{ "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": 2.996458666666667e-05, "loss": 0.0182, "step": 500 }, { "epoch": 0.01, "learning_rate": 2.992903111111111e-05, "loss": 0.0023, "step": 1000 }, { "epoch": 0.01, "learning_rate": 2.9893475555555557e-05, "loss": 0.0018, "step": 1500 }, { "epoch": 0.01, "learning_rate": 2.9857920000000002e-05, "loss": 0.0015, "step": 2000 }, { "epoch": 0.02, "learning_rate": 2.9822364444444448e-05, "loss": 0.0015, "step": 2500 }, { "epoch": 0.02, "learning_rate": 2.978680888888889e-05, "loss": 0.0013, "step": 3000 }, { "epoch": 0.02, "learning_rate": 2.9751253333333332e-05, "loss": 0.0013, "step": 3500 }, { "epoch": 0.03, "learning_rate": 2.9715697777777778e-05, "loss": 0.0012, "step": 4000 }, { "epoch": 0.03, "learning_rate": 2.9680142222222223e-05, "loss": 0.0012, "step": 4500 }, { "epoch": 0.04, "learning_rate": 2.9644586666666665e-05, "loss": 0.0011, "step": 5000 }, { "epoch": 0.04, "learning_rate": 2.960903111111111e-05, "loss": 0.0011, "step": 5500 }, { "epoch": 0.04, "learning_rate": 2.9573475555555556e-05, "loss": 0.001, "step": 6000 }, { "epoch": 0.05, "learning_rate": 2.9537920000000002e-05, "loss": 0.0011, "step": 6500 }, { "epoch": 0.05, "learning_rate": 2.9502364444444444e-05, "loss": 0.001, "step": 7000 }, { "epoch": 0.05, "learning_rate": 2.946680888888889e-05, "loss": 0.001, "step": 7500 }, { "epoch": 0.06, "learning_rate": 2.9431253333333335e-05, "loss": 0.0009, "step": 8000 }, { "epoch": 0.06, "learning_rate": 2.939569777777778e-05, "loss": 0.0009, "step": 8500 }, { "epoch": 0.06, "learning_rate": 2.9360142222222223e-05, "loss": 0.001, "step": 9000 }, { "epoch": 0.07, "learning_rate": 2.932458666666667e-05, "loss": 0.0009, "step": 9500 }, { "epoch": 0.07, "learning_rate": 2.9289031111111114e-05, "loss": 0.0009, "step": 10000 }, { "epoch": 0.07, "learning_rate": 2.9253475555555556e-05, "loss": 0.0009, "step": 10500 }, { "epoch": 0.08, "learning_rate": 2.9217919999999998e-05, "loss": 0.0009, "step": 11000 }, { "epoch": 0.08, "learning_rate": 2.9182364444444444e-05, "loss": 0.0008, "step": 11500 }, { "epoch": 0.09, "learning_rate": 2.914680888888889e-05, "loss": 0.0008, "step": 12000 }, { "epoch": 0.09, "learning_rate": 2.911125333333333e-05, "loss": 0.0008, "step": 12500 }, { "epoch": 0.09, "learning_rate": 2.907576888888889e-05, "loss": 0.0008, "step": 13000 }, { "epoch": 0.1, "learning_rate": 2.9040213333333336e-05, "loss": 0.0008, "step": 13500 }, { "epoch": 0.1, "learning_rate": 2.900465777777778e-05, "loss": 0.0008, "step": 14000 }, { "epoch": 0.1, "learning_rate": 2.8969102222222223e-05, "loss": 0.0008, "step": 14500 }, { "epoch": 0.11, "learning_rate": 2.8933546666666665e-05, "loss": 0.0008, "step": 15000 }, { "epoch": 0.11, "learning_rate": 2.889799111111111e-05, "loss": 0.0008, "step": 15500 }, { "epoch": 0.11, "learning_rate": 2.8862435555555557e-05, "loss": 0.0007, "step": 16000 }, { "epoch": 0.12, "learning_rate": 2.882688e-05, "loss": 0.0007, "step": 16500 }, { "epoch": 0.12, "learning_rate": 2.8791395555555557e-05, "loss": 0.0007, "step": 17000 }, { "epoch": 0.12, "learning_rate": 2.8755840000000003e-05, "loss": 0.0008, "step": 17500 }, { "epoch": 0.13, "learning_rate": 2.8720355555555558e-05, "loss": 0.0008, "step": 18000 }, { "epoch": 0.13, "learning_rate": 2.86848e-05, "loss": 0.0007, "step": 18500 }, { "epoch": 0.14, "learning_rate": 2.8649244444444446e-05, "loss": 0.0007, "step": 19000 }, { "epoch": 0.14, "learning_rate": 2.861368888888889e-05, "loss": 0.0007, "step": 19500 }, { "epoch": 0.14, "learning_rate": 2.8578133333333333e-05, "loss": 0.0007, "step": 20000 }, { "epoch": 0.15, "learning_rate": 2.854257777777778e-05, "loss": 0.0007, "step": 20500 }, { "epoch": 0.15, "learning_rate": 2.8507022222222225e-05, "loss": 0.0007, "step": 21000 }, { "epoch": 0.15, "learning_rate": 2.847153777777778e-05, "loss": 0.0007, "step": 21500 }, { "epoch": 0.16, "learning_rate": 2.8435982222222222e-05, "loss": 0.0007, "step": 22000 }, { "epoch": 0.16, "learning_rate": 2.8400426666666667e-05, "loss": 0.0007, "step": 22500 }, { "epoch": 0.16, "learning_rate": 2.8364871111111113e-05, "loss": 0.0007, "step": 23000 }, { "epoch": 0.17, "learning_rate": 2.832931555555556e-05, "loss": 0.0007, "step": 23500 }, { "epoch": 0.17, "learning_rate": 2.829376e-05, "loss": 0.0007, "step": 24000 }, { "epoch": 0.17, "learning_rate": 2.8258204444444446e-05, "loss": 0.0007, "step": 24500 }, { "epoch": 0.18, "learning_rate": 2.8222648888888892e-05, "loss": 0.0007, "step": 25000 }, { "epoch": 0.18, "learning_rate": 2.8187164444444447e-05, "loss": 0.0007, "step": 25500 }, { "epoch": 0.18, "learning_rate": 2.815160888888889e-05, "loss": 0.0006, "step": 26000 }, { "epoch": 0.19, "learning_rate": 2.8116053333333335e-05, "loss": 0.0007, "step": 26500 }, { "epoch": 0.19, "learning_rate": 2.808049777777778e-05, "loss": 0.0006, "step": 27000 }, { "epoch": 0.2, "learning_rate": 2.8045013333333332e-05, "loss": 0.0007, "step": 27500 }, { "epoch": 0.2, "learning_rate": 2.8009457777777778e-05, "loss": 0.0006, "step": 28000 }, { "epoch": 0.2, "learning_rate": 2.7973902222222223e-05, "loss": 0.0007, "step": 28500 }, { "epoch": 0.21, "learning_rate": 2.793834666666667e-05, "loss": 0.0006, "step": 29000 }, { "epoch": 0.21, "learning_rate": 2.790286222222222e-05, "loss": 0.0006, "step": 29500 }, { "epoch": 0.21, "learning_rate": 2.7867306666666666e-05, "loss": 0.0006, "step": 30000 }, { "epoch": 0.22, "learning_rate": 2.783175111111111e-05, "loss": 0.0006, "step": 30500 }, { "epoch": 0.22, "learning_rate": 2.7796195555555557e-05, "loss": 0.0007, "step": 31000 }, { "epoch": 0.22, "learning_rate": 2.7760711111111112e-05, "loss": 0.0006, "step": 31500 }, { "epoch": 0.23, "learning_rate": 2.7725155555555554e-05, "loss": 0.0006, "step": 32000 }, { "epoch": 0.23, "learning_rate": 2.76896e-05, "loss": 0.0006, "step": 32500 }, { "epoch": 0.23, "learning_rate": 2.7654044444444446e-05, "loss": 0.0006, "step": 33000 }, { "epoch": 0.24, "learning_rate": 2.761856e-05, "loss": 0.0006, "step": 33500 }, { "epoch": 0.24, "learning_rate": 2.7583004444444446e-05, "loss": 0.0006, "step": 34000 }, { "epoch": 0.25, "learning_rate": 2.7547448888888892e-05, "loss": 0.0006, "step": 34500 }, { "epoch": 0.25, "learning_rate": 2.7511964444444444e-05, "loss": 0.0006, "step": 35000 }, { "epoch": 0.25, "learning_rate": 2.747640888888889e-05, "loss": 0.0006, "step": 35500 }, { "epoch": 0.26, "learning_rate": 2.7440853333333335e-05, "loss": 0.0006, "step": 36000 }, { "epoch": 0.26, "learning_rate": 2.740529777777778e-05, "loss": 0.0006, "step": 36500 }, { "epoch": 0.26, "learning_rate": 2.7369742222222222e-05, "loss": 0.0006, "step": 37000 }, { "epoch": 0.27, "learning_rate": 2.7334186666666668e-05, "loss": 0.0006, "step": 37500 }, { "epoch": 0.27, "learning_rate": 2.7298631111111113e-05, "loss": 0.0006, "step": 38000 }, { "epoch": 0.27, "learning_rate": 2.726307555555556e-05, "loss": 0.0006, "step": 38500 }, { "epoch": 0.28, "learning_rate": 2.722759111111111e-05, "loss": 0.0006, "step": 39000 }, { "epoch": 0.28, "learning_rate": 2.7192035555555556e-05, "loss": 0.0006, "step": 39500 }, { "epoch": 0.28, "learning_rate": 2.7156480000000002e-05, "loss": 0.0006, "step": 40000 }, { "epoch": 0.29, "learning_rate": 2.7120924444444447e-05, "loss": 0.0006, "step": 40500 }, { "epoch": 0.29, "learning_rate": 2.708544e-05, "loss": 0.0006, "step": 41000 }, { "epoch": 0.3, "learning_rate": 2.7049884444444445e-05, "loss": 0.0006, "step": 41500 }, { "epoch": 0.3, "learning_rate": 2.701432888888889e-05, "loss": 0.0006, "step": 42000 }, { "epoch": 0.3, "learning_rate": 2.6978773333333333e-05, "loss": 0.0006, "step": 42500 }, { "epoch": 0.31, "learning_rate": 2.6943217777777778e-05, "loss": 0.0006, "step": 43000 }, { "epoch": 0.31, "learning_rate": 2.6907733333333333e-05, "loss": 0.0006, "step": 43500 }, { "epoch": 0.31, "learning_rate": 2.687217777777778e-05, "loss": 0.0006, "step": 44000 }, { "epoch": 0.32, "learning_rate": 2.683662222222222e-05, "loss": 0.0006, "step": 44500 }, { "epoch": 0.32, "learning_rate": 2.6801066666666667e-05, "loss": 0.0006, "step": 45000 }, { "epoch": 0.32, "learning_rate": 2.6765582222222222e-05, "loss": 0.0006, "step": 45500 }, { "epoch": 0.33, "learning_rate": 2.6730026666666667e-05, "loss": 0.0006, "step": 46000 }, { "epoch": 0.33, "learning_rate": 2.669447111111111e-05, "loss": 0.0006, "step": 46500 }, { "epoch": 0.33, "learning_rate": 2.6658915555555555e-05, "loss": 0.0006, "step": 47000 }, { "epoch": 0.34, "learning_rate": 2.6623431111111114e-05, "loss": 0.0006, "step": 47500 }, { "epoch": 0.34, "learning_rate": 2.658787555555556e-05, "loss": 0.0006, "step": 48000 }, { "epoch": 0.34, "learning_rate": 2.655232e-05, "loss": 0.0005, "step": 48500 }, { "epoch": 0.35, "learning_rate": 2.6516764444444443e-05, "loss": 0.0005, "step": 49000 }, { "epoch": 0.35, "learning_rate": 2.6481280000000002e-05, "loss": 0.0006, "step": 49500 }, { "epoch": 0.36, "learning_rate": 2.6445724444444444e-05, "loss": 0.0005, "step": 50000 }, { "epoch": 0.36, "learning_rate": 2.641016888888889e-05, "loss": 0.0006, "step": 50500 }, { "epoch": 0.36, "learning_rate": 2.6374613333333335e-05, "loss": 0.0005, "step": 51000 }, { "epoch": 0.37, "learning_rate": 2.633912888888889e-05, "loss": 0.0006, "step": 51500 }, { "epoch": 0.37, "learning_rate": 2.6303573333333333e-05, "loss": 0.0005, "step": 52000 }, { "epoch": 0.37, "learning_rate": 2.6268017777777778e-05, "loss": 0.0006, "step": 52500 }, { "epoch": 0.38, "learning_rate": 2.6232462222222224e-05, "loss": 0.0005, "step": 53000 }, { "epoch": 0.38, "learning_rate": 2.619690666666667e-05, "loss": 0.0005, "step": 53500 }, { "epoch": 0.38, "learning_rate": 2.616142222222222e-05, "loss": 0.0005, "step": 54000 }, { "epoch": 0.39, "learning_rate": 2.6125866666666667e-05, "loss": 0.0005, "step": 54500 }, { "epoch": 0.39, "learning_rate": 2.6090311111111112e-05, "loss": 0.0005, "step": 55000 }, { "epoch": 0.39, "learning_rate": 2.6054755555555558e-05, "loss": 0.0005, "step": 55500 }, { "epoch": 0.4, "learning_rate": 2.601927111111111e-05, "loss": 0.0005, "step": 56000 }, { "epoch": 0.4, "learning_rate": 2.5983715555555555e-05, "loss": 0.0005, "step": 56500 }, { "epoch": 0.41, "learning_rate": 2.594816e-05, "loss": 0.0005, "step": 57000 }, { "epoch": 0.41, "learning_rate": 2.5912604444444443e-05, "loss": 0.0006, "step": 57500 }, { "epoch": 0.41, "learning_rate": 2.587712e-05, "loss": 0.0005, "step": 58000 }, { "epoch": 0.42, "learning_rate": 2.5841564444444447e-05, "loss": 0.0005, "step": 58500 }, { "epoch": 0.42, "learning_rate": 2.5806008888888892e-05, "loss": 0.0005, "step": 59000 }, { "epoch": 0.42, "learning_rate": 2.577045333333333e-05, "loss": 0.0005, "step": 59500 }, { "epoch": 0.43, "learning_rate": 2.573496888888889e-05, "loss": 0.0005, "step": 60000 }, { "epoch": 0.43, "learning_rate": 2.5699413333333335e-05, "loss": 0.0005, "step": 60500 }, { "epoch": 0.43, "learning_rate": 2.566385777777778e-05, "loss": 0.0005, "step": 61000 }, { "epoch": 0.44, "learning_rate": 2.5628302222222223e-05, "loss": 0.0005, "step": 61500 }, { "epoch": 0.44, "learning_rate": 2.5592817777777778e-05, "loss": 0.0005, "step": 62000 }, { "epoch": 0.44, "learning_rate": 2.5557262222222224e-05, "loss": 0.0005, "step": 62500 }, { "epoch": 0.45, "learning_rate": 2.5521706666666666e-05, "loss": 0.0005, "step": 63000 }, { "epoch": 0.45, "learning_rate": 2.548615111111111e-05, "loss": 0.0005, "step": 63500 }, { "epoch": 0.46, "learning_rate": 2.5450595555555557e-05, "loss": 0.0005, "step": 64000 }, { "epoch": 0.46, "learning_rate": 2.5415111111111112e-05, "loss": 0.0005, "step": 64500 }, { "epoch": 0.46, "learning_rate": 2.5379555555555554e-05, "loss": 0.0005, "step": 65000 }, { "epoch": 0.47, "learning_rate": 2.5344e-05, "loss": 0.0005, "step": 65500 }, { "epoch": 0.47, "learning_rate": 2.5308444444444445e-05, "loss": 0.0005, "step": 66000 }, { "epoch": 0.47, "learning_rate": 2.527296e-05, "loss": 0.0005, "step": 66500 }, { "epoch": 0.48, "learning_rate": 2.5237404444444443e-05, "loss": 0.0005, "step": 67000 }, { "epoch": 0.48, "learning_rate": 2.520184888888889e-05, "loss": 0.0005, "step": 67500 }, { "epoch": 0.48, "learning_rate": 2.5166293333333334e-05, "loss": 0.0005, "step": 68000 }, { "epoch": 0.49, "learning_rate": 2.5130808888888893e-05, "loss": 0.0005, "step": 68500 }, { "epoch": 0.49, "learning_rate": 2.5095253333333335e-05, "loss": 0.0005, "step": 69000 }, { "epoch": 0.49, "learning_rate": 2.5059697777777777e-05, "loss": 0.0005, "step": 69500 }, { "epoch": 0.5, "learning_rate": 2.5024142222222222e-05, "loss": 0.0005, "step": 70000 }, { "epoch": 0.5, "learning_rate": 2.4988657777777778e-05, "loss": 0.0005, "step": 70500 }, { "epoch": 0.5, "learning_rate": 2.4953102222222223e-05, "loss": 0.0005, "step": 71000 }, { "epoch": 0.51, "learning_rate": 2.491754666666667e-05, "loss": 0.0005, "step": 71500 }, { "epoch": 0.51, "learning_rate": 2.4881991111111114e-05, "loss": 0.0005, "step": 72000 }, { "epoch": 0.52, "learning_rate": 2.4846506666666666e-05, "loss": 0.0005, "step": 72500 }, { "epoch": 0.52, "learning_rate": 2.481095111111111e-05, "loss": 0.0005, "step": 73000 }, { "epoch": 0.52, "learning_rate": 2.4775395555555557e-05, "loss": 0.0005, "step": 73500 }, { "epoch": 0.53, "learning_rate": 2.4739840000000003e-05, "loss": 0.0005, "step": 74000 }, { "epoch": 0.53, "learning_rate": 2.4704355555555554e-05, "loss": 0.0005, "step": 74500 }, { "epoch": 0.53, "learning_rate": 2.46688e-05, "loss": 0.0005, "step": 75000 }, { "epoch": 0.54, "learning_rate": 2.4633244444444446e-05, "loss": 0.0005, "step": 75500 }, { "epoch": 0.54, "learning_rate": 2.459768888888889e-05, "loss": 0.0005, "step": 76000 }, { "epoch": 0.54, "learning_rate": 2.4562204444444443e-05, "loss": 0.0005, "step": 76500 }, { "epoch": 0.55, "learning_rate": 2.452664888888889e-05, "loss": 0.0005, "step": 77000 }, { "epoch": 0.55, "learning_rate": 2.4491093333333334e-05, "loss": 0.0005, "step": 77500 }, { "epoch": 0.55, "learning_rate": 2.4455537777777776e-05, "loss": 0.0005, "step": 78000 }, { "epoch": 0.56, "learning_rate": 2.4419982222222222e-05, "loss": 0.0005, "step": 78500 }, { "epoch": 0.56, "learning_rate": 2.438449777777778e-05, "loss": 0.0005, "step": 79000 }, { "epoch": 0.57, "learning_rate": 2.4348942222222222e-05, "loss": 0.0005, "step": 79500 }, { "epoch": 0.57, "learning_rate": 2.4313386666666665e-05, "loss": 0.0005, "step": 80000 }, { "epoch": 0.57, "learning_rate": 2.427783111111111e-05, "loss": 0.0005, "step": 80500 }, { "epoch": 0.58, "learning_rate": 2.424234666666667e-05, "loss": 0.0005, "step": 81000 }, { "epoch": 0.58, "learning_rate": 2.4206791111111114e-05, "loss": 0.0005, "step": 81500 }, { "epoch": 0.58, "learning_rate": 2.4171235555555556e-05, "loss": 0.0005, "step": 82000 }, { "epoch": 0.59, "learning_rate": 2.4135680000000002e-05, "loss": 0.0005, "step": 82500 }, { "epoch": 0.59, "learning_rate": 2.4100124444444444e-05, "loss": 0.0005, "step": 83000 }, { "epoch": 0.59, "learning_rate": 2.4064640000000003e-05, "loss": 0.0005, "step": 83500 }, { "epoch": 0.6, "learning_rate": 2.4029084444444445e-05, "loss": 0.0005, "step": 84000 }, { "epoch": 0.6, "learning_rate": 2.399352888888889e-05, "loss": 0.0005, "step": 84500 }, { "epoch": 0.6, "learning_rate": 2.3957973333333336e-05, "loss": 0.0005, "step": 85000 }, { "epoch": 0.61, "learning_rate": 2.3922488888888888e-05, "loss": 0.0005, "step": 85500 }, { "epoch": 0.61, "learning_rate": 2.3886933333333333e-05, "loss": 0.0005, "step": 86000 }, { "epoch": 0.62, "learning_rate": 2.385137777777778e-05, "loss": 0.0005, "step": 86500 }, { "epoch": 0.62, "learning_rate": 2.3815822222222224e-05, "loss": 0.0005, "step": 87000 }, { "epoch": 0.62, "learning_rate": 2.3780337777777776e-05, "loss": 0.0005, "step": 87500 }, { "epoch": 0.63, "learning_rate": 2.3744782222222222e-05, "loss": 0.0005, "step": 88000 }, { "epoch": 0.63, "learning_rate": 2.3709226666666667e-05, "loss": 0.0005, "step": 88500 }, { "epoch": 0.63, "learning_rate": 2.3673671111111113e-05, "loss": 0.0005, "step": 89000 }, { "epoch": 0.64, "learning_rate": 2.3638186666666665e-05, "loss": 0.0005, "step": 89500 }, { "epoch": 0.64, "learning_rate": 2.360263111111111e-05, "loss": 0.0005, "step": 90000 }, { "epoch": 0.64, "learning_rate": 2.3567075555555556e-05, "loss": 0.0005, "step": 90500 }, { "epoch": 0.65, "learning_rate": 2.353152e-05, "loss": 0.0005, "step": 91000 }, { "epoch": 0.65, "learning_rate": 2.3495964444444443e-05, "loss": 0.0005, "step": 91500 }, { "epoch": 0.65, "learning_rate": 2.3460480000000002e-05, "loss": 0.0005, "step": 92000 }, { "epoch": 0.66, "learning_rate": 2.3424924444444448e-05, "loss": 0.0005, "step": 92500 }, { "epoch": 0.66, "learning_rate": 2.3389368888888886e-05, "loss": 0.0005, "step": 93000 }, { "epoch": 0.66, "learning_rate": 2.3353813333333332e-05, "loss": 0.0005, "step": 93500 }, { "epoch": 0.67, "learning_rate": 2.331832888888889e-05, "loss": 0.0005, "step": 94000 }, { "epoch": 0.67, "learning_rate": 2.3282773333333336e-05, "loss": 0.0005, "step": 94500 }, { "epoch": 0.68, "learning_rate": 2.3247217777777778e-05, "loss": 0.0005, "step": 95000 }, { "epoch": 0.68, "learning_rate": 2.3211662222222224e-05, "loss": 0.0004, "step": 95500 }, { "epoch": 0.68, "learning_rate": 2.317610666666667e-05, "loss": 0.0005, "step": 96000 }, { "epoch": 0.69, "learning_rate": 2.3140622222222225e-05, "loss": 0.0004, "step": 96500 }, { "epoch": 0.69, "learning_rate": 2.3105066666666667e-05, "loss": 0.0005, "step": 97000 }, { "epoch": 0.69, "learning_rate": 2.3069511111111112e-05, "loss": 0.0005, "step": 97500 }, { "epoch": 0.7, "learning_rate": 2.3033955555555558e-05, "loss": 0.0005, "step": 98000 }, { "epoch": 0.7, "learning_rate": 2.2998400000000003e-05, "loss": 0.0005, "step": 98500 }, { "epoch": 0.7, "learning_rate": 2.2962915555555555e-05, "loss": 0.0005, "step": 99000 }, { "epoch": 0.71, "learning_rate": 2.292736e-05, "loss": 0.0005, "step": 99500 }, { "epoch": 0.71, "learning_rate": 2.2891804444444446e-05, "loss": 0.0005, "step": 100000 }, { "epoch": 0.71, "learning_rate": 2.285624888888889e-05, "loss": 0.0004, "step": 100500 }, { "epoch": 0.72, "learning_rate": 2.2820764444444444e-05, "loss": 0.0005, "step": 101000 }, { "epoch": 0.72, "learning_rate": 2.278520888888889e-05, "loss": 0.0005, "step": 101500 }, { "epoch": 0.73, "learning_rate": 2.2749653333333335e-05, "loss": 0.0005, "step": 102000 }, { "epoch": 0.73, "learning_rate": 2.2714097777777777e-05, "loss": 0.0004, "step": 102500 }, { "epoch": 0.73, "learning_rate": 2.2678542222222222e-05, "loss": 0.0004, "step": 103000 }, { "epoch": 0.74, "learning_rate": 2.2643057777777778e-05, "loss": 0.0004, "step": 103500 }, { "epoch": 0.74, "learning_rate": 2.2607502222222223e-05, "loss": 0.0004, "step": 104000 }, { "epoch": 0.74, "learning_rate": 2.2571946666666665e-05, "loss": 0.0005, "step": 104500 }, { "epoch": 0.75, "learning_rate": 2.253639111111111e-05, "loss": 0.0004, "step": 105000 }, { "epoch": 0.75, "learning_rate": 2.250090666666667e-05, "loss": 0.0005, "step": 105500 }, { "epoch": 0.75, "learning_rate": 2.246535111111111e-05, "loss": 0.0004, "step": 106000 }, { "epoch": 0.76, "learning_rate": 2.2429795555555554e-05, "loss": 0.0005, "step": 106500 }, { "epoch": 0.76, "learning_rate": 2.239424e-05, "loss": 0.0004, "step": 107000 }, { "epoch": 0.76, "learning_rate": 2.2358755555555558e-05, "loss": 0.0005, "step": 107500 }, { "epoch": 0.77, "learning_rate": 2.23232e-05, "loss": 0.0005, "step": 108000 }, { "epoch": 0.77, "learning_rate": 2.2287644444444446e-05, "loss": 0.0004, "step": 108500 }, { "epoch": 0.78, "learning_rate": 2.225208888888889e-05, "loss": 0.0004, "step": 109000 }, { "epoch": 0.78, "learning_rate": 2.2216604444444446e-05, "loss": 0.0004, "step": 109500 }, { "epoch": 0.78, "learning_rate": 2.218104888888889e-05, "loss": 0.0004, "step": 110000 }, { "epoch": 0.79, "learning_rate": 2.2145493333333334e-05, "loss": 0.0004, "step": 110500 }, { "epoch": 0.79, "learning_rate": 2.210993777777778e-05, "loss": 0.0004, "step": 111000 }, { "epoch": 0.79, "learning_rate": 2.2074382222222225e-05, "loss": 0.0004, "step": 111500 }, { "epoch": 0.8, "learning_rate": 2.2038897777777777e-05, "loss": 0.0004, "step": 112000 }, { "epoch": 0.8, "learning_rate": 2.2003342222222222e-05, "loss": 0.0004, "step": 112500 }, { "epoch": 0.8, "learning_rate": 2.1967786666666668e-05, "loss": 0.0005, "step": 113000 }, { "epoch": 0.81, "learning_rate": 2.193223111111111e-05, "loss": 0.0004, "step": 113500 }, { "epoch": 0.81, "learning_rate": 2.1896675555555556e-05, "loss": 0.0004, "step": 114000 }, { "epoch": 0.81, "learning_rate": 2.186112e-05, "loss": 0.0004, "step": 114500 }, { "epoch": 0.82, "learning_rate": 2.1825635555555556e-05, "loss": 0.0004, "step": 115000 }, { "epoch": 0.82, "learning_rate": 2.179008e-05, "loss": 0.0004, "step": 115500 }, { "epoch": 0.82, "learning_rate": 2.1754524444444444e-05, "loss": 0.0004, "step": 116000 }, { "epoch": 0.83, "learning_rate": 2.171896888888889e-05, "loss": 0.0004, "step": 116500 }, { "epoch": 0.83, "learning_rate": 2.1683484444444445e-05, "loss": 0.0004, "step": 117000 }, { "epoch": 0.84, "learning_rate": 2.1647928888888887e-05, "loss": 0.0004, "step": 117500 }, { "epoch": 0.84, "learning_rate": 2.1612373333333333e-05, "loss": 0.0004, "step": 118000 }, { "epoch": 0.84, "learning_rate": 2.1576817777777778e-05, "loss": 0.0004, "step": 118500 }, { "epoch": 0.85, "learning_rate": 2.1541333333333337e-05, "loss": 0.0004, "step": 119000 }, { "epoch": 0.85, "learning_rate": 2.150577777777778e-05, "loss": 0.0004, "step": 119500 }, { "epoch": 0.85, "learning_rate": 2.1470222222222224e-05, "loss": 0.0004, "step": 120000 }, { "epoch": 0.86, "learning_rate": 2.1434666666666667e-05, "loss": 0.0004, "step": 120500 }, { "epoch": 0.86, "learning_rate": 2.1399182222222222e-05, "loss": 0.0004, "step": 121000 }, { "epoch": 0.86, "learning_rate": 2.1363626666666667e-05, "loss": 0.0004, "step": 121500 }, { "epoch": 0.87, "learning_rate": 2.1328071111111113e-05, "loss": 0.0004, "step": 122000 }, { "epoch": 0.87, "learning_rate": 2.129251555555556e-05, "loss": 0.0004, "step": 122500 }, { "epoch": 0.87, "learning_rate": 2.125696e-05, "loss": 0.0004, "step": 123000 }, { "epoch": 0.88, "learning_rate": 2.1221475555555556e-05, "loss": 0.0004, "step": 123500 }, { "epoch": 0.88, "learning_rate": 2.118592e-05, "loss": 0.0004, "step": 124000 }, { "epoch": 0.89, "learning_rate": 2.1150364444444447e-05, "loss": 0.0004, "step": 124500 }, { "epoch": 0.89, "learning_rate": 2.111480888888889e-05, "loss": 0.0004, "step": 125000 }, { "epoch": 0.89, "learning_rate": 2.1079324444444444e-05, "loss": 0.0004, "step": 125500 }, { "epoch": 0.9, "learning_rate": 2.104376888888889e-05, "loss": 0.0004, "step": 126000 }, { "epoch": 0.9, "learning_rate": 2.1008213333333335e-05, "loss": 0.0004, "step": 126500 }, { "epoch": 0.9, "learning_rate": 2.0972657777777777e-05, "loss": 0.0004, "step": 127000 }, { "epoch": 0.91, "learning_rate": 2.0937102222222223e-05, "loss": 0.0004, "step": 127500 }, { "epoch": 0.91, "learning_rate": 2.0901617777777778e-05, "loss": 0.0004, "step": 128000 }, { "epoch": 0.91, "learning_rate": 2.086606222222222e-05, "loss": 0.0004, "step": 128500 }, { "epoch": 0.92, "learning_rate": 2.0830506666666666e-05, "loss": 0.0004, "step": 129000 }, { "epoch": 0.92, "learning_rate": 2.079495111111111e-05, "loss": 0.0004, "step": 129500 }, { "epoch": 0.92, "learning_rate": 2.0759395555555557e-05, "loss": 0.0004, "step": 130000 }, { "epoch": 0.93, "learning_rate": 2.072391111111111e-05, "loss": 0.0004, "step": 130500 }, { "epoch": 0.93, "learning_rate": 2.0688355555555554e-05, "loss": 0.0004, "step": 131000 }, { "epoch": 0.94, "learning_rate": 2.06528e-05, "loss": 0.0004, "step": 131500 }, { "epoch": 0.94, "learning_rate": 2.0617244444444445e-05, "loss": 0.0004, "step": 132000 }, { "epoch": 0.94, "learning_rate": 2.0581688888888888e-05, "loss": 0.0004, "step": 132500 }, { "epoch": 0.95, "learning_rate": 2.0546204444444446e-05, "loss": 0.0004, "step": 133000 }, { "epoch": 0.95, "learning_rate": 2.0510648888888892e-05, "loss": 0.0004, "step": 133500 }, { "epoch": 0.95, "learning_rate": 2.0475093333333334e-05, "loss": 0.0004, "step": 134000 }, { "epoch": 0.96, "learning_rate": 2.0439537777777776e-05, "loss": 0.0004, "step": 134500 }, { "epoch": 0.96, "learning_rate": 2.040398222222222e-05, "loss": 0.0004, "step": 135000 }, { "epoch": 0.96, "learning_rate": 2.0368426666666667e-05, "loss": 0.0004, "step": 135500 }, { "epoch": 0.97, "learning_rate": 2.0332942222222222e-05, "loss": 0.0004, "step": 136000 }, { "epoch": 0.97, "learning_rate": 2.0297386666666668e-05, "loss": 0.0004, "step": 136500 }, { "epoch": 0.97, "learning_rate": 2.0261831111111113e-05, "loss": 0.0004, "step": 137000 }, { "epoch": 0.98, "learning_rate": 2.0226275555555556e-05, "loss": 0.0004, "step": 137500 }, { "epoch": 0.98, "learning_rate": 2.019079111111111e-05, "loss": 0.0004, "step": 138000 }, { "epoch": 0.98, "learning_rate": 2.0155235555555556e-05, "loss": 0.0004, "step": 138500 }, { "epoch": 0.99, "learning_rate": 2.0119680000000002e-05, "loss": 0.0004, "step": 139000 }, { "epoch": 0.99, "learning_rate": 2.0084124444444447e-05, "loss": 0.0004, "step": 139500 }, { "epoch": 1.0, "learning_rate": 2.004856888888889e-05, "loss": 0.0004, "step": 140000 }, { "epoch": 1.0, "learning_rate": 2.0013084444444445e-05, "loss": 0.0004, "step": 140500 }, { "epoch": 1.0, "learning_rate": 1.997752888888889e-05, "loss": 0.0004, "step": 141000 }, { "epoch": 1.01, "learning_rate": 1.9941973333333336e-05, "loss": 0.0004, "step": 141500 }, { "epoch": 1.01, "learning_rate": 1.9906417777777778e-05, "loss": 0.0004, "step": 142000 }, { "epoch": 1.01, "learning_rate": 1.9870862222222223e-05, "loss": 0.0004, "step": 142500 }, { "epoch": 1.02, "learning_rate": 1.983530666666667e-05, "loss": 0.0004, "step": 143000 }, { "epoch": 1.02, "learning_rate": 1.9799751111111115e-05, "loss": 0.0004, "step": 143500 }, { "epoch": 1.02, "learning_rate": 1.9764195555555557e-05, "loss": 0.0004, "step": 144000 }, { "epoch": 1.03, "learning_rate": 1.9728711111111112e-05, "loss": 0.0004, "step": 144500 }, { "epoch": 1.03, "learning_rate": 1.9693155555555557e-05, "loss": 0.0004, "step": 145000 }, { "epoch": 1.03, "learning_rate": 1.96576e-05, "loss": 0.0004, "step": 145500 }, { "epoch": 1.04, "learning_rate": 1.9622044444444445e-05, "loss": 0.0004, "step": 146000 }, { "epoch": 1.04, "learning_rate": 1.958648888888889e-05, "loss": 0.0004, "step": 146500 }, { "epoch": 1.05, "learning_rate": 1.9550933333333336e-05, "loss": 0.0004, "step": 147000 }, { "epoch": 1.05, "learning_rate": 1.951537777777778e-05, "loss": 0.0004, "step": 147500 }, { "epoch": 1.05, "learning_rate": 1.9479893333333334e-05, "loss": 0.0004, "step": 148000 }, { "epoch": 1.06, "learning_rate": 1.944433777777778e-05, "loss": 0.0004, "step": 148500 }, { "epoch": 1.06, "learning_rate": 1.9408782222222225e-05, "loss": 0.0004, "step": 149000 }, { "epoch": 1.06, "learning_rate": 1.9373226666666667e-05, "loss": 0.0004, "step": 149500 }, { "epoch": 1.07, "learning_rate": 1.9337742222222222e-05, "loss": 0.0004, "step": 150000 }, { "epoch": 1.07, "learning_rate": 1.9302186666666668e-05, "loss": 0.0004, "step": 150500 }, { "epoch": 1.07, "learning_rate": 1.9266631111111113e-05, "loss": 0.0004, "step": 151000 }, { "epoch": 1.08, "learning_rate": 1.9231075555555555e-05, "loss": 0.0004, "step": 151500 }, { "epoch": 1.08, "learning_rate": 1.919552e-05, "loss": 0.0004, "step": 152000 }, { "epoch": 1.08, "learning_rate": 1.9160035555555556e-05, "loss": 0.0004, "step": 152500 }, { "epoch": 1.09, "learning_rate": 1.9124479999999998e-05, "loss": 0.0004, "step": 153000 }, { "epoch": 1.09, "learning_rate": 1.9088924444444444e-05, "loss": 0.0004, "step": 153500 }, { "epoch": 1.1, "learning_rate": 1.905336888888889e-05, "loss": 0.0004, "step": 154000 }, { "epoch": 1.1, "learning_rate": 1.9017813333333335e-05, "loss": 0.0004, "step": 154500 }, { "epoch": 1.1, "learning_rate": 1.8982257777777777e-05, "loss": 0.0004, "step": 155000 }, { "epoch": 1.11, "learning_rate": 1.8946702222222223e-05, "loss": 0.0004, "step": 155500 }, { "epoch": 1.11, "learning_rate": 1.8911146666666668e-05, "loss": 0.0004, "step": 156000 }, { "epoch": 1.11, "learning_rate": 1.8875662222222223e-05, "loss": 0.0004, "step": 156500 }, { "epoch": 1.12, "learning_rate": 1.884017777777778e-05, "loss": 0.0004, "step": 157000 }, { "epoch": 1.12, "learning_rate": 1.8804622222222224e-05, "loss": 0.0004, "step": 157500 }, { "epoch": 1.12, "learning_rate": 1.8769066666666666e-05, "loss": 0.0004, "step": 158000 }, { "epoch": 1.13, "learning_rate": 1.8733511111111112e-05, "loss": 0.0004, "step": 158500 }, { "epoch": 1.13, "learning_rate": 1.8698026666666667e-05, "loss": 0.0004, "step": 159000 }, { "epoch": 1.13, "learning_rate": 1.8662471111111112e-05, "loss": 0.0004, "step": 159500 }, { "epoch": 1.14, "learning_rate": 1.8626915555555558e-05, "loss": 0.0004, "step": 160000 }, { "epoch": 1.14, "learning_rate": 1.859136e-05, "loss": 0.0004, "step": 160500 }, { "epoch": 1.14, "learning_rate": 1.8555875555555555e-05, "loss": 0.0004, "step": 161000 }, { "epoch": 1.15, "learning_rate": 1.852032e-05, "loss": 0.0004, "step": 161500 }, { "epoch": 1.15, "learning_rate": 1.8484764444444446e-05, "loss": 0.0003, "step": 162000 }, { "epoch": 1.16, "learning_rate": 1.844920888888889e-05, "loss": 0.0004, "step": 162500 }, { "epoch": 1.16, "learning_rate": 1.8413653333333334e-05, "loss": 0.0004, "step": 163000 }, { "epoch": 1.16, "learning_rate": 1.837816888888889e-05, "loss": 0.0004, "step": 163500 }, { "epoch": 1.17, "learning_rate": 1.8342613333333335e-05, "loss": 0.0004, "step": 164000 }, { "epoch": 1.17, "learning_rate": 1.8307057777777777e-05, "loss": 0.0004, "step": 164500 }, { "epoch": 1.17, "learning_rate": 1.8271502222222223e-05, "loss": 0.0004, "step": 165000 }, { "epoch": 1.18, "learning_rate": 1.8235946666666668e-05, "loss": 0.0004, "step": 165500 }, { "epoch": 1.18, "learning_rate": 1.8200391111111114e-05, "loss": 0.0004, "step": 166000 }, { "epoch": 1.18, "learning_rate": 1.8164906666666666e-05, "loss": 0.0004, "step": 166500 }, { "epoch": 1.19, "learning_rate": 1.812935111111111e-05, "loss": 0.0004, "step": 167000 }, { "epoch": 1.19, "learning_rate": 1.8093795555555557e-05, "loss": 0.0004, "step": 167500 }, { "epoch": 1.19, "learning_rate": 1.805824e-05, "loss": 0.0003, "step": 168000 }, { "epoch": 1.2, "learning_rate": 1.8022684444444444e-05, "loss": 0.0004, "step": 168500 }, { "epoch": 1.2, "learning_rate": 1.798712888888889e-05, "loss": 0.0004, "step": 169000 }, { "epoch": 1.21, "learning_rate": 1.7951573333333335e-05, "loss": 0.0004, "step": 169500 }, { "epoch": 1.21, "learning_rate": 1.7916017777777777e-05, "loss": 0.0004, "step": 170000 }, { "epoch": 1.21, "learning_rate": 1.7880533333333333e-05, "loss": 0.0004, "step": 170500 }, { "epoch": 1.22, "learning_rate": 1.784504888888889e-05, "loss": 0.0004, "step": 171000 }, { "epoch": 1.22, "learning_rate": 1.7809493333333337e-05, "loss": 0.0004, "step": 171500 }, { "epoch": 1.22, "learning_rate": 1.7773937777777776e-05, "loss": 0.0004, "step": 172000 }, { "epoch": 1.23, "learning_rate": 1.773838222222222e-05, "loss": 0.0004, "step": 172500 }, { "epoch": 1.23, "learning_rate": 1.7702826666666667e-05, "loss": 0.0004, "step": 173000 }, { "epoch": 1.23, "learning_rate": 1.7667342222222222e-05, "loss": 0.0004, "step": 173500 }, { "epoch": 1.24, "learning_rate": 1.7631786666666667e-05, "loss": 0.0004, "step": 174000 }, { "epoch": 1.24, "learning_rate": 1.7596231111111113e-05, "loss": 0.0004, "step": 174500 }, { "epoch": 1.24, "learning_rate": 1.756067555555556e-05, "loss": 0.0004, "step": 175000 }, { "epoch": 1.25, "learning_rate": 1.7525119999999997e-05, "loss": 0.0004, "step": 175500 }, { "epoch": 1.25, "learning_rate": 1.7489635555555556e-05, "loss": 0.0004, "step": 176000 }, { "epoch": 1.26, "learning_rate": 1.745408e-05, "loss": 0.0003, "step": 176500 }, { "epoch": 1.26, "learning_rate": 1.7418524444444447e-05, "loss": 0.0004, "step": 177000 }, { "epoch": 1.26, "learning_rate": 1.738296888888889e-05, "loss": 0.0003, "step": 177500 }, { "epoch": 1.27, "learning_rate": 1.7347413333333335e-05, "loss": 0.0004, "step": 178000 }, { "epoch": 1.27, "learning_rate": 1.731192888888889e-05, "loss": 0.0004, "step": 178500 }, { "epoch": 1.27, "learning_rate": 1.7276373333333335e-05, "loss": 0.0003, "step": 179000 }, { "epoch": 1.28, "learning_rate": 1.7240817777777778e-05, "loss": 0.0003, "step": 179500 }, { "epoch": 1.28, "learning_rate": 1.7205262222222223e-05, "loss": 0.0003, "step": 180000 }, { "epoch": 1.28, "learning_rate": 1.716970666666667e-05, "loss": 0.0004, "step": 180500 }, { "epoch": 1.29, "learning_rate": 1.7134151111111114e-05, "loss": 0.0004, "step": 181000 }, { "epoch": 1.29, "learning_rate": 1.7098666666666666e-05, "loss": 0.0004, "step": 181500 }, { "epoch": 1.29, "learning_rate": 1.706311111111111e-05, "loss": 0.0004, "step": 182000 }, { "epoch": 1.3, "learning_rate": 1.7027555555555557e-05, "loss": 0.0003, "step": 182500 }, { "epoch": 1.3, "learning_rate": 1.6992e-05, "loss": 0.0003, "step": 183000 }, { "epoch": 1.3, "learning_rate": 1.6956444444444445e-05, "loss": 0.0003, "step": 183500 }, { "epoch": 1.31, "learning_rate": 1.692088888888889e-05, "loss": 0.0003, "step": 184000 }, { "epoch": 1.31, "learning_rate": 1.6885404444444446e-05, "loss": 0.0003, "step": 184500 }, { "epoch": 1.32, "learning_rate": 1.6849848888888888e-05, "loss": 0.0003, "step": 185000 }, { "epoch": 1.32, "learning_rate": 1.6814293333333333e-05, "loss": 0.0003, "step": 185500 }, { "epoch": 1.32, "learning_rate": 1.677873777777778e-05, "loss": 0.0003, "step": 186000 }, { "epoch": 1.33, "learning_rate": 1.6743182222222224e-05, "loss": 0.0004, "step": 186500 }, { "epoch": 1.33, "learning_rate": 1.6707626666666666e-05, "loss": 0.0003, "step": 187000 }, { "epoch": 1.33, "learning_rate": 1.6672071111111112e-05, "loss": 0.0003, "step": 187500 }, { "epoch": 1.34, "learning_rate": 1.6636515555555558e-05, "loss": 0.0003, "step": 188000 }, { "epoch": 1.34, "learning_rate": 1.6601102222222226e-05, "loss": 0.0003, "step": 188500 }, { "epoch": 1.34, "learning_rate": 1.6565546666666665e-05, "loss": 0.0003, "step": 189000 }, { "epoch": 1.35, "learning_rate": 1.652999111111111e-05, "loss": 0.0003, "step": 189500 }, { "epoch": 1.35, "learning_rate": 1.6494435555555556e-05, "loss": 0.0003, "step": 190000 }, { "epoch": 1.35, "learning_rate": 1.6458879999999998e-05, "loss": 0.0003, "step": 190500 }, { "epoch": 1.36, "learning_rate": 1.6423324444444443e-05, "loss": 0.0003, "step": 191000 }, { "epoch": 1.36, "learning_rate": 1.638776888888889e-05, "loss": 0.0003, "step": 191500 }, { "epoch": 1.37, "learning_rate": 1.6352284444444448e-05, "loss": 0.0003, "step": 192000 }, { "epoch": 1.37, "learning_rate": 1.6316728888888886e-05, "loss": 0.0003, "step": 192500 }, { "epoch": 1.37, "learning_rate": 1.6281173333333332e-05, "loss": 0.0003, "step": 193000 }, { "epoch": 1.38, "learning_rate": 1.6245617777777777e-05, "loss": 0.0003, "step": 193500 }, { "epoch": 1.38, "learning_rate": 1.6210062222222223e-05, "loss": 0.0003, "step": 194000 }, { "epoch": 1.38, "learning_rate": 1.6174577777777778e-05, "loss": 0.0003, "step": 194500 }, { "epoch": 1.39, "learning_rate": 1.6139022222222224e-05, "loss": 0.0003, "step": 195000 }, { "epoch": 1.39, "learning_rate": 1.610346666666667e-05, "loss": 0.0003, "step": 195500 }, { "epoch": 1.39, "learning_rate": 1.606791111111111e-05, "loss": 0.0003, "step": 196000 }, { "epoch": 1.4, "learning_rate": 1.6032355555555553e-05, "loss": 0.0003, "step": 196500 }, { "epoch": 1.4, "learning_rate": 1.5996871111111112e-05, "loss": 0.0003, "step": 197000 }, { "epoch": 1.4, "learning_rate": 1.5961315555555558e-05, "loss": 0.0003, "step": 197500 }, { "epoch": 1.41, "learning_rate": 1.592576e-05, "loss": 0.0003, "step": 198000 }, { "epoch": 1.41, "learning_rate": 1.5890204444444445e-05, "loss": 0.0003, "step": 198500 }, { "epoch": 1.42, "learning_rate": 1.585464888888889e-05, "loss": 0.0003, "step": 199000 }, { "epoch": 1.42, "learning_rate": 1.5819093333333333e-05, "loss": 0.0003, "step": 199500 }, { "epoch": 1.42, "learning_rate": 1.5783608888888888e-05, "loss": 0.0003, "step": 200000 }, { "epoch": 1.43, "learning_rate": 1.5748053333333334e-05, "loss": 0.0003, "step": 200500 }, { "epoch": 1.43, "learning_rate": 1.571249777777778e-05, "loss": 0.0003, "step": 201000 }, { "epoch": 1.43, "learning_rate": 1.5676942222222225e-05, "loss": 0.0003, "step": 201500 }, { "epoch": 1.44, "learning_rate": 1.5641386666666667e-05, "loss": 0.0003, "step": 202000 }, { "epoch": 1.44, "learning_rate": 1.5605831111111113e-05, "loss": 0.0003, "step": 202500 }, { "epoch": 1.44, "learning_rate": 1.5570275555555558e-05, "loss": 0.0003, "step": 203000 }, { "epoch": 1.45, "learning_rate": 1.553472e-05, "loss": 0.0003, "step": 203500 }, { "epoch": 1.45, "learning_rate": 1.5499235555555555e-05, "loss": 0.0003, "step": 204000 }, { "epoch": 1.45, "learning_rate": 1.546375111111111e-05, "loss": 0.0003, "step": 204500 }, { "epoch": 1.46, "learning_rate": 1.5428195555555556e-05, "loss": 0.0003, "step": 205000 }, { "epoch": 1.46, "learning_rate": 1.539264e-05, "loss": 0.0003, "step": 205500 }, { "epoch": 1.46, "learning_rate": 1.5357084444444444e-05, "loss": 0.0003, "step": 206000 }, { "epoch": 1.47, "learning_rate": 1.532152888888889e-05, "loss": 0.0003, "step": 206500 }, { "epoch": 1.47, "learning_rate": 1.5285973333333335e-05, "loss": 0.0003, "step": 207000 }, { "epoch": 1.48, "learning_rate": 1.5250417777777779e-05, "loss": 0.0003, "step": 207500 }, { "epoch": 1.48, "learning_rate": 1.5214862222222223e-05, "loss": 0.0003, "step": 208000 }, { "epoch": 1.48, "learning_rate": 1.5179377777777778e-05, "loss": 0.0003, "step": 208500 }, { "epoch": 1.49, "learning_rate": 1.5143822222222222e-05, "loss": 0.0003, "step": 209000 }, { "epoch": 1.49, "learning_rate": 1.510833777777778e-05, "loss": 0.0003, "step": 209500 }, { "epoch": 1.49, "learning_rate": 1.507278222222222e-05, "loss": 0.0003, "step": 210000 }, { "epoch": 1.5, "learning_rate": 1.5037226666666666e-05, "loss": 0.0003, "step": 210500 }, { "epoch": 1.5, "learning_rate": 1.500167111111111e-05, "loss": 0.0003, "step": 211000 }, { "epoch": 1.5, "learning_rate": 1.4966115555555556e-05, "loss": 0.0003, "step": 211500 }, { "epoch": 1.51, "learning_rate": 1.4930631111111111e-05, "loss": 0.0003, "step": 212000 }, { "epoch": 1.51, "learning_rate": 1.4895075555555556e-05, "loss": 0.0003, "step": 212500 }, { "epoch": 1.51, "learning_rate": 1.485952e-05, "loss": 0.0003, "step": 213000 }, { "epoch": 1.52, "learning_rate": 1.4823964444444444e-05, "loss": 0.0003, "step": 213500 }, { "epoch": 1.52, "learning_rate": 1.478840888888889e-05, "loss": 0.0003, "step": 214000 }, { "epoch": 1.53, "learning_rate": 1.4752853333333334e-05, "loss": 0.0003, "step": 214500 }, { "epoch": 1.53, "learning_rate": 1.4717297777777779e-05, "loss": 0.0003, "step": 215000 }, { "epoch": 1.53, "learning_rate": 1.4681742222222221e-05, "loss": 0.0003, "step": 215500 }, { "epoch": 1.54, "learning_rate": 1.4646257777777778e-05, "loss": 0.0003, "step": 216000 }, { "epoch": 1.54, "learning_rate": 1.4610773333333333e-05, "loss": 0.0003, "step": 216500 }, { "epoch": 1.54, "learning_rate": 1.4575217777777779e-05, "loss": 0.0003, "step": 217000 }, { "epoch": 1.55, "learning_rate": 1.4539662222222223e-05, "loss": 0.0003, "step": 217500 }, { "epoch": 1.55, "learning_rate": 1.4504106666666668e-05, "loss": 0.0003, "step": 218000 }, { "epoch": 1.55, "learning_rate": 1.446855111111111e-05, "loss": 0.0003, "step": 218500 }, { "epoch": 1.56, "learning_rate": 1.4432995555555556e-05, "loss": 0.0003, "step": 219000 }, { "epoch": 1.56, "learning_rate": 1.439744e-05, "loss": 0.0003, "step": 219500 }, { "epoch": 1.56, "learning_rate": 1.4361884444444445e-05, "loss": 0.0003, "step": 220000 }, { "epoch": 1.57, "learning_rate": 1.43264e-05, "loss": 0.0003, "step": 220500 }, { "epoch": 1.57, "learning_rate": 1.4290844444444444e-05, "loss": 0.0003, "step": 221000 }, { "epoch": 1.58, "learning_rate": 1.425528888888889e-05, "loss": 0.0003, "step": 221500 }, { "epoch": 1.58, "learning_rate": 1.4219733333333332e-05, "loss": 0.0003, "step": 222000 }, { "epoch": 1.58, "learning_rate": 1.4184248888888889e-05, "loss": 0.0003, "step": 222500 }, { "epoch": 1.59, "learning_rate": 1.4148693333333333e-05, "loss": 0.0003, "step": 223000 }, { "epoch": 1.59, "learning_rate": 1.4113137777777778e-05, "loss": 0.0003, "step": 223500 }, { "epoch": 1.59, "learning_rate": 1.4077582222222222e-05, "loss": 0.0003, "step": 224000 }, { "epoch": 1.6, "learning_rate": 1.404209777777778e-05, "loss": 0.0003, "step": 224500 }, { "epoch": 1.6, "learning_rate": 1.4006542222222223e-05, "loss": 0.0003, "step": 225000 }, { "epoch": 1.6, "learning_rate": 1.3970986666666667e-05, "loss": 0.0003, "step": 225500 }, { "epoch": 1.61, "learning_rate": 1.393543111111111e-05, "loss": 0.0003, "step": 226000 }, { "epoch": 1.61, "learning_rate": 1.3899946666666668e-05, "loss": 0.0003, "step": 226500 }, { "epoch": 1.61, "learning_rate": 1.3864391111111111e-05, "loss": 0.0003, "step": 227000 }, { "epoch": 1.62, "learning_rate": 1.3828835555555557e-05, "loss": 0.0003, "step": 227500 }, { "epoch": 1.62, "learning_rate": 1.3793351111111112e-05, "loss": 0.0003, "step": 228000 }, { "epoch": 1.62, "learning_rate": 1.3757795555555554e-05, "loss": 0.0003, "step": 228500 }, { "epoch": 1.63, "learning_rate": 1.372224e-05, "loss": 0.0003, "step": 229000 }, { "epoch": 1.63, "learning_rate": 1.3686684444444444e-05, "loss": 0.0003, "step": 229500 }, { "epoch": 1.64, "learning_rate": 1.365112888888889e-05, "loss": 0.0003, "step": 230000 }, { "epoch": 1.64, "learning_rate": 1.3615644444444445e-05, "loss": 0.0003, "step": 230500 }, { "epoch": 1.64, "learning_rate": 1.358008888888889e-05, "loss": 0.0003, "step": 231000 }, { "epoch": 1.65, "learning_rate": 1.3544533333333334e-05, "loss": 0.0003, "step": 231500 }, { "epoch": 1.65, "learning_rate": 1.3508977777777778e-05, "loss": 0.0003, "step": 232000 }, { "epoch": 1.65, "learning_rate": 1.3473422222222222e-05, "loss": 0.0003, "step": 232500 }, { "epoch": 1.66, "learning_rate": 1.3437937777777779e-05, "loss": 0.0003, "step": 233000 }, { "epoch": 1.66, "learning_rate": 1.3402382222222222e-05, "loss": 0.0003, "step": 233500 }, { "epoch": 1.66, "learning_rate": 1.3366826666666668e-05, "loss": 0.0003, "step": 234000 }, { "epoch": 1.67, "learning_rate": 1.3331271111111112e-05, "loss": 0.0003, "step": 234500 }, { "epoch": 1.67, "learning_rate": 1.3295715555555557e-05, "loss": 0.0003, "step": 235000 }, { "epoch": 1.67, "learning_rate": 1.326016e-05, "loss": 0.0003, "step": 235500 }, { "epoch": 1.68, "learning_rate": 1.3224604444444445e-05, "loss": 0.0003, "step": 236000 }, { "epoch": 1.68, "learning_rate": 1.3189048888888889e-05, "loss": 0.0003, "step": 236500 }, { "epoch": 1.69, "learning_rate": 1.3153564444444444e-05, "loss": 0.0003, "step": 237000 }, { "epoch": 1.69, "learning_rate": 1.311800888888889e-05, "loss": 0.0003, "step": 237500 }, { "epoch": 1.69, "learning_rate": 1.3082453333333333e-05, "loss": 0.0003, "step": 238000 }, { "epoch": 1.7, "learning_rate": 1.3046897777777779e-05, "loss": 0.0003, "step": 238500 }, { "epoch": 1.7, "learning_rate": 1.3011413333333332e-05, "loss": 0.0003, "step": 239000 }, { "epoch": 1.7, "learning_rate": 1.2975857777777778e-05, "loss": 0.0003, "step": 239500 }, { "epoch": 1.71, "learning_rate": 1.2940302222222222e-05, "loss": 0.0003, "step": 240000 }, { "epoch": 1.71, "learning_rate": 1.2904746666666667e-05, "loss": 0.0003, "step": 240500 }, { "epoch": 1.71, "learning_rate": 1.2869333333333334e-05, "loss": 0.0003, "step": 241000 }, { "epoch": 1.72, "learning_rate": 1.283377777777778e-05, "loss": 0.0003, "step": 241500 }, { "epoch": 1.72, "learning_rate": 1.2798222222222223e-05, "loss": 0.0003, "step": 242000 }, { "epoch": 1.72, "learning_rate": 1.2762666666666666e-05, "loss": 0.0003, "step": 242500 }, { "epoch": 1.73, "learning_rate": 1.2727111111111111e-05, "loss": 0.0003, "step": 243000 }, { "epoch": 1.73, "learning_rate": 1.2691555555555555e-05, "loss": 0.0003, "step": 243500 }, { "epoch": 1.74, "learning_rate": 1.2656e-05, "loss": 0.0003, "step": 244000 }, { "epoch": 1.74, "learning_rate": 1.2620444444444444e-05, "loss": 0.0003, "step": 244500 }, { "epoch": 1.74, "learning_rate": 1.2584960000000001e-05, "loss": 0.0003, "step": 245000 }, { "epoch": 1.75, "learning_rate": 1.2549404444444445e-05, "loss": 0.0003, "step": 245500 }, { "epoch": 1.75, "learning_rate": 1.2513848888888889e-05, "loss": 0.0003, "step": 246000 }, { "epoch": 1.75, "learning_rate": 1.2478293333333333e-05, "loss": 0.0003, "step": 246500 }, { "epoch": 1.76, "learning_rate": 1.244280888888889e-05, "loss": 0.0003, "step": 247000 }, { "epoch": 1.76, "learning_rate": 1.2407253333333334e-05, "loss": 0.0003, "step": 247500 }, { "epoch": 1.76, "learning_rate": 1.2371697777777779e-05, "loss": 0.0003, "step": 248000 }, { "epoch": 1.77, "learning_rate": 1.2336142222222223e-05, "loss": 0.0003, "step": 248500 }, { "epoch": 1.77, "learning_rate": 1.2300657777777778e-05, "loss": 0.0003, "step": 249000 }, { "epoch": 1.77, "learning_rate": 1.2265102222222222e-05, "loss": 0.0003, "step": 249500 }, { "epoch": 1.78, "learning_rate": 1.2229546666666666e-05, "loss": 0.0003, "step": 250000 }, { "epoch": 1.78, "learning_rate": 1.2193991111111111e-05, "loss": 0.0003, "step": 250500 }, { "epoch": 1.78, "learning_rate": 1.2158506666666667e-05, "loss": 0.0003, "step": 251000 }, { "epoch": 1.79, "learning_rate": 1.2122951111111112e-05, "loss": 0.0003, "step": 251500 }, { "epoch": 1.79, "learning_rate": 1.2087395555555556e-05, "loss": 0.0003, "step": 252000 }, { "epoch": 1.8, "learning_rate": 1.205184e-05, "loss": 0.0003, "step": 252500 }, { "epoch": 1.8, "learning_rate": 1.2016355555555555e-05, "loss": 0.0003, "step": 253000 }, { "epoch": 1.8, "learning_rate": 1.19808e-05, "loss": 0.0003, "step": 253500 }, { "epoch": 1.81, "learning_rate": 1.1945244444444444e-05, "loss": 0.0003, "step": 254000 }, { "epoch": 1.81, "learning_rate": 1.190968888888889e-05, "loss": 0.0003, "step": 254500 }, { "epoch": 1.81, "learning_rate": 1.1874275555555557e-05, "loss": 0.0003, "step": 255000 }, { "epoch": 1.82, "learning_rate": 1.183872e-05, "loss": 0.0003, "step": 255500 }, { "epoch": 1.82, "learning_rate": 1.1803164444444444e-05, "loss": 0.0003, "step": 256000 }, { "epoch": 1.82, "learning_rate": 1.1767608888888888e-05, "loss": 0.0003, "step": 256500 }, { "epoch": 1.83, "learning_rate": 1.1732053333333334e-05, "loss": 0.0003, "step": 257000 }, { "epoch": 1.83, "learning_rate": 1.1696497777777777e-05, "loss": 0.0003, "step": 257500 }, { "epoch": 1.83, "learning_rate": 1.1661013333333334e-05, "loss": 0.0003, "step": 258000 }, { "epoch": 1.84, "learning_rate": 1.1625457777777778e-05, "loss": 0.0003, "step": 258500 }, { "epoch": 1.84, "learning_rate": 1.1589902222222222e-05, "loss": 0.0003, "step": 259000 }, { "epoch": 1.85, "learning_rate": 1.1554346666666666e-05, "loss": 0.0003, "step": 259500 }, { "epoch": 1.85, "learning_rate": 1.1518791111111111e-05, "loss": 0.0003, "step": 260000 }, { "epoch": 1.85, "learning_rate": 1.1483235555555555e-05, "loss": 0.0003, "step": 260500 }, { "epoch": 1.86, "learning_rate": 1.144768e-05, "loss": 0.0003, "step": 261000 }, { "epoch": 1.86, "learning_rate": 1.1412124444444445e-05, "loss": 0.0003, "step": 261500 }, { "epoch": 1.86, "learning_rate": 1.1376640000000002e-05, "loss": 0.0003, "step": 262000 }, { "epoch": 1.87, "learning_rate": 1.1341084444444444e-05, "loss": 0.0003, "step": 262500 }, { "epoch": 1.87, "learning_rate": 1.130552888888889e-05, "loss": 0.0003, "step": 263000 }, { "epoch": 1.87, "learning_rate": 1.1269973333333333e-05, "loss": 0.0003, "step": 263500 }, { "epoch": 1.88, "learning_rate": 1.1234488888888888e-05, "loss": 0.0003, "step": 264000 }, { "epoch": 1.88, "learning_rate": 1.1198933333333334e-05, "loss": 0.0003, "step": 264500 }, { "epoch": 1.88, "learning_rate": 1.1163377777777778e-05, "loss": 0.0003, "step": 265000 }, { "epoch": 1.89, "learning_rate": 1.1127822222222223e-05, "loss": 0.0003, "step": 265500 }, { "epoch": 1.89, "learning_rate": 1.1092337777777777e-05, "loss": 0.0003, "step": 266000 }, { "epoch": 1.9, "learning_rate": 1.1056782222222222e-05, "loss": 0.0003, "step": 266500 }, { "epoch": 1.9, "learning_rate": 1.1021226666666666e-05, "loss": 0.0003, "step": 267000 }, { "epoch": 1.9, "learning_rate": 1.0985671111111112e-05, "loss": 0.0003, "step": 267500 }, { "epoch": 1.91, "learning_rate": 1.0950186666666667e-05, "loss": 0.0003, "step": 268000 }, { "epoch": 1.91, "learning_rate": 1.0914631111111112e-05, "loss": 0.0003, "step": 268500 }, { "epoch": 1.91, "learning_rate": 1.0879075555555556e-05, "loss": 0.0003, "step": 269000 }, { "epoch": 1.92, "learning_rate": 1.084352e-05, "loss": 0.0003, "step": 269500 }, { "epoch": 1.92, "learning_rate": 1.0808035555555555e-05, "loss": 0.0003, "step": 270000 }, { "epoch": 1.92, "learning_rate": 1.0772480000000001e-05, "loss": 0.0003, "step": 270500 }, { "epoch": 1.93, "learning_rate": 1.0736924444444445e-05, "loss": 0.0003, "step": 271000 }, { "epoch": 1.93, "learning_rate": 1.0701368888888889e-05, "loss": 0.0003, "step": 271500 }, { "epoch": 1.93, "learning_rate": 1.0665884444444446e-05, "loss": 0.0003, "step": 272000 }, { "epoch": 1.94, "learning_rate": 1.063032888888889e-05, "loss": 0.0003, "step": 272500 }, { "epoch": 1.94, "learning_rate": 1.0594844444444445e-05, "loss": 0.0003, "step": 273000 }, { "epoch": 1.94, "learning_rate": 1.0559288888888888e-05, "loss": 0.0003, "step": 273500 }, { "epoch": 1.95, "learning_rate": 1.0523733333333334e-05, "loss": 0.0003, "step": 274000 }, { "epoch": 1.95, "learning_rate": 1.0488177777777778e-05, "loss": 0.0003, "step": 274500 }, { "epoch": 1.96, "learning_rate": 1.0452622222222223e-05, "loss": 0.0003, "step": 275000 }, { "epoch": 1.96, "learning_rate": 1.0417066666666667e-05, "loss": 0.0003, "step": 275500 }, { "epoch": 1.96, "learning_rate": 1.0381511111111113e-05, "loss": 0.0003, "step": 276000 }, { "epoch": 1.97, "learning_rate": 1.0345955555555555e-05, "loss": 0.0003, "step": 276500 }, { "epoch": 1.97, "learning_rate": 1.0310471111111112e-05, "loss": 0.0003, "step": 277000 }, { "epoch": 1.97, "learning_rate": 1.0274915555555556e-05, "loss": 0.0003, "step": 277500 }, { "epoch": 1.98, "learning_rate": 1.0239360000000001e-05, "loss": 0.0003, "step": 278000 }, { "epoch": 1.98, "learning_rate": 1.0203804444444445e-05, "loss": 0.0003, "step": 278500 }, { "epoch": 1.98, "learning_rate": 1.016832e-05, "loss": 0.0003, "step": 279000 }, { "epoch": 1.99, "learning_rate": 1.0132764444444444e-05, "loss": 0.0003, "step": 279500 }, { "epoch": 1.99, "learning_rate": 1.0097208888888888e-05, "loss": 0.0003, "step": 280000 }, { "epoch": 1.99, "learning_rate": 1.0061653333333333e-05, "loss": 0.0003, "step": 280500 }, { "epoch": 2.0, "learning_rate": 1.0026168888888889e-05, "loss": 0.0003, "step": 281000 }, { "epoch": 2.0, "learning_rate": 9.990613333333334e-06, "loss": 0.0003, "step": 281500 }, { "epoch": 2.01, "learning_rate": 9.955057777777778e-06, "loss": 0.0003, "step": 282000 }, { "epoch": 2.01, "learning_rate": 9.919502222222224e-06, "loss": 0.0003, "step": 282500 }, { "epoch": 2.01, "learning_rate": 9.884017777777777e-06, "loss": 0.0003, "step": 283000 }, { "epoch": 2.02, "learning_rate": 9.848462222222223e-06, "loss": 0.0003, "step": 283500 }, { "epoch": 2.02, "learning_rate": 9.812906666666667e-06, "loss": 0.0003, "step": 284000 }, { "epoch": 2.02, "learning_rate": 9.777351111111112e-06, "loss": 0.0003, "step": 284500 }, { "epoch": 2.03, "learning_rate": 9.741866666666667e-06, "loss": 0.0003, "step": 285000 }, { "epoch": 2.03, "learning_rate": 9.706311111111111e-06, "loss": 0.0003, "step": 285500 }, { "epoch": 2.03, "learning_rate": 9.670755555555557e-06, "loss": 0.0003, "step": 286000 }, { "epoch": 2.04, "learning_rate": 9.635199999999999e-06, "loss": 0.0003, "step": 286500 }, { "epoch": 2.04, "learning_rate": 9.599715555555556e-06, "loss": 0.0003, "step": 287000 }, { "epoch": 2.04, "learning_rate": 9.56416e-06, "loss": 0.0003, "step": 287500 }, { "epoch": 2.05, "learning_rate": 9.528604444444445e-06, "loss": 0.0003, "step": 288000 }, { "epoch": 2.05, "learning_rate": 9.493048888888889e-06, "loss": 0.0003, "step": 288500 }, { "epoch": 2.06, "learning_rate": 9.457564444444446e-06, "loss": 0.0003, "step": 289000 }, { "epoch": 2.06, "learning_rate": 9.422008888888888e-06, "loss": 0.0003, "step": 289500 }, { "epoch": 2.06, "learning_rate": 9.386453333333334e-06, "loss": 0.0003, "step": 290000 }, { "epoch": 2.07, "learning_rate": 9.350897777777777e-06, "loss": 0.0003, "step": 290500 }, { "epoch": 2.07, "learning_rate": 9.315413333333334e-06, "loss": 0.0003, "step": 291000 }, { "epoch": 2.07, "learning_rate": 9.279857777777778e-06, "loss": 0.0003, "step": 291500 }, { "epoch": 2.08, "learning_rate": 9.244302222222224e-06, "loss": 0.0003, "step": 292000 }, { "epoch": 2.08, "learning_rate": 9.208746666666668e-06, "loss": 0.0003, "step": 292500 }, { "epoch": 2.08, "learning_rate": 9.173262222222221e-06, "loss": 0.0003, "step": 293000 }, { "epoch": 2.09, "learning_rate": 9.137706666666667e-06, "loss": 0.0003, "step": 293500 }, { "epoch": 2.09, "learning_rate": 9.10215111111111e-06, "loss": 0.0003, "step": 294000 }, { "epoch": 2.09, "learning_rate": 9.066595555555556e-06, "loss": 0.0003, "step": 294500 }, { "epoch": 2.1, "learning_rate": 9.031111111111111e-06, "loss": 0.0003, "step": 295000 }, { "epoch": 2.1, "learning_rate": 8.995555555555557e-06, "loss": 0.0003, "step": 295500 }, { "epoch": 2.1, "learning_rate": 8.96e-06, "loss": 0.0003, "step": 296000 }, { "epoch": 2.11, "learning_rate": 8.924444444444444e-06, "loss": 0.0003, "step": 296500 }, { "epoch": 2.11, "learning_rate": 8.88896e-06, "loss": 0.0003, "step": 297000 }, { "epoch": 2.12, "learning_rate": 8.853404444444445e-06, "loss": 0.0003, "step": 297500 }, { "epoch": 2.12, "learning_rate": 8.817848888888889e-06, "loss": 0.0003, "step": 298000 }, { "epoch": 2.12, "learning_rate": 8.782293333333335e-06, "loss": 0.0003, "step": 298500 }, { "epoch": 2.13, "learning_rate": 8.74680888888889e-06, "loss": 0.0003, "step": 299000 }, { "epoch": 2.13, "learning_rate": 8.711253333333332e-06, "loss": 0.0003, "step": 299500 }, { "epoch": 2.13, "learning_rate": 8.675768888888889e-06, "loss": 0.0003, "step": 300000 }, { "epoch": 2.14, "learning_rate": 8.640213333333333e-06, "loss": 0.0003, "step": 300500 }, { "epoch": 2.14, "learning_rate": 8.604657777777778e-06, "loss": 0.0003, "step": 301000 }, { "epoch": 2.14, "learning_rate": 8.569102222222222e-06, "loss": 0.0003, "step": 301500 }, { "epoch": 2.15, "learning_rate": 8.533546666666668e-06, "loss": 0.0003, "step": 302000 }, { "epoch": 2.15, "learning_rate": 8.497991111111111e-06, "loss": 0.0003, "step": 302500 }, { "epoch": 2.15, "learning_rate": 8.462435555555557e-06, "loss": 0.0003, "step": 303000 }, { "epoch": 2.16, "learning_rate": 8.42688e-06, "loss": 0.0003, "step": 303500 }, { "epoch": 2.16, "learning_rate": 8.391395555555556e-06, "loss": 0.0003, "step": 304000 }, { "epoch": 2.17, "learning_rate": 8.35584e-06, "loss": 0.0003, "step": 304500 }, { "epoch": 2.17, "learning_rate": 8.320284444444445e-06, "loss": 0.0003, "step": 305000 }, { "epoch": 2.17, "learning_rate": 8.28472888888889e-06, "loss": 0.0003, "step": 305500 }, { "epoch": 2.18, "learning_rate": 8.249244444444445e-06, "loss": 0.0003, "step": 306000 }, { "epoch": 2.18, "learning_rate": 8.213688888888888e-06, "loss": 0.0003, "step": 306500 }, { "epoch": 2.18, "learning_rate": 8.178204444444444e-06, "loss": 0.0003, "step": 307000 }, { "epoch": 2.19, "learning_rate": 8.14264888888889e-06, "loss": 0.0003, "step": 307500 }, { "epoch": 2.19, "learning_rate": 8.107093333333333e-06, "loss": 0.0003, "step": 308000 }, { "epoch": 2.19, "learning_rate": 8.071537777777779e-06, "loss": 0.0003, "step": 308500 }, { "epoch": 2.2, "learning_rate": 8.035982222222222e-06, "loss": 0.0003, "step": 309000 }, { "epoch": 2.2, "learning_rate": 8.000426666666668e-06, "loss": 0.0003, "step": 309500 }, { "epoch": 2.2, "learning_rate": 7.96487111111111e-06, "loss": 0.0003, "step": 310000 }, { "epoch": 2.21, "learning_rate": 7.929315555555556e-06, "loss": 0.0003, "step": 310500 }, { "epoch": 2.21, "learning_rate": 7.89383111111111e-06, "loss": 0.0003, "step": 311000 }, { "epoch": 2.22, "learning_rate": 7.858275555555556e-06, "loss": 0.0003, "step": 311500 }, { "epoch": 2.22, "learning_rate": 7.82272e-06, "loss": 0.0003, "step": 312000 }, { "epoch": 2.22, "learning_rate": 7.787164444444446e-06, "loss": 0.0003, "step": 312500 }, { "epoch": 2.23, "learning_rate": 7.751680000000001e-06, "loss": 0.0003, "step": 313000 }, { "epoch": 2.23, "learning_rate": 7.716124444444445e-06, "loss": 0.0003, "step": 313500 }, { "epoch": 2.23, "learning_rate": 7.680568888888889e-06, "loss": 0.0003, "step": 314000 }, { "epoch": 2.24, "learning_rate": 7.645013333333333e-06, "loss": 0.0003, "step": 314500 }, { "epoch": 2.24, "learning_rate": 7.609457777777778e-06, "loss": 0.0003, "step": 315000 }, { "epoch": 2.24, "learning_rate": 7.573973333333334e-06, "loss": 0.0003, "step": 315500 }, { "epoch": 2.25, "learning_rate": 7.538417777777779e-06, "loss": 0.0003, "step": 316000 }, { "epoch": 2.25, "learning_rate": 7.5028622222222235e-06, "loss": 0.0003, "step": 316500 }, { "epoch": 2.25, "learning_rate": 7.467306666666667e-06, "loss": 0.0002, "step": 317000 }, { "epoch": 2.26, "learning_rate": 7.431822222222222e-06, "loss": 0.0003, "step": 317500 }, { "epoch": 2.26, "learning_rate": 7.396266666666666e-06, "loss": 0.0003, "step": 318000 }, { "epoch": 2.26, "learning_rate": 7.360711111111111e-06, "loss": 0.0003, "step": 318500 }, { "epoch": 2.27, "learning_rate": 7.325155555555556e-06, "loss": 0.0003, "step": 319000 }, { "epoch": 2.27, "learning_rate": 7.289671111111111e-06, "loss": 0.0003, "step": 319500 }, { "epoch": 2.28, "learning_rate": 7.254115555555556e-06, "loss": 0.0002, "step": 320000 }, { "epoch": 2.28, "learning_rate": 7.21856e-06, "loss": 0.0003, "step": 320500 }, { "epoch": 2.28, "learning_rate": 7.183004444444444e-06, "loss": 0.0003, "step": 321000 }, { "epoch": 2.29, "learning_rate": 7.14752e-06, "loss": 0.0003, "step": 321500 }, { "epoch": 2.29, "learning_rate": 7.111964444444445e-06, "loss": 0.0003, "step": 322000 }, { "epoch": 2.29, "learning_rate": 7.07640888888889e-06, "loss": 0.0003, "step": 322500 }, { "epoch": 2.3, "learning_rate": 7.0408533333333335e-06, "loss": 0.0003, "step": 323000 }, { "epoch": 2.3, "learning_rate": 7.00536888888889e-06, "loss": 0.0003, "step": 323500 }, { "epoch": 2.3, "learning_rate": 6.969813333333334e-06, "loss": 0.0003, "step": 324000 }, { "epoch": 2.31, "learning_rate": 6.934257777777777e-06, "loss": 0.0002, "step": 324500 }, { "epoch": 2.31, "learning_rate": 6.898702222222222e-06, "loss": 0.0003, "step": 325000 }, { "epoch": 2.31, "learning_rate": 6.863217777777778e-06, "loss": 0.0003, "step": 325500 }, { "epoch": 2.32, "learning_rate": 6.827662222222222e-06, "loss": 0.0003, "step": 326000 }, { "epoch": 2.32, "learning_rate": 6.792106666666667e-06, "loss": 0.0003, "step": 326500 }, { "epoch": 2.33, "learning_rate": 6.756551111111111e-06, "loss": 0.0003, "step": 327000 }, { "epoch": 2.33, "learning_rate": 6.7210666666666665e-06, "loss": 0.0003, "step": 327500 }, { "epoch": 2.33, "learning_rate": 6.685511111111111e-06, "loss": 0.0003, "step": 328000 }, { "epoch": 2.34, "learning_rate": 6.649955555555556e-06, "loss": 0.0003, "step": 328500 }, { "epoch": 2.34, "learning_rate": 6.6144000000000006e-06, "loss": 0.0003, "step": 329000 }, { "epoch": 2.34, "learning_rate": 6.578915555555556e-06, "loss": 0.0003, "step": 329500 }, { "epoch": 2.35, "learning_rate": 6.5433600000000005e-06, "loss": 0.0003, "step": 330000 }, { "epoch": 2.35, "learning_rate": 6.507804444444445e-06, "loss": 0.0003, "step": 330500 }, { "epoch": 2.35, "learning_rate": 6.472248888888889e-06, "loss": 0.0003, "step": 331000 }, { "epoch": 2.36, "learning_rate": 6.436764444444444e-06, "loss": 0.0003, "step": 331500 }, { "epoch": 2.36, "learning_rate": 6.401208888888889e-06, "loss": 0.0003, "step": 332000 }, { "epoch": 2.36, "learning_rate": 6.365653333333334e-06, "loss": 0.0003, "step": 332500 }, { "epoch": 2.37, "learning_rate": 6.3300977777777775e-06, "loss": 0.0003, "step": 333000 }, { "epoch": 2.37, "learning_rate": 6.2946133333333335e-06, "loss": 0.0003, "step": 333500 }, { "epoch": 2.38, "learning_rate": 6.259057777777778e-06, "loss": 0.0003, "step": 334000 }, { "epoch": 2.38, "learning_rate": 6.223502222222222e-06, "loss": 0.0003, "step": 334500 }, { "epoch": 2.38, "learning_rate": 6.187946666666667e-06, "loss": 0.0003, "step": 335000 }, { "epoch": 2.39, "learning_rate": 6.1523911111111115e-06, "loss": 0.0003, "step": 335500 }, { "epoch": 2.39, "learning_rate": 6.116906666666667e-06, "loss": 0.0002, "step": 336000 }, { "epoch": 2.39, "learning_rate": 6.081351111111111e-06, "loss": 0.0003, "step": 336500 }, { "epoch": 2.4, "learning_rate": 6.045795555555556e-06, "loss": 0.0002, "step": 337000 }, { "epoch": 2.4, "learning_rate": 6.010240000000001e-06, "loss": 0.0003, "step": 337500 }, { "epoch": 2.4, "learning_rate": 5.974755555555555e-06, "loss": 0.0002, "step": 338000 }, { "epoch": 2.41, "learning_rate": 5.9392e-06, "loss": 0.0002, "step": 338500 }, { "epoch": 2.41, "learning_rate": 5.9036444444444445e-06, "loss": 0.0002, "step": 339000 }, { "epoch": 2.41, "learning_rate": 5.868088888888888e-06, "loss": 0.0002, "step": 339500 }, { "epoch": 2.42, "learning_rate": 5.8326044444444444e-06, "loss": 0.0003, "step": 340000 }, { "epoch": 2.42, "learning_rate": 5.797048888888889e-06, "loss": 0.0003, "step": 340500 }, { "epoch": 2.42, "learning_rate": 5.761493333333334e-06, "loss": 0.0003, "step": 341000 }, { "epoch": 2.43, "learning_rate": 5.725937777777778e-06, "loss": 0.0003, "step": 341500 }, { "epoch": 2.43, "learning_rate": 5.690453333333334e-06, "loss": 0.0002, "step": 342000 }, { "epoch": 2.44, "learning_rate": 5.654897777777778e-06, "loss": 0.0003, "step": 342500 }, { "epoch": 2.44, "learning_rate": 5.619342222222222e-06, "loss": 0.0002, "step": 343000 }, { "epoch": 2.44, "learning_rate": 5.583786666666667e-06, "loss": 0.0002, "step": 343500 }, { "epoch": 2.45, "learning_rate": 5.548302222222223e-06, "loss": 0.0003, "step": 344000 }, { "epoch": 2.45, "learning_rate": 5.512746666666667e-06, "loss": 0.0002, "step": 344500 }, { "epoch": 2.45, "learning_rate": 5.4771911111111116e-06, "loss": 0.0003, "step": 345000 }, { "epoch": 2.46, "learning_rate": 5.441635555555555e-06, "loss": 0.0003, "step": 345500 }, { "epoch": 2.46, "learning_rate": 5.406151111111111e-06, "loss": 0.0002, "step": 346000 }, { "epoch": 2.46, "learning_rate": 5.370595555555555e-06, "loss": 0.0002, "step": 346500 }, { "epoch": 2.47, "learning_rate": 5.33504e-06, "loss": 0.0002, "step": 347000 }, { "epoch": 2.47, "learning_rate": 5.299484444444445e-06, "loss": 0.0002, "step": 347500 }, { "epoch": 2.47, "learning_rate": 5.264e-06, "loss": 0.0002, "step": 348000 }, { "epoch": 2.48, "learning_rate": 5.228444444444445e-06, "loss": 0.0002, "step": 348500 }, { "epoch": 2.48, "learning_rate": 5.192888888888889e-06, "loss": 0.0002, "step": 349000 }, { "epoch": 2.49, "learning_rate": 5.157333333333333e-06, "loss": 0.0002, "step": 349500 }, { "epoch": 2.49, "learning_rate": 5.121777777777778e-06, "loss": 0.0002, "step": 350000 }, { "epoch": 2.49, "learning_rate": 5.086293333333334e-06, "loss": 0.0002, "step": 350500 }, { "epoch": 2.5, "learning_rate": 5.050737777777779e-06, "loss": 0.0002, "step": 351000 }, { "epoch": 2.5, "learning_rate": 5.0151822222222224e-06, "loss": 0.0002, "step": 351500 }, { "epoch": 2.5, "learning_rate": 4.979626666666667e-06, "loss": 0.0002, "step": 352000 }, { "epoch": 2.51, "learning_rate": 4.944142222222222e-06, "loss": 0.0002, "step": 352500 }, { "epoch": 2.51, "learning_rate": 4.908586666666666e-06, "loss": 0.0002, "step": 353000 }, { "epoch": 2.51, "learning_rate": 4.873031111111111e-06, "loss": 0.0002, "step": 353500 }, { "epoch": 2.52, "learning_rate": 4.8374755555555556e-06, "loss": 0.0003, "step": 354000 }, { "epoch": 2.52, "learning_rate": 4.801991111111111e-06, "loss": 0.0002, "step": 354500 }, { "epoch": 2.52, "learning_rate": 4.7664355555555555e-06, "loss": 0.0002, "step": 355000 }, { "epoch": 2.53, "learning_rate": 4.73088e-06, "loss": 0.0002, "step": 355500 }, { "epoch": 2.53, "learning_rate": 4.695324444444445e-06, "loss": 0.0002, "step": 356000 }, { "epoch": 2.54, "learning_rate": 4.65984e-06, "loss": 0.0002, "step": 356500 }, { "epoch": 2.54, "learning_rate": 4.624284444444445e-06, "loss": 0.0002, "step": 357000 }, { "epoch": 2.54, "learning_rate": 4.5887288888888895e-06, "loss": 0.0002, "step": 357500 }, { "epoch": 2.55, "learning_rate": 4.553173333333333e-06, "loss": 0.0002, "step": 358000 }, { "epoch": 2.55, "learning_rate": 4.517688888888889e-06, "loss": 0.0002, "step": 358500 }, { "epoch": 2.55, "learning_rate": 4.482133333333333e-06, "loss": 0.0002, "step": 359000 }, { "epoch": 2.56, "learning_rate": 4.446577777777778e-06, "loss": 0.0002, "step": 359500 }, { "epoch": 2.56, "learning_rate": 4.411022222222222e-06, "loss": 0.0002, "step": 360000 }, { "epoch": 2.56, "learning_rate": 4.375537777777778e-06, "loss": 0.0002, "step": 360500 }, { "epoch": 2.57, "learning_rate": 4.3399822222222225e-06, "loss": 0.0002, "step": 361000 }, { "epoch": 2.57, "learning_rate": 4.304426666666666e-06, "loss": 0.0003, "step": 361500 }, { "epoch": 2.57, "learning_rate": 4.268871111111111e-06, "loss": 0.0002, "step": 362000 }, { "epoch": 2.58, "learning_rate": 4.233386666666667e-06, "loss": 0.0002, "step": 362500 }, { "epoch": 2.58, "learning_rate": 4.197831111111111e-06, "loss": 0.0002, "step": 363000 }, { "epoch": 2.58, "learning_rate": 4.162275555555556e-06, "loss": 0.0002, "step": 363500 }, { "epoch": 2.59, "learning_rate": 4.12672e-06, "loss": 0.0002, "step": 364000 }, { "epoch": 2.59, "learning_rate": 4.091235555555556e-06, "loss": 0.0002, "step": 364500 }, { "epoch": 2.6, "learning_rate": 4.05568e-06, "loss": 0.0002, "step": 365000 }, { "epoch": 2.6, "learning_rate": 4.020124444444445e-06, "loss": 0.0002, "step": 365500 }, { "epoch": 2.6, "learning_rate": 3.98456888888889e-06, "loss": 0.0002, "step": 366000 }, { "epoch": 2.61, "learning_rate": 3.949084444444444e-06, "loss": 0.0002, "step": 366500 }, { "epoch": 2.61, "learning_rate": 3.913528888888889e-06, "loss": 0.0002, "step": 367000 }, { "epoch": 2.61, "learning_rate": 3.877973333333333e-06, "loss": 0.0002, "step": 367500 }, { "epoch": 2.62, "learning_rate": 3.842417777777777e-06, "loss": 0.0002, "step": 368000 }, { "epoch": 2.62, "learning_rate": 3.806862222222222e-06, "loss": 0.0002, "step": 368500 }, { "epoch": 2.62, "learning_rate": 3.771377777777778e-06, "loss": 0.0002, "step": 369000 }, { "epoch": 2.63, "learning_rate": 3.7358222222222223e-06, "loss": 0.0002, "step": 369500 }, { "epoch": 2.63, "learning_rate": 3.700266666666667e-06, "loss": 0.0002, "step": 370000 }, { "epoch": 2.63, "learning_rate": 3.6647111111111112e-06, "loss": 0.0002, "step": 370500 }, { "epoch": 2.64, "learning_rate": 3.6292266666666665e-06, "loss": 0.0002, "step": 371000 }, { "epoch": 2.64, "learning_rate": 3.593671111111111e-06, "loss": 0.0002, "step": 371500 }, { "epoch": 2.65, "learning_rate": 3.5581155555555554e-06, "loss": 0.0002, "step": 372000 }, { "epoch": 2.65, "learning_rate": 3.52256e-06, "loss": 0.0002, "step": 372500 }, { "epoch": 2.65, "learning_rate": 3.4870755555555558e-06, "loss": 0.0002, "step": 373000 }, { "epoch": 2.66, "learning_rate": 3.45152e-06, "loss": 0.0002, "step": 373500 }, { "epoch": 2.66, "learning_rate": 3.4159644444444447e-06, "loss": 0.0002, "step": 374000 }, { "epoch": 2.66, "learning_rate": 3.380408888888889e-06, "loss": 0.0002, "step": 374500 }, { "epoch": 2.67, "learning_rate": 3.3449244444444442e-06, "loss": 0.0002, "step": 375000 }, { "epoch": 2.67, "learning_rate": 3.309368888888889e-06, "loss": 0.0002, "step": 375500 }, { "epoch": 2.67, "learning_rate": 3.273813333333333e-06, "loss": 0.0002, "step": 376000 }, { "epoch": 2.68, "learning_rate": 3.238257777777778e-06, "loss": 0.0002, "step": 376500 }, { "epoch": 2.68, "learning_rate": 3.2027733333333335e-06, "loss": 0.0002, "step": 377000 }, { "epoch": 2.68, "learning_rate": 3.167217777777778e-06, "loss": 0.0002, "step": 377500 }, { "epoch": 2.69, "learning_rate": 3.1316622222222225e-06, "loss": 0.0002, "step": 378000 }, { "epoch": 2.69, "learning_rate": 3.0961066666666667e-06, "loss": 0.0002, "step": 378500 }, { "epoch": 2.7, "learning_rate": 3.0606222222222224e-06, "loss": 0.0002, "step": 379000 }, { "epoch": 2.7, "learning_rate": 3.0250666666666666e-06, "loss": 0.0002, "step": 379500 }, { "epoch": 2.7, "learning_rate": 2.9895111111111113e-06, "loss": 0.0002, "step": 380000 }, { "epoch": 2.71, "learning_rate": 2.9539555555555556e-06, "loss": 0.0002, "step": 380500 }, { "epoch": 2.71, "learning_rate": 2.9184000000000003e-06, "loss": 0.0002, "step": 381000 }, { "epoch": 2.71, "learning_rate": 2.8829155555555555e-06, "loss": 0.0002, "step": 381500 }, { "epoch": 2.72, "learning_rate": 2.84736e-06, "loss": 0.0002, "step": 382000 }, { "epoch": 2.72, "learning_rate": 2.8118044444444445e-06, "loss": 0.0002, "step": 382500 }, { "epoch": 2.72, "learning_rate": 2.7762488888888887e-06, "loss": 0.0002, "step": 383000 }, { "epoch": 2.73, "learning_rate": 2.7407644444444444e-06, "loss": 0.0002, "step": 383500 }, { "epoch": 2.73, "learning_rate": 2.705208888888889e-06, "loss": 0.0002, "step": 384000 }, { "epoch": 2.73, "learning_rate": 2.6696533333333334e-06, "loss": 0.0002, "step": 384500 }, { "epoch": 2.74, "learning_rate": 2.634097777777778e-06, "loss": 0.0002, "step": 385000 }, { "epoch": 2.74, "learning_rate": 2.5986133333333333e-06, "loss": 0.0002, "step": 385500 }, { "epoch": 2.74, "learning_rate": 2.5630577777777775e-06, "loss": 0.0002, "step": 386000 }, { "epoch": 2.75, "learning_rate": 2.5275022222222222e-06, "loss": 0.0002, "step": 386500 }, { "epoch": 2.75, "learning_rate": 2.4919466666666665e-06, "loss": 0.0002, "step": 387000 }, { "epoch": 2.76, "learning_rate": 2.4564622222222226e-06, "loss": 0.0002, "step": 387500 }, { "epoch": 2.76, "learning_rate": 2.420906666666667e-06, "loss": 0.0002, "step": 388000 }, { "epoch": 2.76, "learning_rate": 2.3853511111111115e-06, "loss": 0.0002, "step": 388500 }, { "epoch": 2.77, "learning_rate": 2.3497955555555558e-06, "loss": 0.0002, "step": 389000 }, { "epoch": 2.77, "learning_rate": 2.314311111111111e-06, "loss": 0.0002, "step": 389500 }, { "epoch": 2.77, "learning_rate": 2.2787555555555557e-06, "loss": 0.0002, "step": 390000 }, { "epoch": 2.78, "learning_rate": 2.2432e-06, "loss": 0.0002, "step": 390500 }, { "epoch": 2.78, "learning_rate": 2.2076444444444447e-06, "loss": 0.0002, "step": 391000 }, { "epoch": 2.78, "learning_rate": 2.172088888888889e-06, "loss": 0.0002, "step": 391500 }, { "epoch": 2.79, "learning_rate": 2.136604444444444e-06, "loss": 0.0002, "step": 392000 }, { "epoch": 2.79, "learning_rate": 2.101048888888889e-06, "loss": 0.0002, "step": 392500 }, { "epoch": 2.79, "learning_rate": 2.065493333333333e-06, "loss": 0.0002, "step": 393000 }, { "epoch": 2.8, "learning_rate": 2.029937777777778e-06, "loss": 0.0002, "step": 393500 }, { "epoch": 2.8, "learning_rate": 1.9944533333333334e-06, "loss": 0.0002, "step": 394000 }, { "epoch": 2.81, "learning_rate": 1.9588977777777777e-06, "loss": 0.0002, "step": 394500 }, { "epoch": 2.81, "learning_rate": 1.9233422222222224e-06, "loss": 0.0002, "step": 395000 }, { "epoch": 2.81, "learning_rate": 1.8877866666666665e-06, "loss": 0.0002, "step": 395500 }, { "epoch": 2.82, "learning_rate": 1.8523022222222223e-06, "loss": 0.0002, "step": 396000 }, { "epoch": 2.82, "learning_rate": 1.8167466666666668e-06, "loss": 0.0002, "step": 396500 }, { "epoch": 2.82, "learning_rate": 1.781191111111111e-06, "loss": 0.0002, "step": 397000 }, { "epoch": 2.83, "learning_rate": 1.7456355555555555e-06, "loss": 0.0002, "step": 397500 }, { "epoch": 2.83, "learning_rate": 1.710151111111111e-06, "loss": 0.0002, "step": 398000 }, { "epoch": 2.83, "learning_rate": 1.6745955555555555e-06, "loss": 0.0002, "step": 398500 }, { "epoch": 2.84, "learning_rate": 1.63904e-06, "loss": 0.0002, "step": 399000 }, { "epoch": 2.84, "learning_rate": 1.6034844444444444e-06, "loss": 0.0002, "step": 399500 }, { "epoch": 2.84, "learning_rate": 1.568e-06, "loss": 0.0002, "step": 400000 }, { "epoch": 2.85, "learning_rate": 1.5324444444444445e-06, "loss": 0.0002, "step": 400500 }, { "epoch": 2.85, "learning_rate": 1.496888888888889e-06, "loss": 0.0002, "step": 401000 }, { "epoch": 2.86, "learning_rate": 1.4613333333333333e-06, "loss": 0.0002, "step": 401500 }, { "epoch": 2.86, "learning_rate": 1.4257777777777778e-06, "loss": 0.0002, "step": 402000 }, { "epoch": 2.86, "learning_rate": 1.3902933333333334e-06, "loss": 0.0002, "step": 402500 }, { "epoch": 2.87, "learning_rate": 1.354737777777778e-06, "loss": 0.0002, "step": 403000 }, { "epoch": 2.87, "learning_rate": 1.3191822222222224e-06, "loss": 0.0002, "step": 403500 }, { "epoch": 2.87, "learning_rate": 1.2836266666666669e-06, "loss": 0.0002, "step": 404000 }, { "epoch": 2.88, "learning_rate": 1.2481422222222223e-06, "loss": 0.0002, "step": 404500 }, { "epoch": 2.88, "learning_rate": 1.2125866666666668e-06, "loss": 0.0002, "step": 405000 }, { "epoch": 2.88, "learning_rate": 1.1770311111111112e-06, "loss": 0.0002, "step": 405500 }, { "epoch": 2.89, "learning_rate": 1.1414755555555555e-06, "loss": 0.0002, "step": 406000 }, { "epoch": 2.89, "learning_rate": 1.1059911111111112e-06, "loss": 0.0002, "step": 406500 }, { "epoch": 2.89, "learning_rate": 1.0704355555555556e-06, "loss": 0.0002, "step": 407000 }, { "epoch": 2.9, "learning_rate": 1.03488e-06, "loss": 0.0002, "step": 407500 }, { "epoch": 2.9, "learning_rate": 9.993244444444444e-07, "loss": 0.0002, "step": 408000 }, { "epoch": 2.9, "learning_rate": 9.6384e-07, "loss": 0.0002, "step": 408500 }, { "epoch": 2.91, "learning_rate": 9.282844444444445e-07, "loss": 0.0002, "step": 409000 }, { "epoch": 2.91, "learning_rate": 8.927288888888889e-07, "loss": 0.0002, "step": 409500 }, { "epoch": 2.92, "learning_rate": 8.571733333333334e-07, "loss": 0.0002, "step": 410000 }, { "epoch": 2.92, "learning_rate": 8.216177777777778e-07, "loss": 0.0002, "step": 410500 }, { "epoch": 2.92, "learning_rate": 7.860622222222223e-07, "loss": 0.0002, "step": 411000 }, { "epoch": 2.93, "learning_rate": 7.505777777777778e-07, "loss": 0.0002, "step": 411500 }, { "epoch": 2.93, "learning_rate": 7.150222222222222e-07, "loss": 0.0002, "step": 412000 }, { "epoch": 2.93, "learning_rate": 6.794666666666667e-07, "loss": 0.0002, "step": 412500 }, { "epoch": 2.94, "learning_rate": 6.439111111111111e-07, "loss": 0.0002, "step": 413000 }, { "epoch": 2.94, "learning_rate": 6.084266666666667e-07, "loss": 0.0002, "step": 413500 }, { "epoch": 2.94, "learning_rate": 5.728711111111112e-07, "loss": 0.0002, "step": 414000 }, { "epoch": 2.95, "learning_rate": 5.373155555555556e-07, "loss": 0.0002, "step": 414500 }, { "epoch": 2.95, "learning_rate": 5.0176e-07, "loss": 0.0002, "step": 415000 }, { "epoch": 2.95, "learning_rate": 4.6620444444444444e-07, "loss": 0.0002, "step": 415500 }, { "epoch": 2.96, "learning_rate": 4.3072e-07, "loss": 0.0002, "step": 416000 }, { "epoch": 2.96, "learning_rate": 3.9516444444444446e-07, "loss": 0.0002, "step": 416500 }, { "epoch": 2.97, "learning_rate": 3.596088888888889e-07, "loss": 0.0002, "step": 417000 }, { "epoch": 2.97, "learning_rate": 3.2405333333333337e-07, "loss": 0.0002, "step": 417500 }, { "epoch": 2.97, "learning_rate": 2.884977777777778e-07, "loss": 0.0002, "step": 418000 }, { "epoch": 2.98, "learning_rate": 2.5301333333333333e-07, "loss": 0.0002, "step": 418500 }, { "epoch": 2.98, "learning_rate": 2.1745777777777779e-07, "loss": 0.0002, "step": 419000 }, { "epoch": 2.98, "learning_rate": 1.819022222222222e-07, "loss": 0.0002, "step": 419500 }, { "epoch": 2.99, "learning_rate": 1.4634666666666666e-07, "loss": 0.0002, "step": 420000 }, { "epoch": 2.99, "learning_rate": 1.1086222222222222e-07, "loss": 0.0002, "step": 420500 }, { "epoch": 2.99, "learning_rate": 7.530666666666667e-08, "loss": 0.0002, "step": 421000 }, { "epoch": 3.0, "learning_rate": 3.975111111111112e-08, "loss": 0.0002, "step": 421500 }, { "epoch": 3.0, "step": 421875, "total_flos": 0, "train_runtime": 120438.6398, "train_samples_per_second": 3.503 } ], "max_steps": 421875, "num_train_epochs": 3, "total_flos": 0, "trial_name": null, "trial_params": null }