500 }, { "epoch": 2.29, "learning_rate": 7.111751111111111e-06, "loss": 0.0012, "step": 322000 }, { "epoch": 2.29, "learning_rate": 7.0761955555555555e-06, "loss": 0.0012, "step": 322500 }, { "epoch": 2.3, "learning_rate": 7.040711111111112e-06, "loss": 0.0012, "step": 323000 }, { "epoch": 2.3, "learning_rate": 7.0051555555555555e-06, "loss": 0.0012, "step": 323500 }, { "epoch": 2.3, "learning_rate": 6.9696e-06, "loss": 0.0012, "step": 324000 }, { "epoch": 2.31, "learning_rate": 6.934044444444445e-06, "loss": 0.0012, "step": 324500 }, { "epoch": 2.31, "learning_rate": 6.89856e-06, "loss": 0.0012, "step": 325000 }, { "epoch": 2.31, "learning_rate": 6.863004444444445e-06, "loss": 0.0012, "step": 325500 }, { "epoch": 2.32, "learning_rate": 6.8274488888888894e-06, "loss": 0.0012, "step": 326000 }, { "epoch": 2.32, "learning_rate": 6.791893333333334e-06, "loss": 0.0012, "step": 326500 }, { "epoch": 2.33, "learning_rate": 6.7564088888888885e-06, "loss": 0.0012, "step": 327000 }, { "epoch": 2.33, "learning_rate": 6.720853333333333e-06, "loss": 0.0012, "step": 327500 }, { "epoch": 2.33, "learning_rate": 6.685297777777778e-06, "loss": 0.0012, "step": 328000 }, { "epoch": 2.34, "learning_rate": 6.649742222222222e-06, "loss": 0.0012, "step": 328500 }, { "epoch": 2.34, "learning_rate": 6.614257777777778e-06, "loss": 0.0012, "step": 329000 }, { "epoch": 2.34, "learning_rate": 6.5787022222222225e-06, "loss": 0.0012, "step": 329500 }, { "epoch": 2.35, "learning_rate": 6.543146666666666e-06, "loss": 0.0012, "step": 330000 }, { "epoch": 2.35, "learning_rate": 6.507591111111111e-06, "loss": 0.0013, "step": 330500 }, { "epoch": 2.35, "learning_rate": 6.472106666666667e-06, "loss": 0.0012, "step": 331000 }, { "epoch": 2.36, "learning_rate": 6.436551111111112e-06, "loss": 0.0012, "step": 331500 }, { "epoch": 2.36, "learning_rate": 6.400995555555556e-06, "loss": 0.0012, "step": 332000 }, { "epoch": 2.36, "learning_rate": 6.36544e-06, "loss": 0.0012, "step": 332500 }, { "epoch": 2.37, "learning_rate": 6.329955555555556e-06, "loss": 0.0012, "step": 333000 }, { "epoch": 2.37, "learning_rate": 6.2944e-06, "loss": 0.0012, "step": 333500 }, { "epoch": 2.38, "learning_rate": 6.258844444444445e-06, "loss": 0.0012, "step": 334000 }, { "epoch": 2.38, "learning_rate": 6.223288888888889e-06, "loss": 0.0012, "step": 334500 }, { "epoch": 2.38, "learning_rate": 6.187804444444444e-06, "loss": 0.0012, "step": 335000 }, { "epoch": 2.39, "learning_rate": 6.152248888888889e-06, "loss": 0.0012, "step": 335500 }, { "epoch": 2.39, "learning_rate": 6.116693333333333e-06, "loss": 0.0012, "step": 336000 }, { "epoch": 2.39, "learning_rate": 6.081137777777778e-06, "loss": 0.0012, "step": 336500 }, { "epoch": 2.4, "learning_rate": 6.045653333333333e-06, "loss": 0.0012, "step": 337000 }, { "epoch": 2.4, "learning_rate": 6.010097777777778e-06, "loss": 0.0012, "step": 337500 }, { "epoch": 2.4, "learning_rate": 5.974542222222223e-06, "loss": 0.0012, "step": 338000 }, { "epoch": 2.41, "learning_rate": 5.9389866666666665e-06, "loss": 0.0012, "step": 338500 }, { "epoch": 2.41, "learning_rate": 5.903502222222223e-06, "loss": 0.0012, "step": 339000 }, { "epoch": 2.41, "learning_rate": 5.867946666666667e-06, "loss": 0.0012, "step": 339500 }, { "epoch": 2.42, "learning_rate": 5.832391111111112e-06, "loss": 0.0012, "step": 340000 }, { "epoch": 2.42, "learning_rate": 5.796906666666666e-06, "loss": 0.0012, "step": 340500 }, { "epoch": 2.42, "learning_rate": 5.761351111111111e-06, "loss": 0.0012, "step": 341000 }, { "epoch": 2.43, "learning_rate": 5.725795555555556e-06, "loss": 0.0012, "step": 341500 }, { "epoch": 2.43, "learning_rate": 5.69024e-06, "loss": 0.0012, "step": 342000 }, { "epoch": 2.44, "learning_rate": 5.654684444444444e-06, "loss": 0.0012, "step": 342500 }, { "epoch": 2.44, "learning_rate": 5.619128888888889e-06, "loss": 0.0012, "step": 343000 }, { "epoch": 2.44, "learning_rate": 5.583644444444444e-06, "loss": 0.0012, "step": 343500 }, { "epoch": 2.45, "learning_rate": 5.548088888888889e-06, "loss": 0.0012, "step": 344000 }, { "epoch": 2.45, "learning_rate": 5.5125333333333336e-06, "loss": 0.0012, "step": 344500 }, { "epoch": 2.45, "learning_rate": 5.476977777777778e-06, "loss": 0.0012, "step": 345000 }, { "epoch": 2.46, "learning_rate": 5.441422222222222e-06, "loss": 0.0012, "step": 345500 }, { "epoch": 2.46, "learning_rate": 5.405866666666667e-06, "loss": 0.0012, "step": 346000 }, { "epoch": 2.46, "learning_rate": 5.3703111111111115e-06, "loss": 0.0012, "step": 346500 }, { "epoch": 2.47, "learning_rate": 5.334755555555556e-06, "loss": 0.0012, "step": 347000 }, { "epoch": 2.47, "learning_rate": 5.299271111111111e-06, "loss": 0.0012, "step": 347500 }, { "epoch": 2.47, "learning_rate": 5.263715555555556e-06, "loss": 0.0012, "step": 348000 }, { "epoch": 2.48, "learning_rate": 5.228160000000001e-06, "loss": 0.0012, "step": 348500 }, { "epoch": 2.48, "learning_rate": 5.192604444444445e-06, "loss": 0.0012, "step": 349000 }, { "epoch": 2.49, "learning_rate": 5.157191111111111e-06, "loss": 0.0012, "step": 349500 }, { "epoch": 2.49, "learning_rate": 5.121635555555556e-06, "loss": 0.0012, "step": 350000 }, { "epoch": 2.49, "learning_rate": 5.08608e-06, "loss": 0.0012, "step": 350500 }, { "epoch": 2.5, "learning_rate": 5.0505244444444445e-06, "loss": 0.0012, "step": 351000 }, { "epoch": 2.5, "learning_rate": 5.014968888888889e-06, "loss": 0.0012, "step": 351500 }, { "epoch": 2.5, "learning_rate": 4.979413333333334e-06, "loss": 0.0012, "step": 352000 }, { "epoch": 2.51, "learning_rate": 4.943857777777778e-06, "loss": 0.0012, "step": 352500 }, { "epoch": 2.51, "learning_rate": 4.908302222222222e-06, "loss": 0.0012, "step": 353000 }, { "epoch": 2.51, "learning_rate": 4.8728177777777784e-06, "loss": 0.0012, "step": 353500 }, { "epoch": 2.52, "learning_rate": 4.837262222222222e-06, "loss": 0.0012, "step": 354000 }, { "epoch": 2.52, "learning_rate": 4.801706666666667e-06, "loss": 0.0012, "step": 354500 }, { "epoch": 2.52, "learning_rate": 4.766151111111112e-06, "loss": 0.0012, "step": 355000 }, { "epoch": 2.53, "learning_rate": 4.730666666666667e-06, "loss": 0.0012, "step": 355500 }, { "epoch": 2.53, "learning_rate": 4.695111111111111e-06, "loss": 0.0012, "step": 356000 }, { "epoch": 2.54, "learning_rate": 4.659555555555555e-06, "loss": 0.0012, "step": 356500 }, { "epoch": 2.54, "learning_rate": 4.624e-06, "loss": 0.0012, "step": 357000 }, { "epoch": 2.54, "learning_rate": 4.588515555555555e-06, "loss": 0.0012, "step": 357500 }, { "epoch": 2.55, "learning_rate": 4.55296e-06, "loss": 0.0012, "step": 358000 }, { "epoch": 2.55, "learning_rate": 4.517404444444445e-06, "loss": 0.0012, "step": 358500 }, { "epoch": 2.55, "learning_rate": 4.4818488888888886e-06, "loss": 0.0012, "step": 359000 }, { "epoch": 2.56, "learning_rate": 4.446364444444445e-06, "loss": 0.0012, "step": 359500 }, { "epoch": 2.56, "learning_rate": 4.410808888888889e-06, "loss": 0.0012, "step": 360000 }, { "epoch": 2.56, "learning_rate": 4.375253333333333e-06, "loss": 0.0012, "step": 360500 }, { "epoch": 2.57, "learning_rate": 4.339697777777778e-06, "loss": 0.0012, "step": 361000 }, { "epoch": 2.57, "learning_rate": 4.304213333333334e-06, "loss": 0.0012, "step": 361500 }, { "epoch": 2.57, "learning_rate": 4.268657777777779e-06, "loss": 0.0012, "step": 362000 }, { "epoch": 2.58, "learning_rate": 4.2331022222222225e-06, "loss": 0.0012, "step": 362500 }, { "epoch": 2.58, "learning_rate": 4.197546666666667e-06, "loss": 0.0012, "step": 363000 }, { "epoch": 2.58, "learning_rate": 4.161991111111111e-06, "loss": 0.0012, "step": 363500 }, { "epoch": 2.59, "learning_rate": 4.126506666666666e-06, "loss": 0.0012, "step": 364000 }, { "epoch": 2.59, "learning_rate": 4.090951111111111e-06, "loss": 0.0012, "step": 364500 }, { "epoch": 2.6, "learning_rate": 4.055395555555556e-06, "loss": 0.0012, "step": 365000 }, { "epoch": 2.6, "learning_rate": 4.01984e-06, "loss": 0.0012, "step": 365500 }, { "epoch": 2.6, "learning_rate": 3.9843555555555555e-06, "loss": 0.0012, "step": 366000 }, { "epoch": 2.61, "learning_rate": 3.9488e-06, "loss": 0.0012, "step": 366500 }, { "epoch": 2.61, "learning_rate": 3.913244444444445e-06, "loss": 0.0012, "step": 367000 }, { "epoch": 2.61, "learning_rate": 3.877688888888889e-06, "loss": 0.0012, "step": 367500 }, { "epoch": 2.62, "learning_rate": 3.842204444444445e-06, "loss": 0.0012, "step": 368000 }, { "epoch": 2.62, "learning_rate": 3.806648888888889e-06, "loss": 0.0012, "step": 368500 }, { "epoch": 2.62, "learning_rate": 3.771093333333333e-06, "loss": 0.0012, "step": 369000 }, { "epoch": 2.63, "learning_rate": 3.7355377777777776e-06, "loss": 0.0011, "step": 369500 }, { "epoch": 2.63, "learning_rate": 3.7000533333333337e-06, "loss": 0.0012, "step": 370000 }, { "epoch": 2.63, "learning_rate": 3.664497777777778e-06, "loss": 0.0012, "step": 370500 }, { "epoch": 2.64, "learning_rate": 3.6289422222222226e-06, "loss": 0.0012, "step": 371000 }, { "epoch": 2.64, "learning_rate": 3.593386666666667e-06, "loss": 0.0012, "step": 371500 }, { "epoch": 2.65, "learning_rate": 3.557902222222222e-06, "loss": 0.0011, "step": 372000 }, { "epoch": 2.65, "learning_rate": 3.522346666666667e-06, "loss": 0.0011, "step": 372500 }, { "epoch": 2.65, "learning_rate": 3.486791111111111e-06, "loss": 0.0012, "step": 373000 }, { "epoch": 2.66, "learning_rate": 3.4512355555555558e-06, "loss": 0.0012, "step": 373500 }, { "epoch": 2.66, "learning_rate": 3.4157511111111114e-06, "loss": 0.0012, "step": 374000 }, { "epoch": 2.66, "learning_rate": 3.3801955555555557e-06, "loss": 0.0012, "step": 374500 }, { "epoch": 2.67, "learning_rate": 3.3446400000000004e-06, "loss": 0.0012, "step": 375000 }, { "epoch": 2.67, "learning_rate": 3.3090844444444442e-06, "loss": 0.0012, "step": 375500 }, { "epoch": 2.67, "learning_rate": 3.2736e-06, "loss": 0.0012, "step": 376000 }, { "epoch": 2.68, "learning_rate": 3.2380444444444446e-06, "loss": 0.0012, "step": 376500 }, { "epoch": 2.68, "learning_rate": 3.202488888888889e-06, "loss": 0.0012, "step": 377000 }, { "epoch": 2.68, "learning_rate": 3.1669333333333335e-06, "loss": 0.0012, "step": 377500 }, { "epoch": 2.69, "learning_rate": 3.131448888888889e-06, "loss": 0.0012, "step": 378000 }, { "epoch": 2.69, "learning_rate": 3.095893333333333e-06, "loss": 0.0012, "step": 378500 }, { "epoch": 2.7, "learning_rate": 3.0603377777777777e-06, "loss": 0.0012, "step": 379000 }, { "epoch": 2.7, "learning_rate": 3.024782222222222e-06, "loss": 0.0012, "step": 379500 }, { "epoch": 2.7, "learning_rate": 2.989297777777778e-06, "loss": 0.0012, "step": 380000 }, { "epoch": 2.71, "learning_rate": 2.9537422222222223e-06, "loss": 0.0012, "step": 380500 }, { "epoch": 2.71, "learning_rate": 2.918186666666667e-06, "loss": 0.0012, "step": 381000 }, { "epoch": 2.71, "learning_rate": 2.8826311111111113e-06, "loss": 0.0012, "step": 381500 }, { "epoch": 2.72, "learning_rate": 2.8471466666666665e-06, "loss": 0.0012, "step": 382000 }, { "epoch": 2.72, "learning_rate": 2.811591111111111e-06, "loss": 0.0012, "step": 382500 }, { "epoch": 2.72, "learning_rate": 2.7760355555555555e-06, "loss": 0.0012, "step": 383000 }, { "epoch": 2.73, "learning_rate": 2.74048e-06, "loss": 0.0012, "step": 383500 }, { "epoch": 2.73, "learning_rate": 2.704995555555556e-06, "loss": 0.0012, "step": 384000 }, { "epoch": 2.73, "learning_rate": 2.66944e-06, "loss": 0.0012, "step": 384500 }, { "epoch": 2.74, "learning_rate": 2.6338844444444448e-06, "loss": 0.0012, "step": 385000 }, { "epoch": 2.74, "learning_rate": 2.598328888888889e-06, "loss": 0.0012, "step": 385500 }, { "epoch": 2.74, "learning_rate": 2.5628444444444442e-06, "loss": 0.0012, "step": 386000 }, { "epoch": 2.75, "learning_rate": 2.527288888888889e-06, "loss": 0.0012, "step": 386500 }, { "epoch": 2.75, "learning_rate": 2.491733333333333e-06, "loss": 0.0012, "step": 387000 }, { "epoch": 2.76, "learning_rate": 2.456177777777778e-06, "loss": 0.0012, "step": 387500 }, { "epoch": 2.76, "learning_rate": 2.4206933333333335e-06, "loss": 0.0012, "step": 388000 }, { "epoch": 2.76, "learning_rate": 2.3851377777777782e-06, "loss": 0.0012, "step": 388500 }, { "epoch": 2.77, "learning_rate": 2.349582222222222e-06, "loss": 0.0012, "step": 389000 }, { "epoch": 2.77, "learning_rate": 2.3140266666666668e-06, "loss": 0.0012, "step": 389500 }, { "epoch": 2.77, "learning_rate": 2.2785422222222224e-06, "loss": 0.0012, "step": 390000 }, { "epoch": 2.78, "learning_rate": 2.2429866666666667e-06, "loss": 0.0012, "step": 390500 }, { "epoch": 2.78, "learning_rate": 2.2074311111111114e-06, "loss": 0.0012, "step": 391000 }, { "epoch": 2.78, "learning_rate": 2.1718755555555556e-06, "loss": 0.0012, "step": 391500 }, { "epoch": 2.79, "learning_rate": 2.136391111111111e-06, "loss": 0.0012, "step": 392000 }, { "epoch": 2.79, "learning_rate": 2.1008355555555555e-06, "loss": 0.0012, "step": 392500 }, { "epoch": 2.79, "learning_rate": 2.06528e-06, "loss": 0.0012, "step": 393000 }, { "epoch": 2.8, "learning_rate": 2.0297244444444445e-06, "loss": 0.0012, "step": 393500 }, { "epoch": 2.8, "learning_rate": 1.99424e-06, "loss": 0.0012, "step": 394000 }, { "epoch": 2.81, "learning_rate": 1.9586844444444444e-06, "loss": 0.0012, "step": 394500 }, { "epoch": 2.81, "learning_rate": 1.923128888888889e-06, "loss": 0.0011, "step": 395000 }, { "epoch": 2.81, "learning_rate": 1.8875733333333332e-06, "loss": 0.0011, "step": 395500 }, { "epoch": 2.82, "learning_rate": 1.852088888888889e-06, "loss": 0.0012, "step": 396000 }, { "epoch": 2.82, "learning_rate": 1.8165333333333333e-06, "loss": 0.0012, "step": 396500 }, { "epoch": 2.82, "learning_rate": 1.7809777777777778e-06, "loss": 0.0012, "step": 397000 }, { "epoch": 2.83, "learning_rate": 1.7454222222222223e-06, "loss": 0.0012, "step": 397500 }, { "epoch": 2.83, "learning_rate": 1.7099377777777777e-06, "loss": 0.0012, "step": 398000 }, { "epoch": 2.83, "learning_rate": 1.6743822222222222e-06, "loss": 0.0012, "step": 398500 }, { "epoch": 2.84, "learning_rate": 1.6388266666666666e-06, "loss": 0.0012, "step": 399000 }, { "epoch": 2.84, "learning_rate": 1.6032711111111111e-06, "loss": 0.0012, "step": 399500 }, { "epoch": 2.84, "learning_rate": 1.5677866666666666e-06, "loss": 0.0012, "step": 400000 }, { "epoch": 2.85, "learning_rate": 1.532231111111111e-06, "loss": 0.0012, "step": 400500 }, { "epoch": 2.85, "learning_rate": 1.4966755555555555e-06, "loss": 0.0012, "step": 401000 }, { "epoch": 2.86, "learning_rate": 1.46112e-06, "loss": 0.0012, "step": 401500 }, { "epoch": 2.86, "learning_rate": 1.4255644444444445e-06, "loss": 0.0012, "step": 402000 }, { "epoch": 2.86, "learning_rate": 1.3900800000000001e-06, "loss": 0.0012, "step": 402500 }, { "epoch": 2.87, "learning_rate": 1.3545244444444446e-06, "loss": 0.0012, "step": 403000 }, { "epoch": 2.87, "learning_rate": 1.318968888888889e-06, "loss": 0.0012, "step": 403500 }, { "epoch": 2.87, "learning_rate": 1.2834133333333333e-06, "loss": 0.0012, "step": 404000 }, { "epoch": 2.88, "learning_rate": 1.247928888888889e-06, "loss": 0.0012, "step": 404500 }, { "epoch": 2.88, "learning_rate": 1.2123733333333335e-06, "loss": 0.0012, "step": 405000 }, { "epoch": 2.88, "learning_rate": 1.1768177777777777e-06, "loss": 0.0012, "step": 405500 }, { "epoch": 2.89, "learning_rate": 1.1412622222222222e-06, "loss": 0.0012, "step": 406000 }, { "epoch": 2.89, "learning_rate": 1.1057777777777779e-06, "loss": 0.0012, "step": 406500 }, { "epoch": 2.89, "learning_rate": 1.0702222222222221e-06, "loss": 0.0012, "step": 407000 }, { "epoch": 2.9, "learning_rate": 1.0346666666666666e-06, "loss": 0.0012, "step": 407500 }, { "epoch": 2.9, "learning_rate": 9.99111111111111e-07, "loss": 0.0012, "step": 408000 }, { "epoch": 2.9, "learning_rate": 9.636266666666665e-07, "loss": 0.0012, "step": 408500 }, { "epoch": 2.91, "learning_rate": 9.280711111111112e-07, "loss": 0.0012, "step": 409000 }, { "epoch": 2.91, "learning_rate": 8.925155555555556e-07, "loss": 0.0012, "step": 409500 }, { "epoch": 2.92, "learning_rate": 8.569600000000001e-07, "loss": 0.0012, "step": 410000 }, { "epoch": 2.92, "learning_rate": 8.214755555555555e-07, "loss": 0.0011, "step": 410500 }, { "epoch": 2.92, "learning_rate": 7.8592e-07, "loss": 0.0012, "step": 411000 }, { "epoch": 2.93, "learning_rate": 7.503644444444445e-07, "loss": 0.0012, "step": 411500 }, { "epoch": 2.93, "learning_rate": 7.148088888888889e-07, "loss": 0.0012, "step": 412000 }, { "epoch": 2.93, "learning_rate": 6.793244444444445e-07, "loss": 0.0012, "step": 412500 }, { "epoch": 2.94, "learning_rate": 6.43768888888889e-07, "loss": 0.0012, "step": 413000 }, { "epoch": 2.94, "learning_rate": 6.082133333333334e-07, "loss": 0.0011, "step": 413500 }, { "epoch": 2.94, "learning_rate": 5.726577777777777e-07, "loss": 0.0012, "step": 414000 }, { "epoch": 2.95, "learning_rate": 5.371733333333334e-07, "loss": 0.0012, "step": 414500 }, { "epoch": 2.95, "learning_rate": 5.016177777777778e-07, "loss": 0.0012, "step": 415000 }, { "epoch": 2.95, "learning_rate": 4.6606222222222226e-07, "loss": 0.0012, "step": 415500 }, { "epoch": 2.96, "learning_rate": 4.305066666666667e-07, "loss": 0.0012, "step": 416000 }, { "epoch": 2.96, "learning_rate": 3.9502222222222223e-07, "loss": 0.0012, "step": 416500 }, { "epoch": 2.97, "learning_rate": 3.5946666666666666e-07, "loss": 0.0012, "step": 417000 }, { "epoch": 2.97, "learning_rate": 3.2391111111111114e-07, "loss": 0.0012, "step": 417500 }, { "epoch": 2.97, "learning_rate": 2.8835555555555556e-07, "loss": 0.0012, "step": 418000 }, { "epoch": 2.98, "learning_rate": 2.528711111111111e-07, "loss": 0.0012, "step": 418500 }, { "epoch": 2.98, "learning_rate": 2.1731555555555556e-07, "loss": 0.0012, "step": 419000 }, { "epoch": 2.98, "learning_rate": 1.8176e-07, "loss": 0.0012, "step": 419500 }, { "epoch": 2.99, "learning_rate": 1.4620444444444443e-07, "loss": 0.0012, "step": 420000 }, { "epoch": 2.99, "learning_rate": 1.1072e-07, "loss": 0.0012, "step": 420500 }, { "epoch": 2.99, "learning_rate": 7.516444444444444e-08, "loss": 0.0011, "step": 421000 }, { "epoch": 3.0, "learning_rate": 3.9608888888888887e-08, "loss": 0.0012, "step": 421500 }, { "epoch": 3.0, "step": 421875, "total_flos": 0, "train_runtime": 37325.4581, "train_samples_per_second": 11.303 } ], "max_steps": 421875, "num_train_epochs": 3, "total_flos": 0, "trial_name": null, "trial_params": null }