diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28414 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.999948237486413,
+  "global_step": 473312,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999260530052059e-05,
+      "loss": 2.919,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9985210601041174e-05,
+      "loss": 2.7423,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9977815901561766e-05,
+      "loss": 2.6872,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997042120208235e-05,
+      "loss": 2.6637,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996302650260294e-05,
+      "loss": 2.6088,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995563180312352e-05,
+      "loss": 2.5799,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.994823710364411e-05,
+      "loss": 2.5585,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9940842404164694e-05,
+      "loss": 2.5208,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993344770468528e-05,
+      "loss": 2.5201,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992605300520587e-05,
+      "loss": 2.4521,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991865830572646e-05,
+      "loss": 2.4482,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9911263606247044e-05,
+      "loss": 2.4095,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.990386890676763e-05,
+      "loss": 2.4293,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989647420728822e-05,
+      "loss": 2.4321,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988907950780881e-05,
+      "loss": 2.3756,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.988168480832939e-05,
+      "loss": 2.3572,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987429010884998e-05,
+      "loss": 2.3661,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9866895409370564e-05,
+      "loss": 2.3842,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9859500709891157e-05,
+      "loss": 2.3186,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9852106010411735e-05,
+      "loss": 2.3267,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984471131093232e-05,
+      "loss": 2.3331,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9837316611452914e-05,
+      "loss": 2.2617,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98299219119735e-05,
+      "loss": 2.3182,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9822527212494085e-05,
+      "loss": 2.3079,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981513251301467e-05,
+      "loss": 2.2781,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980773781353526e-05,
+      "loss": 2.2769,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980034311405585e-05,
+      "loss": 2.3029,
+      "step": 2700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9792948414576434e-05,
+      "loss": 2.2748,
+      "step": 2800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978555371509702e-05,
+      "loss": 2.2387,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977815901561761e-05,
+      "loss": 2.2111,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97707643161382e-05,
+      "loss": 2.2477,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9763369616658783e-05,
+      "loss": 2.229,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975597491717937e-05,
+      "loss": 2.256,
+      "step": 3300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9748580217699955e-05,
+      "loss": 2.2034,
+      "step": 3400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974118551822054e-05,
+      "loss": 2.241,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9733790818741126e-05,
+      "loss": 2.2128,
+      "step": 3600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972639611926171e-05,
+      "loss": 2.2185,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9719001419782304e-05,
+      "loss": 2.2213,
+      "step": 3800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.971160672030289e-05,
+      "loss": 2.1884,
+      "step": 3900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9704212020823475e-05,
+      "loss": 2.1957,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.969681732134406e-05,
+      "loss": 2.204,
+      "step": 4100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.968949656885945e-05,
+      "loss": 2.1919,
+      "step": 4200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9682101869380034e-05,
+      "loss": 2.1941,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.967470716990062e-05,
+      "loss": 2.1981,
+      "step": 4400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9667312470421205e-05,
+      "loss": 2.1546,
+      "step": 4500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.965991777094179e-05,
+      "loss": 2.1156,
+      "step": 4600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9652523071462376e-05,
+      "loss": 2.142,
+      "step": 4700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.964512837198296e-05,
+      "loss": 2.1557,
+      "step": 4800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963773367250355e-05,
+      "loss": 2.1273,
+      "step": 4900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.963033897302414e-05,
+      "loss": 2.1309,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9622944273544726e-05,
+      "loss": 2.1652,
+      "step": 5100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961554957406531e-05,
+      "loss": 2.149,
+      "step": 5200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96081548745859e-05,
+      "loss": 2.1456,
+      "step": 5300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.960076017510649e-05,
+      "loss": 2.1594,
+      "step": 5400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9593365475627075e-05,
+      "loss": 2.1398,
+      "step": 5500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.958597077614766e-05,
+      "loss": 2.1148,
+      "step": 5600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9578576076668246e-05,
+      "loss": 2.1583,
+      "step": 5700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957118137718884e-05,
+      "loss": 2.1114,
+      "step": 5800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9563786677709424e-05,
+      "loss": 2.1155,
+      "step": 5900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955639197823e-05,
+      "loss": 2.1098,
+      "step": 6000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954899727875059e-05,
+      "loss": 2.0925,
+      "step": 6100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9541676526265976e-05,
+      "loss": 2.1266,
+      "step": 6200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953428182678656e-05,
+      "loss": 2.1477,
+      "step": 6300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952688712730715e-05,
+      "loss": 2.1214,
+      "step": 6400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.951949242782773e-05,
+      "loss": 2.093,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9512097728348325e-05,
+      "loss": 2.0955,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.950470302886891e-05,
+      "loss": 2.0927,
+      "step": 6700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94973083293895e-05,
+      "loss": 2.0823,
+      "step": 6800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.948991362991008e-05,
+      "loss": 2.0801,
+      "step": 6900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.948251893043067e-05,
+      "loss": 2.1013,
+      "step": 7000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.947512423095126e-05,
+      "loss": 2.0966,
+      "step": 7100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9467729531471846e-05,
+      "loss": 2.0613,
+      "step": 7200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9460334831992425e-05,
+      "loss": 2.063,
+      "step": 7300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.945294013251302e-05,
+      "loss": 2.0941,
+      "step": 7400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.94455454330336e-05,
+      "loss": 2.0996,
+      "step": 7500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.943815073355419e-05,
+      "loss": 2.0831,
+      "step": 7600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9430756034074774e-05,
+      "loss": 2.0941,
+      "step": 7700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942336133459537e-05,
+      "loss": 2.1278,
+      "step": 7800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.941596663511595e-05,
+      "loss": 2.0738,
+      "step": 7900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.940857193563654e-05,
+      "loss": 2.0666,
+      "step": 8000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9401177236157124e-05,
+      "loss": 2.0527,
+      "step": 8100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9393782536677716e-05,
+      "loss": 2.0532,
+      "step": 8200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.93863878371983e-05,
+      "loss": 2.0297,
+      "step": 8300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.937899313771889e-05,
+      "loss": 2.0484,
+      "step": 8400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937159843823947e-05,
+      "loss": 2.0632,
+      "step": 8500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.936420373876006e-05,
+      "loss": 2.0324,
+      "step": 8600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.935688298627544e-05,
+      "loss": 2.0349,
+      "step": 8700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9349488286796025e-05,
+      "loss": 2.0879,
+      "step": 8800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.934209358731661e-05,
+      "loss": 2.0388,
+      "step": 8900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.93346988878372e-05,
+      "loss": 2.0342,
+      "step": 9000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.932730418835779e-05,
+      "loss": 2.0462,
+      "step": 9100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9319909488878374e-05,
+      "loss": 2.0469,
+      "step": 9200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.931251478939896e-05,
+      "loss": 2.0628,
+      "step": 9300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.930512008991955e-05,
+      "loss": 2.0183,
+      "step": 9400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.929772539044014e-05,
+      "loss": 2.0261,
+      "step": 9500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.929033069096072e-05,
+      "loss": 2.0298,
+      "step": 9600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.928293599148131e-05,
+      "loss": 2.0427,
+      "step": 9700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9275541292001895e-05,
+      "loss": 2.0566,
+      "step": 9800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.926814659252249e-05,
+      "loss": 1.998,
+      "step": 9900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9260751893043066e-05,
+      "loss": 2.0508,
+      "step": 10000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.925335719356365e-05,
+      "loss": 1.9931,
+      "step": 10100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9245962494084244e-05,
+      "loss": 2.04,
+      "step": 10200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.923856779460483e-05,
+      "loss": 1.9981,
+      "step": 10300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9231173095125415e-05,
+      "loss": 2.0008,
+      "step": 10400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9223778395646e-05,
+      "loss": 1.9967,
+      "step": 10500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.921638369616659e-05,
+      "loss": 2.013,
+      "step": 10600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.920898899668718e-05,
+      "loss": 2.0167,
+      "step": 10700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9201594297207764e-05,
+      "loss": 2.012,
+      "step": 10800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.919419959772835e-05,
+      "loss": 1.9876,
+      "step": 10900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.918680489824894e-05,
+      "loss": 2.0142,
+      "step": 11000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.917941019876953e-05,
+      "loss": 1.9965,
+      "step": 11100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.917208944628491e-05,
+      "loss": 1.9778,
+      "step": 11200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.916469474680549e-05,
+      "loss": 2.0163,
+      "step": 11300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.915730004732608e-05,
+      "loss": 2.0398,
+      "step": 11400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9149905347846666e-05,
+      "loss": 2.0177,
+      "step": 11500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.914251064836725e-05,
+      "loss": 1.9847,
+      "step": 11600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.913511594888784e-05,
+      "loss": 1.9609,
+      "step": 11700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.912772124940843e-05,
+      "loss": 1.9641,
+      "step": 11800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9120326549929015e-05,
+      "loss": 2.0005,
+      "step": 11900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.91129318504496e-05,
+      "loss": 1.9991,
+      "step": 12000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9105537150970186e-05,
+      "loss": 1.9912,
+      "step": 12100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.909814245149077e-05,
+      "loss": 1.987,
+      "step": 12200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9090747752011364e-05,
+      "loss": 1.9887,
+      "step": 12300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.908335305253195e-05,
+      "loss": 1.9956,
+      "step": 12400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9075958353052535e-05,
+      "loss": 1.9633,
+      "step": 12500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.906856365357312e-05,
+      "loss": 1.9802,
+      "step": 12600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.906116895409371e-05,
+      "loss": 2.0055,
+      "step": 12700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.905377425461429e-05,
+      "loss": 1.9626,
+      "step": 12800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.904637955513488e-05,
+      "loss": 1.9506,
+      "step": 12900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.903898485565547e-05,
+      "loss": 1.963,
+      "step": 13000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9031590156176056e-05,
+      "loss": 1.9647,
+      "step": 13100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9024269403691436e-05,
+      "loss": 1.9834,
+      "step": 13200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.901687470421202e-05,
+      "loss": 1.9763,
+      "step": 13300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.900948000473261e-05,
+      "loss": 1.971,
+      "step": 13400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.90020853052532e-05,
+      "loss": 1.9911,
+      "step": 13500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8994690605773786e-05,
+      "loss": 1.9565,
+      "step": 13600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.898729590629437e-05,
+      "loss": 1.9361,
+      "step": 13700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.897990120681496e-05,
+      "loss": 1.9895,
+      "step": 13800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.897250650733554e-05,
+      "loss": 1.9424,
+      "step": 13900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.896511180785613e-05,
+      "loss": 1.9614,
+      "step": 14000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8957717108376714e-05,
+      "loss": 1.9662,
+      "step": 14100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8950322408897306e-05,
+      "loss": 2.0129,
+      "step": 14200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.894292770941789e-05,
+      "loss": 1.9652,
+      "step": 14300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.893553300993848e-05,
+      "loss": 1.9475,
+      "step": 14400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.892813831045906e-05,
+      "loss": 1.9473,
+      "step": 14500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8920743610979656e-05,
+      "loss": 1.944,
+      "step": 14600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.891334891150024e-05,
+      "loss": 1.9409,
+      "step": 14700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.890595421202083e-05,
+      "loss": 1.949,
+      "step": 14800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.889855951254141e-05,
+      "loss": 1.9782,
+      "step": 14900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8891164813062e-05,
+      "loss": 1.9485,
+      "step": 15000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.888377011358259e-05,
+      "loss": 1.9484,
+      "step": 15100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.887637541410317e-05,
+      "loss": 1.9551,
+      "step": 15200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8868980714623755e-05,
+      "loss": 1.9261,
+      "step": 15300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.886158601514435e-05,
+      "loss": 1.9625,
+      "step": 15400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.885419131566493e-05,
+      "loss": 1.9619,
+      "step": 15500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.884679661618552e-05,
+      "loss": 1.9496,
+      "step": 15600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8839401916706105e-05,
+      "loss": 1.975,
+      "step": 15700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.88320072172267e-05,
+      "loss": 1.9324,
+      "step": 15800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.882461251774728e-05,
+      "loss": 1.9643,
+      "step": 15900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.881721781826787e-05,
+      "loss": 1.9366,
+      "step": 16000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8809823118788454e-05,
+      "loss": 1.934,
+      "step": 16100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8802502366303834e-05,
+      "loss": 1.9236,
+      "step": 16200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.879510766682443e-05,
+      "loss": 1.9284,
+      "step": 16300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.878771296734501e-05,
+      "loss": 1.9677,
+      "step": 16400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.87803182678656e-05,
+      "loss": 1.9315,
+      "step": 16500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8772923568386184e-05,
+      "loss": 1.9345,
+      "step": 16600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.876552886890677e-05,
+      "loss": 1.9373,
+      "step": 16700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8758134169427355e-05,
+      "loss": 1.932,
+      "step": 16800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.875073946994794e-05,
+      "loss": 1.9039,
+      "step": 16900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.874334477046853e-05,
+      "loss": 1.9559,
+      "step": 17000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.873595007098912e-05,
+      "loss": 1.9104,
+      "step": 17100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.8728555371509704e-05,
+      "loss": 1.9314,
+      "step": 17200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.872116067203029e-05,
+      "loss": 1.9595,
+      "step": 17300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8713765972550876e-05,
+      "loss": 1.9264,
+      "step": 17400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.870637127307147e-05,
+      "loss": 1.8972,
+      "step": 17500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8698976573592054e-05,
+      "loss": 1.9434,
+      "step": 17600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.869158187411264e-05,
+      "loss": 1.9297,
+      "step": 17700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8684187174633225e-05,
+      "loss": 1.9031,
+      "step": 17800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.867679247515381e-05,
+      "loss": 1.9358,
+      "step": 17900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8669397775674396e-05,
+      "loss": 1.8736,
+      "step": 18000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.866200307619498e-05,
+      "loss": 1.9253,
+      "step": 18100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8654608376715574e-05,
+      "loss": 1.9201,
+      "step": 18200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.864721367723616e-05,
+      "loss": 1.9401,
+      "step": 18300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8639818977756745e-05,
+      "loss": 1.9467,
+      "step": 18400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8632498225272126e-05,
+      "loss": 1.9439,
+      "step": 18500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.862510352579271e-05,
+      "loss": 1.9153,
+      "step": 18600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8617708826313304e-05,
+      "loss": 1.9149,
+      "step": 18700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.861031412683389e-05,
+      "loss": 1.9285,
+      "step": 18800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8602919427354475e-05,
+      "loss": 1.9138,
+      "step": 18900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.859552472787506e-05,
+      "loss": 1.9528,
+      "step": 19000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.858813002839565e-05,
+      "loss": 1.8888,
+      "step": 19100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.858073532891623e-05,
+      "loss": 1.9066,
+      "step": 19200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857334062943682e-05,
+      "loss": 1.932,
+      "step": 19300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.856594592995741e-05,
+      "loss": 1.9112,
+      "step": 19400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8558551230477996e-05,
+      "loss": 1.9011,
+      "step": 19500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.855115653099858e-05,
+      "loss": 1.9064,
+      "step": 19600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.854376183151917e-05,
+      "loss": 1.9445,
+      "step": 19700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.853636713203976e-05,
+      "loss": 1.9436,
+      "step": 19800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8528972432560345e-05,
+      "loss": 1.8964,
+      "step": 19900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.852157773308093e-05,
+      "loss": 1.8949,
+      "step": 20000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8514183033601516e-05,
+      "loss": 1.9179,
+      "step": 20100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.85067883341221e-05,
+      "loss": 1.9006,
+      "step": 20200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8499393634642694e-05,
+      "loss": 1.8683,
+      "step": 20300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.849199893516328e-05,
+      "loss": 1.8887,
+      "step": 20400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.848460423568386e-05,
+      "loss": 1.8973,
+      "step": 20500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.847720953620445e-05,
+      "loss": 1.8831,
+      "step": 20600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.846981483672504e-05,
+      "loss": 1.8713,
+      "step": 20700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.846242013724562e-05,
+      "loss": 1.9004,
+      "step": 20800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.845502543776621e-05,
+      "loss": 1.8996,
+      "step": 20900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.84476307382868e-05,
+      "loss": 1.9057,
+      "step": 21000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8440236038807386e-05,
+      "loss": 1.9165,
+      "step": 21100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.843284133932797e-05,
+      "loss": 1.885,
+      "step": 21200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.842544663984856e-05,
+      "loss": 1.8674,
+      "step": 21300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.841812588736394e-05,
+      "loss": 1.9026,
+      "step": 21400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.841073118788453e-05,
+      "loss": 1.8985,
+      "step": 21500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8403336488405116e-05,
+      "loss": 1.8923,
+      "step": 21600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.83959417889257e-05,
+      "loss": 1.8978,
+      "step": 21700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.838854708944629e-05,
+      "loss": 1.8871,
+      "step": 21800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.838115238996687e-05,
+      "loss": 1.8718,
+      "step": 21900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.837375769048746e-05,
+      "loss": 1.9119,
+      "step": 22000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8366362991008044e-05,
+      "loss": 1.8625,
+      "step": 22100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.835896829152864e-05,
+      "loss": 1.8834,
+      "step": 22200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.835157359204922e-05,
+      "loss": 1.8734,
+      "step": 22300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.834417889256981e-05,
+      "loss": 1.8839,
+      "step": 22400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8336784193090394e-05,
+      "loss": 1.8718,
+      "step": 22500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.832938949361098e-05,
+      "loss": 1.9251,
+      "step": 22600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.832199479413157e-05,
+      "loss": 1.8665,
+      "step": 22700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.831460009465216e-05,
+      "loss": 1.8725,
+      "step": 22800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.830720539517274e-05,
+      "loss": 1.881,
+      "step": 22900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.829981069569333e-05,
+      "loss": 1.875,
+      "step": 23000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.829241599621392e-05,
+      "loss": 1.8611,
+      "step": 23100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.82850212967345e-05,
+      "loss": 1.8724,
+      "step": 23200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8277626597255086e-05,
+      "loss": 1.8807,
+      "step": 23300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827023189777568e-05,
+      "loss": 1.8552,
+      "step": 23400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8262837198296264e-05,
+      "loss": 1.853,
+      "step": 23500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.825544249881685e-05,
+      "loss": 1.8351,
+      "step": 23600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8248047799337435e-05,
+      "loss": 1.875,
+      "step": 23700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.824065309985803e-05,
+      "loss": 1.8807,
+      "step": 23800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.823325840037861e-05,
+      "loss": 1.8734,
+      "step": 23900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.82258637008992e-05,
+      "loss": 1.8642,
+      "step": 24000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8218469001419784e-05,
+      "loss": 1.8599,
+      "step": 24100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8211074301940377e-05,
+      "loss": 1.8656,
+      "step": 24200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.820367960246096e-05,
+      "loss": 1.8629,
+      "step": 24300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.819628490298155e-05,
+      "loss": 1.8598,
+      "step": 24400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.818889020350213e-05,
+      "loss": 1.8552,
+      "step": 24500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.818149550402272e-05,
+      "loss": 1.8565,
+      "step": 24600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.81741747515381e-05,
+      "loss": 1.8701,
+      "step": 24700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8166780052058685e-05,
+      "loss": 1.8765,
+      "step": 24800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815938535257927e-05,
+      "loss": 1.8726,
+      "step": 24900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.815199065309986e-05,
+      "loss": 1.8706,
+      "step": 25000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.814459595362045e-05,
+      "loss": 1.873,
+      "step": 25100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8137201254141035e-05,
+      "loss": 1.8408,
+      "step": 25200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.812980655466162e-05,
+      "loss": 1.8744,
+      "step": 25300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8122411855182206e-05,
+      "loss": 1.885,
+      "step": 25400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.81150171557028e-05,
+      "loss": 1.8794,
+      "step": 25500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8107622456223384e-05,
+      "loss": 1.842,
+      "step": 25600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.810022775674397e-05,
+      "loss": 1.8753,
+      "step": 25700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8092833057264555e-05,
+      "loss": 1.8326,
+      "step": 25800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.808543835778514e-05,
+      "loss": 1.8327,
+      "step": 25900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8078043658305726e-05,
+      "loss": 1.8257,
+      "step": 26000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.807064895882631e-05,
+      "loss": 1.8652,
+      "step": 26100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8063254259346904e-05,
+      "loss": 1.8509,
+      "step": 26200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.805585955986749e-05,
+      "loss": 1.8735,
+      "step": 26300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8048464860388076e-05,
+      "loss": 1.8531,
+      "step": 26400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.804107016090866e-05,
+      "loss": 1.8617,
+      "step": 26500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8033675461429254e-05,
+      "loss": 1.8583,
+      "step": 26600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.802628076194984e-05,
+      "loss": 1.8491,
+      "step": 26700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8018886062470425e-05,
+      "loss": 1.8717,
+      "step": 26800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8011565309985806e-05,
+      "loss": 1.8676,
+      "step": 26900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.800417061050639e-05,
+      "loss": 1.8635,
+      "step": 27000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.799677591102698e-05,
+      "loss": 1.8291,
+      "step": 27100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.798938121154756e-05,
+      "loss": 1.859,
+      "step": 27200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.798198651206815e-05,
+      "loss": 1.8158,
+      "step": 27300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.797459181258874e-05,
+      "loss": 1.8652,
+      "step": 27400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7967197113109326e-05,
+      "loss": 1.8637,
+      "step": 27500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.795980241362991e-05,
+      "loss": 1.8458,
+      "step": 27600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.79524077141505e-05,
+      "loss": 1.8086,
+      "step": 27700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.794501301467108e-05,
+      "loss": 1.8499,
+      "step": 27800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.7937618315191675e-05,
+      "loss": 1.879,
+      "step": 27900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.793022361571226e-05,
+      "loss": 1.8575,
+      "step": 28000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.792282891623285e-05,
+      "loss": 1.8326,
+      "step": 28100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.791543421675343e-05,
+      "loss": 1.875,
+      "step": 28200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7908039517274025e-05,
+      "loss": 1.8682,
+      "step": 28300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.790064481779461e-05,
+      "loss": 1.8373,
+      "step": 28400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.789325011831519e-05,
+      "loss": 1.7958,
+      "step": 28500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.788585541883578e-05,
+      "loss": 1.8596,
+      "step": 28600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.787846071935637e-05,
+      "loss": 1.8747,
+      "step": 28700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.787106601987695e-05,
+      "loss": 1.8629,
+      "step": 28800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.786367132039754e-05,
+      "loss": 1.8307,
+      "step": 28900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.785627662091813e-05,
+      "loss": 1.8578,
+      "step": 29000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.784888192143872e-05,
+      "loss": 1.8484,
+      "step": 29100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.78414872219593e-05,
+      "loss": 1.8428,
+      "step": 29200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.783409252247989e-05,
+      "loss": 1.8385,
+      "step": 29300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7826697823000474e-05,
+      "loss": 1.8431,
+      "step": 29400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7819303123521066e-05,
+      "loss": 1.8292,
+      "step": 29500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.781190842404165e-05,
+      "loss": 1.8305,
+      "step": 29600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.780451372456224e-05,
+      "loss": 1.8412,
+      "step": 29700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.779711902508282e-05,
+      "loss": 1.8244,
+      "step": 29800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.778972432560341e-05,
+      "loss": 1.8344,
+      "step": 29900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.7782329626123994e-05,
+      "loss": 1.8502,
+      "step": 30000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.777493492664458e-05,
+      "loss": 1.8605,
+      "step": 30100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.776754022716517e-05,
+      "loss": 1.8364,
+      "step": 30200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.776021947468055e-05,
+      "loss": 1.8621,
+      "step": 30300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.775282477520114e-05,
+      "loss": 1.8395,
+      "step": 30400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7745430075721724e-05,
+      "loss": 1.7994,
+      "step": 30500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.773803537624231e-05,
+      "loss": 1.8275,
+      "step": 30600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.77306406767629e-05,
+      "loss": 1.8363,
+      "step": 30700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.772324597728349e-05,
+      "loss": 1.8223,
+      "step": 30800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.771585127780407e-05,
+      "loss": 1.8288,
+      "step": 30900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.770845657832466e-05,
+      "loss": 1.8547,
+      "step": 31000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7701061878845245e-05,
+      "loss": 1.8582,
+      "step": 31100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.769366717936583e-05,
+      "loss": 1.8086,
+      "step": 31200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.7686272479886416e-05,
+      "loss": 1.8162,
+      "step": 31300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.767887778040701e-05,
+      "loss": 1.8085,
+      "step": 31400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7671483080927594e-05,
+      "loss": 1.8304,
+      "step": 31500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.766408838144818e-05,
+      "loss": 1.8654,
+      "step": 31600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7656693681968765e-05,
+      "loss": 1.8079,
+      "step": 31700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.764929898248936e-05,
+      "loss": 1.8224,
+      "step": 31800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.764190428300994e-05,
+      "loss": 1.8346,
+      "step": 31900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.763450958353053e-05,
+      "loss": 1.8433,
+      "step": 32000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.7627114884051115e-05,
+      "loss": 1.8315,
+      "step": 32100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.76197201845717e-05,
+      "loss": 1.8166,
+      "step": 32200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.761232548509229e-05,
+      "loss": 1.8255,
+      "step": 32300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.760493078561287e-05,
+      "loss": 1.7861,
+      "step": 32400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.759753608613346e-05,
+      "loss": 1.8063,
+      "step": 32500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.759014138665405e-05,
+      "loss": 1.8162,
+      "step": 32600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.758282063416943e-05,
+      "loss": 1.8133,
+      "step": 32700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7575425934690016e-05,
+      "loss": 1.8252,
+      "step": 32800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.75680312352106e-05,
+      "loss": 1.8241,
+      "step": 32900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.756063653573119e-05,
+      "loss": 1.8034,
+      "step": 33000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.755324183625178e-05,
+      "loss": 1.8063,
+      "step": 33100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7545847136772365e-05,
+      "loss": 1.846,
+      "step": 33200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.753845243729295e-05,
+      "loss": 1.8431,
+      "step": 33300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7531057737813536e-05,
+      "loss": 1.8001,
+      "step": 33400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.752366303833413e-05,
+      "loss": 1.8041,
+      "step": 33500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7516268338854714e-05,
+      "loss": 1.8031,
+      "step": 33600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.750887363937529e-05,
+      "loss": 1.8613,
+      "step": 33700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.7501478939895885e-05,
+      "loss": 1.8254,
+      "step": 33800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.749408424041647e-05,
+      "loss": 1.8084,
+      "step": 33900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.748668954093706e-05,
+      "loss": 1.8285,
+      "step": 34000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.747929484145764e-05,
+      "loss": 1.8077,
+      "step": 34100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7471900141978235e-05,
+      "loss": 1.797,
+      "step": 34200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.746450544249882e-05,
+      "loss": 1.8165,
+      "step": 34300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7457110743019406e-05,
+      "loss": 1.8253,
+      "step": 34400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.744971604353999e-05,
+      "loss": 1.8192,
+      "step": 34500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.744232134406058e-05,
+      "loss": 1.817,
+      "step": 34600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.743492664458117e-05,
+      "loss": 1.8394,
+      "step": 34700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7427531945101755e-05,
+      "loss": 1.8201,
+      "step": 34800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.742013724562234e-05,
+      "loss": 1.8149,
+      "step": 34900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.741274254614293e-05,
+      "loss": 1.8006,
+      "step": 35000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.740534784666351e-05,
+      "loss": 1.8284,
+      "step": 35100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.73979531471841e-05,
+      "loss": 1.834,
+      "step": 35200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7390558447704684e-05,
+      "loss": 1.8289,
+      "step": 35300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7383163748225276e-05,
+      "loss": 1.776,
+      "step": 35400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.737576904874586e-05,
+      "loss": 1.7996,
+      "step": 35500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736837434926645e-05,
+      "loss": 1.8297,
+      "step": 35600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736105359678183e-05,
+      "loss": 1.8323,
+      "step": 35700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735365889730241e-05,
+      "loss": 1.7792,
+      "step": 35800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7346264197823006e-05,
+      "loss": 1.8141,
+      "step": 35900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733886949834359e-05,
+      "loss": 1.8022,
+      "step": 36000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733147479886418e-05,
+      "loss": 1.8013,
+      "step": 36100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.732408009938476e-05,
+      "loss": 1.8144,
+      "step": 36200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7316685399905355e-05,
+      "loss": 1.8317,
+      "step": 36300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7309290700425934e-05,
+      "loss": 1.7999,
+      "step": 36400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.730189600094652e-05,
+      "loss": 1.838,
+      "step": 36500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.729450130146711e-05,
+      "loss": 1.8051,
+      "step": 36600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.72871066019877e-05,
+      "loss": 1.828,
+      "step": 36700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.727971190250828e-05,
+      "loss": 1.7934,
+      "step": 36800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727231720302887e-05,
+      "loss": 1.8061,
+      "step": 36900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.726492250354946e-05,
+      "loss": 1.8287,
+      "step": 37000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.725752780407005e-05,
+      "loss": 1.7928,
+      "step": 37100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.725013310459063e-05,
+      "loss": 1.797,
+      "step": 37200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.724273840511122e-05,
+      "loss": 1.8524,
+      "step": 37300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7235343705631804e-05,
+      "loss": 1.7954,
+      "step": 37400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7227949006152396e-05,
+      "loss": 1.8148,
+      "step": 37500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.722055430667298e-05,
+      "loss": 1.808,
+      "step": 37600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.721315960719356e-05,
+      "loss": 1.7973,
+      "step": 37700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.720576490771415e-05,
+      "loss": 1.8017,
+      "step": 37800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.719837020823474e-05,
+      "loss": 1.8122,
+      "step": 37900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7190975508755325e-05,
+      "loss": 1.8046,
+      "step": 38000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.718358080927591e-05,
+      "loss": 1.7849,
+      "step": 38100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.71761861097965e-05,
+      "loss": 1.8101,
+      "step": 38200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.716879141031709e-05,
+      "loss": 1.7937,
+      "step": 38300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7161396710837674e-05,
+      "loss": 1.8027,
+      "step": 38400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.715400201135826e-05,
+      "loss": 1.8289,
+      "step": 38500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.714660731187885e-05,
+      "loss": 1.7869,
+      "step": 38600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.713921261239944e-05,
+      "loss": 1.7739,
+      "step": 38700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.713181791292002e-05,
+      "loss": 1.7466,
+      "step": 38800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.712442321344061e-05,
+      "loss": 1.8165,
+      "step": 38900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7117028513961194e-05,
+      "loss": 1.7868,
+      "step": 39000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.710963381448178e-05,
+      "loss": 1.7665,
+      "step": 39100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.710231306199716e-05,
+      "loss": 1.7795,
+      "step": 39200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7094918362517746e-05,
+      "loss": 1.7745,
+      "step": 39300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.708752366303834e-05,
+      "loss": 1.7927,
+      "step": 39400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7080128963558924e-05,
+      "loss": 1.7867,
+      "step": 39500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707273426407951e-05,
+      "loss": 1.7866,
+      "step": 39600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7065339564600095e-05,
+      "loss": 1.7803,
+      "step": 39700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.705794486512068e-05,
+      "loss": 1.7881,
+      "step": 39800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7050550165641274e-05,
+      "loss": 1.7949,
+      "step": 39900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.704315546616186e-05,
+      "loss": 1.7869,
+      "step": 40000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7035760766682445e-05,
+      "loss": 1.7688,
+      "step": 40100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.702836606720303e-05,
+      "loss": 1.8111,
+      "step": 40200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7020971367723616e-05,
+      "loss": 1.7692,
+      "step": 40300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.70135766682442e-05,
+      "loss": 1.7678,
+      "step": 40400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.700618196876479e-05,
+      "loss": 1.7915,
+      "step": 40500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.699878726928538e-05,
+      "loss": 1.761,
+      "step": 40600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.6991392569805965e-05,
+      "loss": 1.7747,
+      "step": 40700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.698399787032655e-05,
+      "loss": 1.7642,
+      "step": 40800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.697660317084714e-05,
+      "loss": 1.7973,
+      "step": 40900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.696920847136773e-05,
+      "loss": 1.7942,
+      "step": 41000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6961813771888315e-05,
+      "loss": 1.7908,
+      "step": 41100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.69544190724089e-05,
+      "loss": 1.7836,
+      "step": 41200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6947024372929486e-05,
+      "loss": 1.803,
+      "step": 41300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.693962967345008e-05,
+      "loss": 1.7714,
+      "step": 41400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6932234973970664e-05,
+      "loss": 1.7912,
+      "step": 41500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692484027449125e-05,
+      "loss": 1.7617,
+      "step": 41600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.691744557501183e-05,
+      "loss": 1.7715,
+      "step": 41700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6910124822527216e-05,
+      "loss": 1.7794,
+      "step": 41800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69027301230478e-05,
+      "loss": 1.7629,
+      "step": 41900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.689533542356839e-05,
+      "loss": 1.7977,
+      "step": 42000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.688794072408897e-05,
+      "loss": 1.7731,
+      "step": 42100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6880546024609565e-05,
+      "loss": 1.7539,
+      "step": 42200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687315132513015e-05,
+      "loss": 1.788,
+      "step": 42300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6865756625650736e-05,
+      "loss": 1.7804,
+      "step": 42400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685836192617132e-05,
+      "loss": 1.7973,
+      "step": 42500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685096722669191e-05,
+      "loss": 1.8139,
+      "step": 42600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68435725272125e-05,
+      "loss": 1.7456,
+      "step": 42700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6836177827733086e-05,
+      "loss": 1.7873,
+      "step": 42800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.682878312825367e-05,
+      "loss": 1.7842,
+      "step": 42900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.682138842877426e-05,
+      "loss": 1.7672,
+      "step": 43000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.681399372929484e-05,
+      "loss": 1.7784,
+      "step": 43100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.680659902981543e-05,
+      "loss": 1.7817,
+      "step": 43200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6799204330336014e-05,
+      "loss": 1.7407,
+      "step": 43300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6791809630856606e-05,
+      "loss": 1.7646,
+      "step": 43400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.678441493137719e-05,
+      "loss": 1.7657,
+      "step": 43500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.677702023189778e-05,
+      "loss": 1.7983,
+      "step": 43600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.676962553241836e-05,
+      "loss": 1.743,
+      "step": 43700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6762230832938956e-05,
+      "loss": 1.8009,
+      "step": 43800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6754910080454336e-05,
+      "loss": 1.8079,
+      "step": 43900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.674751538097492e-05,
+      "loss": 1.7845,
+      "step": 44000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.674012068149551e-05,
+      "loss": 1.7908,
+      "step": 44100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.673272598201609e-05,
+      "loss": 1.7645,
+      "step": 44200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.672533128253668e-05,
+      "loss": 1.792,
+      "step": 44300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6717936583057264e-05,
+      "loss": 1.7845,
+      "step": 44400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.671054188357785e-05,
+      "loss": 1.7503,
+      "step": 44500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.670314718409844e-05,
+      "loss": 1.7454,
+      "step": 44600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.669575248461903e-05,
+      "loss": 1.7689,
+      "step": 44700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6688357785139614e-05,
+      "loss": 1.7625,
+      "step": 44800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.66809630856602e-05,
+      "loss": 1.7851,
+      "step": 44900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6673568386180785e-05,
+      "loss": 1.7418,
+      "step": 45000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.666617368670138e-05,
+      "loss": 1.7588,
+      "step": 45100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.665877898722196e-05,
+      "loss": 1.7847,
+      "step": 45200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.665138428774255e-05,
+      "loss": 1.7753,
+      "step": 45300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6643989588263134e-05,
+      "loss": 1.782,
+      "step": 45400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.663659488878373e-05,
+      "loss": 1.7734,
+      "step": 45500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6629200189304306e-05,
+      "loss": 1.765,
+      "step": 45600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662180548982489e-05,
+      "loss": 1.7741,
+      "step": 45700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6614410790345484e-05,
+      "loss": 1.766,
+      "step": 45800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.660701609086607e-05,
+      "loss": 1.7594,
+      "step": 45900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6599621391386655e-05,
+      "loss": 1.781,
+      "step": 46000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.659222669190724e-05,
+      "loss": 1.7948,
+      "step": 46100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.658483199242783e-05,
+      "loss": 1.7626,
+      "step": 46200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.657743729294842e-05,
+      "loss": 1.7658,
+      "step": 46300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6570042593469004e-05,
+      "loss": 1.7611,
+      "step": 46400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656264789398959e-05,
+      "loss": 1.7607,
+      "step": 46500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655532714150497e-05,
+      "loss": 1.7708,
+      "step": 46600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654793244202556e-05,
+      "loss": 1.7282,
+      "step": 46700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654053774254615e-05,
+      "loss": 1.7611,
+      "step": 46800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653314304306673e-05,
+      "loss": 1.7682,
+      "step": 46900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.652574834358732e-05,
+      "loss": 1.7663,
+      "step": 47000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6518353644107905e-05,
+      "loss": 1.7679,
+      "step": 47100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.651095894462849e-05,
+      "loss": 1.7655,
+      "step": 47200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6503564245149076e-05,
+      "loss": 1.7753,
+      "step": 47300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.649616954566967e-05,
+      "loss": 1.7546,
+      "step": 47400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6488774846190255e-05,
+      "loss": 1.7677,
+      "step": 47500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.648138014671084e-05,
+      "loss": 1.786,
+      "step": 47600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6473985447231426e-05,
+      "loss": 1.7512,
+      "step": 47700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.646659074775201e-05,
+      "loss": 1.7708,
+      "step": 47800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6459196048272604e-05,
+      "loss": 1.7607,
+      "step": 47900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.645180134879319e-05,
+      "loss": 1.78,
+      "step": 48000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6444406649313775e-05,
+      "loss": 1.794,
+      "step": 48100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.643701194983436e-05,
+      "loss": 1.7604,
+      "step": 48200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6429617250354946e-05,
+      "loss": 1.7383,
+      "step": 48300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.642222255087553e-05,
+      "loss": 1.7602,
+      "step": 48400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.641482785139612e-05,
+      "loss": 1.7424,
+      "step": 48500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.640743315191671e-05,
+      "loss": 1.7605,
+      "step": 48600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.640011239943209e-05,
+      "loss": 1.771,
+      "step": 48700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.6392717699952676e-05,
+      "loss": 1.7781,
+      "step": 48800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.638532300047326e-05,
+      "loss": 1.7859,
+      "step": 48900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.637792830099385e-05,
+      "loss": 1.7836,
+      "step": 49000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.637053360151444e-05,
+      "loss": 1.7654,
+      "step": 49100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6363138902035026e-05,
+      "loss": 1.7866,
+      "step": 49200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.635574420255561e-05,
+      "loss": 1.7538,
+      "step": 49300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.63483495030762e-05,
+      "loss": 1.7573,
+      "step": 49400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.634095480359679e-05,
+      "loss": 1.7741,
+      "step": 49500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.633356010411737e-05,
+      "loss": 1.7288,
+      "step": 49600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6326165404637954e-05,
+      "loss": 1.789,
+      "step": 49700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6318770705158546e-05,
+      "loss": 1.7548,
+      "step": 49800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.631137600567913e-05,
+      "loss": 1.75,
+      "step": 49900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.630398130619972e-05,
+      "loss": 1.7316,
+      "step": 50000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.62965866067203e-05,
+      "loss": 1.7567,
+      "step": 50100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.628919190724089e-05,
+      "loss": 1.7842,
+      "step": 50200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.628179720776148e-05,
+      "loss": 1.7256,
+      "step": 50300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.627440250828207e-05,
+      "loss": 1.8103,
+      "step": 50400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.626700780880265e-05,
+      "loss": 1.753,
+      "step": 50500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.625961310932324e-05,
+      "loss": 1.7531,
+      "step": 50600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6252292356838625e-05,
+      "loss": 1.7673,
+      "step": 50700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.624489765735921e-05,
+      "loss": 1.6973,
+      "step": 50800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623750295787979e-05,
+      "loss": 1.75,
+      "step": 50900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6230108258400375e-05,
+      "loss": 1.7644,
+      "step": 51000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.622271355892097e-05,
+      "loss": 1.7536,
+      "step": 51100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6215318859441553e-05,
+      "loss": 1.7453,
+      "step": 51200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.620792415996214e-05,
+      "loss": 1.7729,
+      "step": 51300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6200529460482725e-05,
+      "loss": 1.7514,
+      "step": 51400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.619313476100332e-05,
+      "loss": 1.7211,
+      "step": 51500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.61857400615239e-05,
+      "loss": 1.777,
+      "step": 51600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.617834536204449e-05,
+      "loss": 1.7453,
+      "step": 51700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6170950662565074e-05,
+      "loss": 1.7488,
+      "step": 51800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6163555963085666e-05,
+      "loss": 1.7438,
+      "step": 51900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.615616126360625e-05,
+      "loss": 1.7071,
+      "step": 52000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.614876656412684e-05,
+      "loss": 1.7346,
+      "step": 52100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.614137186464742e-05,
+      "loss": 1.7436,
+      "step": 52200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613397716516801e-05,
+      "loss": 1.761,
+      "step": 52300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6126582465688595e-05,
+      "loss": 1.7495,
+      "step": 52400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.611918776620918e-05,
+      "loss": 1.7501,
+      "step": 52500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6111793066729766e-05,
+      "loss": 1.7356,
+      "step": 52600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.610439836725036e-05,
+      "loss": 1.7315,
+      "step": 52700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6097003667770944e-05,
+      "loss": 1.7563,
+      "step": 52800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.608960896829153e-05,
+      "loss": 1.7512,
+      "step": 52900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6082214268812115e-05,
+      "loss": 1.7517,
+      "step": 53000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.607481956933271e-05,
+      "loss": 1.7447,
+      "step": 53100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.606742486985329e-05,
+      "loss": 1.7181,
+      "step": 53200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.606003017037388e-05,
+      "loss": 1.7548,
+      "step": 53300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6052635470894465e-05,
+      "loss": 1.7634,
+      "step": 53400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.604524077141506e-05,
+      "loss": 1.761,
+      "step": 53500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6037846071935636e-05,
+      "loss": 1.7367,
+      "step": 53600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.603045137245622e-05,
+      "loss": 1.7466,
+      "step": 53700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6023056672976814e-05,
+      "loss": 1.7266,
+      "step": 53800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.60156619734974e-05,
+      "loss": 1.7491,
+      "step": 53900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6008267274017985e-05,
+      "loss": 1.7448,
+      "step": 54000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.600087257453857e-05,
+      "loss": 1.7238,
+      "step": 54100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.599347787505916e-05,
+      "loss": 1.7352,
+      "step": 54200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.598608317557975e-05,
+      "loss": 1.7386,
+      "step": 54300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5978688476100334e-05,
+      "loss": 1.7512,
+      "step": 54400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597129377662092e-05,
+      "loss": 1.7432,
+      "step": 54500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5963899077141506e-05,
+      "loss": 1.7693,
+      "step": 54600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.59565043776621e-05,
+      "loss": 1.7302,
+      "step": 54700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5949109678182684e-05,
+      "loss": 1.751,
+      "step": 54800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.594171497870326e-05,
+      "loss": 1.7279,
+      "step": 54900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5934320279223855e-05,
+      "loss": 1.746,
+      "step": 55000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.592692557974444e-05,
+      "loss": 1.7433,
+      "step": 55100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5919530880265026e-05,
+      "loss": 1.76,
+      "step": 55200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.591213618078561e-05,
+      "loss": 1.7278,
+      "step": 55300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.590481542830099e-05,
+      "loss": 1.7517,
+      "step": 55400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5897420728821585e-05,
+      "loss": 1.7349,
+      "step": 55500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.589002602934217e-05,
+      "loss": 1.7422,
+      "step": 55600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.5882631329862756e-05,
+      "loss": 1.7348,
+      "step": 55700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.587523663038334e-05,
+      "loss": 1.7244,
+      "step": 55800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5867841930903934e-05,
+      "loss": 1.7639,
+      "step": 55900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.586044723142452e-05,
+      "loss": 1.7663,
+      "step": 56000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5853052531945105e-05,
+      "loss": 1.7351,
+      "step": 56100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.584565783246569e-05,
+      "loss": 1.719,
+      "step": 56200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.583826313298628e-05,
+      "loss": 1.7408,
+      "step": 56300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.583086843350686e-05,
+      "loss": 1.7307,
+      "step": 56400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.582347373402745e-05,
+      "loss": 1.7479,
+      "step": 56500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.581607903454804e-05,
+      "loss": 1.7718,
+      "step": 56600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5808684335068626e-05,
+      "loss": 1.7378,
+      "step": 56700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.580128963558921e-05,
+      "loss": 1.741,
+      "step": 56800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.57938949361098e-05,
+      "loss": 1.7423,
+      "step": 56900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.578650023663038e-05,
+      "loss": 1.7297,
+      "step": 57000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5779105537150975e-05,
+      "loss": 1.7669,
+      "step": 57100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.577171083767156e-05,
+      "loss": 1.7625,
+      "step": 57200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.576431613819215e-05,
+      "loss": 1.745,
+      "step": 57300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.575692143871273e-05,
+      "loss": 1.7359,
+      "step": 57400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.574952673923332e-05,
+      "loss": 1.724,
+      "step": 57500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5742132039753904e-05,
+      "loss": 1.698,
+      "step": 57600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.573473734027449e-05,
+      "loss": 1.7201,
+      "step": 57700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.572734264079508e-05,
+      "loss": 1.75,
+      "step": 57800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.571994794131567e-05,
+      "loss": 1.7518,
+      "step": 57900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.571255324183625e-05,
+      "loss": 1.7491,
+      "step": 58000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.570515854235684e-05,
+      "loss": 1.7018,
+      "step": 58100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.569776384287743e-05,
+      "loss": 1.7489,
+      "step": 58200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5690369143398017e-05,
+      "loss": 1.7427,
+      "step": 58300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.56830483909134e-05,
+      "loss": 1.748,
+      "step": 58400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567565369143398e-05,
+      "loss": 1.7254,
+      "step": 58500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566825899195457e-05,
+      "loss": 1.748,
+      "step": 58600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.566086429247516e-05,
+      "loss": 1.7413,
+      "step": 58700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.565346959299574e-05,
+      "loss": 1.7595,
+      "step": 58800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5646074893516325e-05,
+      "loss": 1.7217,
+      "step": 58900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.563868019403692e-05,
+      "loss": 1.7287,
+      "step": 59000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56312854945575e-05,
+      "loss": 1.7052,
+      "step": 59100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.562389079507809e-05,
+      "loss": 1.721,
+      "step": 59200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5616496095598675e-05,
+      "loss": 1.7357,
+      "step": 59300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.560910139611927e-05,
+      "loss": 1.74,
+      "step": 59400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.560170669663985e-05,
+      "loss": 1.7229,
+      "step": 59500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.559431199716044e-05,
+      "loss": 1.7402,
+      "step": 59600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5586917297681024e-05,
+      "loss": 1.7735,
+      "step": 59700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.557952259820161e-05,
+      "loss": 1.7196,
+      "step": 59800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.55721278987222e-05,
+      "loss": 1.7319,
+      "step": 59900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.556473319924279e-05,
+      "loss": 1.7132,
+      "step": 60000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.555733849976337e-05,
+      "loss": 1.737,
+      "step": 60100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.554994380028396e-05,
+      "loss": 1.7046,
+      "step": 60200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5542549100804544e-05,
+      "loss": 1.7156,
+      "step": 60300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.553515440132513e-05,
+      "loss": 1.7484,
+      "step": 60400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5527759701845716e-05,
+      "loss": 1.6991,
+      "step": 60500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.552036500236631e-05,
+      "loss": 1.7218,
+      "step": 60600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5512970302886894e-05,
+      "loss": 1.7342,
+      "step": 60700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5505649550402274e-05,
+      "loss": 1.7479,
+      "step": 60800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.549825485092286e-05,
+      "loss": 1.7051,
+      "step": 60900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5490860151443446e-05,
+      "loss": 1.7051,
+      "step": 61000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.548346545196404e-05,
+      "loss": 1.7394,
+      "step": 61100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5476070752484624e-05,
+      "loss": 1.7232,
+      "step": 61200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.546867605300521e-05,
+      "loss": 1.7034,
+      "step": 61300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5461281353525795e-05,
+      "loss": 1.7138,
+      "step": 61400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545388665404638e-05,
+      "loss": 1.728,
+      "step": 61500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5446491954566966e-05,
+      "loss": 1.7295,
+      "step": 61600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.543909725508755e-05,
+      "loss": 1.7085,
+      "step": 61700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5431702555608144e-05,
+      "loss": 1.7272,
+      "step": 61800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.542430785612873e-05,
+      "loss": 1.744,
+      "step": 61900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5416913156649315e-05,
+      "loss": 1.739,
+      "step": 62000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.54095184571699e-05,
+      "loss": 1.7242,
+      "step": 62100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.540212375769049e-05,
+      "loss": 1.7257,
+      "step": 62200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.539472905821108e-05,
+      "loss": 1.7315,
+      "step": 62300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5387334358731665e-05,
+      "loss": 1.7385,
+      "step": 62400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.537993965925225e-05,
+      "loss": 1.7208,
+      "step": 62500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5372544959772836e-05,
+      "loss": 1.7236,
+      "step": 62600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.536515026029343e-05,
+      "loss": 1.7295,
+      "step": 62700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.535775556081401e-05,
+      "loss": 1.6798,
+      "step": 62800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.535036086133459e-05,
+      "loss": 1.7287,
+      "step": 62900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5343040108849973e-05,
+      "loss": 1.7127,
+      "step": 63000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5335645409370566e-05,
+      "loss": 1.7257,
+      "step": 63100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.532825070989115e-05,
+      "loss": 1.7089,
+      "step": 63200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.532085601041174e-05,
+      "loss": 1.7232,
+      "step": 63300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.531346131093232e-05,
+      "loss": 1.6997,
+      "step": 63400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5306066611452915e-05,
+      "loss": 1.7407,
+      "step": 63500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.52986719119735e-05,
+      "loss": 1.7252,
+      "step": 63600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5291277212494086e-05,
+      "loss": 1.7325,
+      "step": 63700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.528388251301467e-05,
+      "loss": 1.703,
+      "step": 63800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5276487813535265e-05,
+      "loss": 1.7222,
+      "step": 63900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.526909311405585e-05,
+      "loss": 1.7445,
+      "step": 64000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.526169841457643e-05,
+      "loss": 1.7151,
+      "step": 64100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.525430371509702e-05,
+      "loss": 1.7084,
+      "step": 64200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.524690901561761e-05,
+      "loss": 1.7056,
+      "step": 64300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.523951431613819e-05,
+      "loss": 1.7038,
+      "step": 64400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.523219356365357e-05,
+      "loss": 1.7043,
+      "step": 64500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.522479886417416e-05,
+      "loss": 1.7337,
+      "step": 64600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.521740416469475e-05,
+      "loss": 1.7563,
+      "step": 64700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.521000946521534e-05,
+      "loss": 1.7321,
+      "step": 64800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.520261476573592e-05,
+      "loss": 1.6985,
+      "step": 64900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.519522006625651e-05,
+      "loss": 1.7249,
+      "step": 65000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.51878253667771e-05,
+      "loss": 1.7415,
+      "step": 65100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5180430667297686e-05,
+      "loss": 1.7054,
+      "step": 65200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517303596781827e-05,
+      "loss": 1.6725,
+      "step": 65300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.516564126833886e-05,
+      "loss": 1.6966,
+      "step": 65400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.515824656885944e-05,
+      "loss": 1.7048,
+      "step": 65500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.515085186938003e-05,
+      "loss": 1.7154,
+      "step": 65600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5143457169900614e-05,
+      "loss": 1.6992,
+      "step": 65700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.51360624704212e-05,
+      "loss": 1.7224,
+      "step": 65800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.512866777094179e-05,
+      "loss": 1.6985,
+      "step": 65900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.512127307146238e-05,
+      "loss": 1.7052,
+      "step": 66000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5113878371982964e-05,
+      "loss": 1.7244,
+      "step": 66100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.510648367250355e-05,
+      "loss": 1.7161,
+      "step": 66200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.509908897302414e-05,
+      "loss": 1.7339,
+      "step": 66300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.509169427354473e-05,
+      "loss": 1.7057,
+      "step": 66400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.508429957406531e-05,
+      "loss": 1.7268,
+      "step": 66500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.50769048745859e-05,
+      "loss": 1.6977,
+      "step": 66600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506951017510649e-05,
+      "loss": 1.7667,
+      "step": 66700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.506211547562707e-05,
+      "loss": 1.7237,
+      "step": 66800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5054720776147656e-05,
+      "loss": 1.6985,
+      "step": 66900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.504732607666825e-05,
+      "loss": 1.7174,
+      "step": 67000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5039931377188834e-05,
+      "loss": 1.7205,
+      "step": 67100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.503253667770942e-05,
+      "loss": 1.7076,
+      "step": 67200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5025141978230005e-05,
+      "loss": 1.6836,
+      "step": 67300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.501774727875059e-05,
+      "loss": 1.7388,
+      "step": 67400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.501035257927118e-05,
+      "loss": 1.7164,
+      "step": 67500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.500295787979177e-05,
+      "loss": 1.7383,
+      "step": 67600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4995563180312354e-05,
+      "loss": 1.669,
+      "step": 67700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.498816848083294e-05,
+      "loss": 1.6484,
+      "step": 67800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.498077378135353e-05,
+      "loss": 1.6639,
+      "step": 67900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.497337908187412e-05,
+      "loss": 1.6289,
+      "step": 68000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.49659843823947e-05,
+      "loss": 1.6387,
+      "step": 68100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.495858968291529e-05,
+      "loss": 1.6487,
+      "step": 68200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4951194983435875e-05,
+      "loss": 1.6556,
+      "step": 68300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.494380028395646e-05,
+      "loss": 1.6545,
+      "step": 68400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4936405584477046e-05,
+      "loss": 1.6228,
+      "step": 68500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.492901088499764e-05,
+      "loss": 1.6589,
+      "step": 68600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4921616185518224e-05,
+      "loss": 1.6365,
+      "step": 68700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.491422148603881e-05,
+      "loss": 1.6372,
+      "step": 68800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4906826786559395e-05,
+      "loss": 1.654,
+      "step": 68900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.489943208707999e-05,
+      "loss": 1.6128,
+      "step": 69000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4892037387600573e-05,
+      "loss": 1.6337,
+      "step": 69100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4884716635115954e-05,
+      "loss": 1.6445,
+      "step": 69200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.487732193563654e-05,
+      "loss": 1.6732,
+      "step": 69300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4869927236157125e-05,
+      "loss": 1.6443,
+      "step": 69400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.486253253667771e-05,
+      "loss": 1.6231,
+      "step": 69500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4855137837198296e-05,
+      "loss": 1.6425,
+      "step": 69600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.484774313771888e-05,
+      "loss": 1.6342,
+      "step": 69700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.4840348438239475e-05,
+      "loss": 1.6311,
+      "step": 69800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.483295373876006e-05,
+      "loss": 1.6294,
+      "step": 69900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4825559039280646e-05,
+      "loss": 1.6494,
+      "step": 70000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481816433980123e-05,
+      "loss": 1.6323,
+      "step": 70100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.481076964032182e-05,
+      "loss": 1.6091,
+      "step": 70200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.480337494084241e-05,
+      "loss": 1.6803,
+      "step": 70300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4795980241362995e-05,
+      "loss": 1.6503,
+      "step": 70400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.478858554188358e-05,
+      "loss": 1.6331,
+      "step": 70500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.4781190842404166e-05,
+      "loss": 1.6153,
+      "step": 70600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.477379614292475e-05,
+      "loss": 1.634,
+      "step": 70700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.476640144344534e-05,
+      "loss": 1.6496,
+      "step": 70800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.475900674396592e-05,
+      "loss": 1.6467,
+      "step": 70900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4751612044486516e-05,
+      "loss": 1.646,
+      "step": 71000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.47442173450071e-05,
+      "loss": 1.6283,
+      "step": 71100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.473682264552769e-05,
+      "loss": 1.6491,
+      "step": 71200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.472942794604827e-05,
+      "loss": 1.6546,
+      "step": 71300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4722033246568865e-05,
+      "loss": 1.658,
+      "step": 71400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.471463854708945e-05,
+      "loss": 1.6407,
+      "step": 71500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4707243847610036e-05,
+      "loss": 1.6168,
+      "step": 71600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.469984914813062e-05,
+      "loss": 1.6646,
+      "step": 71700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4692528395646e-05,
+      "loss": 1.6625,
+      "step": 71800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4685133696166595e-05,
+      "loss": 1.6334,
+      "step": 71900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.467773899668718e-05,
+      "loss": 1.6648,
+      "step": 72000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.467034429720776e-05,
+      "loss": 1.6427,
+      "step": 72100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.466294959772835e-05,
+      "loss": 1.6329,
+      "step": 72200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.465555489824894e-05,
+      "loss": 1.6358,
+      "step": 72300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.464816019876952e-05,
+      "loss": 1.6472,
+      "step": 72400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.464076549929011e-05,
+      "loss": 1.6301,
+      "step": 72500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.4633370799810694e-05,
+      "loss": 1.6448,
+      "step": 72600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.462597610033129e-05,
+      "loss": 1.6268,
+      "step": 72700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461858140085187e-05,
+      "loss": 1.6422,
+      "step": 72800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461118670137246e-05,
+      "loss": 1.6508,
+      "step": 72900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4603792001893044e-05,
+      "loss": 1.6502,
+      "step": 73000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.4596397302413636e-05,
+      "loss": 1.6549,
+      "step": 73100
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.458900260293422e-05,
+      "loss": 1.6403,
+      "step": 73200
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.458160790345481e-05,
+      "loss": 1.6211,
+      "step": 73300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.457421320397539e-05,
+      "loss": 1.6717,
+      "step": 73400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456681850449598e-05,
+      "loss": 1.623,
+      "step": 73500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4559423805016564e-05,
+      "loss": 1.636,
+      "step": 73600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.455202910553715e-05,
+      "loss": 1.653,
+      "step": 73700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.454463440605774e-05,
+      "loss": 1.6364,
+      "step": 73800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.453723970657833e-05,
+      "loss": 1.6604,
+      "step": 73900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.4529845007098914e-05,
+      "loss": 1.6192,
+      "step": 74000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.45224503076195e-05,
+      "loss": 1.6533,
+      "step": 74100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.451505560814009e-05,
+      "loss": 1.658,
+      "step": 74200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.450766090866068e-05,
+      "loss": 1.6628,
+      "step": 74300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.450026620918126e-05,
+      "loss": 1.6612,
+      "step": 74400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.449294545669664e-05,
+      "loss": 1.6311,
+      "step": 74500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.448555075721723e-05,
+      "loss": 1.6191,
+      "step": 74600
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.4478156057737815e-05,
+      "loss": 1.6699,
+      "step": 74700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.44707613582584e-05,
+      "loss": 1.6596,
+      "step": 74800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4463366658778986e-05,
+      "loss": 1.6416,
+      "step": 74900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.445597195929958e-05,
+      "loss": 1.658,
+      "step": 75000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4448577259820164e-05,
+      "loss": 1.6409,
+      "step": 75100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.444118256034075e-05,
+      "loss": 1.6429,
+      "step": 75200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4433787860861335e-05,
+      "loss": 1.62,
+      "step": 75300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.442639316138192e-05,
+      "loss": 1.6508,
+      "step": 75400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.441899846190251e-05,
+      "loss": 1.6689,
+      "step": 75500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.44116037624231e-05,
+      "loss": 1.6373,
+      "step": 75600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.4404209062943685e-05,
+      "loss": 1.6261,
+      "step": 75700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.439681436346427e-05,
+      "loss": 1.6451,
+      "step": 75800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.438941966398486e-05,
+      "loss": 1.6637,
+      "step": 75900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.438202496450544e-05,
+      "loss": 1.6368,
+      "step": 76000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.437463026502603e-05,
+      "loss": 1.6531,
+      "step": 76100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.436723556554662e-05,
+      "loss": 1.6539,
+      "step": 76200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4359840866067205e-05,
+      "loss": 1.6337,
+      "step": 76300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.435244616658779e-05,
+      "loss": 1.6454,
+      "step": 76400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.434512541410317e-05,
+      "loss": 1.6357,
+      "step": 76500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.433773071462376e-05,
+      "loss": 1.6724,
+      "step": 76600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.433033601514435e-05,
+      "loss": 1.6366,
+      "step": 76700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4322941315664935e-05,
+      "loss": 1.6473,
+      "step": 76800
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.431554661618552e-05,
+      "loss": 1.6549,
+      "step": 76900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4308151916706106e-05,
+      "loss": 1.6261,
+      "step": 77000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.43007572172267e-05,
+      "loss": 1.6345,
+      "step": 77100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4293362517747284e-05,
+      "loss": 1.6534,
+      "step": 77200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.428596781826786e-05,
+      "loss": 1.6261,
+      "step": 77300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4278573118788456e-05,
+      "loss": 1.6517,
+      "step": 77400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.427117841930904e-05,
+      "loss": 1.6511,
+      "step": 77500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.426378371982963e-05,
+      "loss": 1.6348,
+      "step": 77600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.425638902035021e-05,
+      "loss": 1.6357,
+      "step": 77700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.42489943208708e-05,
+      "loss": 1.6104,
+      "step": 77800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.424159962139139e-05,
+      "loss": 1.662,
+      "step": 77900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4234204921911976e-05,
+      "loss": 1.6615,
+      "step": 78000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.422681022243256e-05,
+      "loss": 1.6471,
+      "step": 78100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.421941552295315e-05,
+      "loss": 1.6302,
+      "step": 78200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.421202082347374e-05,
+      "loss": 1.6192,
+      "step": 78300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4204626123994325e-05,
+      "loss": 1.6047,
+      "step": 78400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.419723142451491e-05,
+      "loss": 1.6212,
+      "step": 78500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.418991067203029e-05,
+      "loss": 1.606,
+      "step": 78600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.418251597255088e-05,
+      "loss": 1.6442,
+      "step": 78700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.417512127307146e-05,
+      "loss": 1.6533,
+      "step": 78800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.416772657359205e-05,
+      "loss": 1.655,
+      "step": 78900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4160331874112634e-05,
+      "loss": 1.6277,
+      "step": 79000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4152937174633227e-05,
+      "loss": 1.6666,
+      "step": 79100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.414554247515381e-05,
+      "loss": 1.6253,
+      "step": 79200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.41381477756744e-05,
+      "loss": 1.6345,
+      "step": 79300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4130753076194983e-05,
+      "loss": 1.6389,
+      "step": 79400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4123358376715576e-05,
+      "loss": 1.6313,
+      "step": 79500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.411596367723616e-05,
+      "loss": 1.6361,
+      "step": 79600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.410856897775675e-05,
+      "loss": 1.6882,
+      "step": 79700
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.410117427827733e-05,
+      "loss": 1.6232,
+      "step": 79800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4093779578797925e-05,
+      "loss": 1.6481,
+      "step": 79900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4086384879318504e-05,
+      "loss": 1.6444,
+      "step": 80000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.407899017983909e-05,
+      "loss": 1.6399,
+      "step": 80100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4071595480359675e-05,
+      "loss": 1.6394,
+      "step": 80200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.406420078088027e-05,
+      "loss": 1.6524,
+      "step": 80300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.405680608140085e-05,
+      "loss": 1.6401,
+      "step": 80400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.404941138192144e-05,
+      "loss": 1.6188,
+      "step": 80500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.404209062943682e-05,
+      "loss": 1.6411,
+      "step": 80600
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.403469592995741e-05,
+      "loss": 1.6328,
+      "step": 80700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4027301230478e-05,
+      "loss": 1.595,
+      "step": 80800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.401990653099858e-05,
+      "loss": 1.6619,
+      "step": 80900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.401251183151917e-05,
+      "loss": 1.61,
+      "step": 81000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.400511713203976e-05,
+      "loss": 1.6605,
+      "step": 81100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.399772243256035e-05,
+      "loss": 1.6238,
+      "step": 81200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.3990327733080926e-05,
+      "loss": 1.6199,
+      "step": 81300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.398293303360151e-05,
+      "loss": 1.6419,
+      "step": 81400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3975538334122104e-05,
+      "loss": 1.632,
+      "step": 81500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.396814363464269e-05,
+      "loss": 1.62,
+      "step": 81600
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3960748935163275e-05,
+      "loss": 1.6128,
+      "step": 81700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.395335423568386e-05,
+      "loss": 1.6094,
+      "step": 81800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.394595953620445e-05,
+      "loss": 1.6465,
+      "step": 81900
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.393856483672504e-05,
+      "loss": 1.6343,
+      "step": 82000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3931170137245624e-05,
+      "loss": 1.6188,
+      "step": 82100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.392377543776621e-05,
+      "loss": 1.6546,
+      "step": 82200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.39163807382868e-05,
+      "loss": 1.6277,
+      "step": 82300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.390898603880739e-05,
+      "loss": 1.635,
+      "step": 82400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3901591339327974e-05,
+      "loss": 1.6634,
+      "step": 82500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.389419663984856e-05,
+      "loss": 1.6109,
+      "step": 82600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.3886801940369145e-05,
+      "loss": 1.6295,
+      "step": 82700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.387940724088973e-05,
+      "loss": 1.6304,
+      "step": 82800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.387208648840511e-05,
+      "loss": 1.6545,
+      "step": 82900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.38646917889257e-05,
+      "loss": 1.6472,
+      "step": 83000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.385729708944629e-05,
+      "loss": 1.6056,
+      "step": 83100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3849902389966875e-05,
+      "loss": 1.6097,
+      "step": 83200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.384250769048746e-05,
+      "loss": 1.6203,
+      "step": 83300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3835112991008046e-05,
+      "loss": 1.6279,
+      "step": 83400
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.382771829152864e-05,
+      "loss": 1.634,
+      "step": 83500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3820323592049224e-05,
+      "loss": 1.6468,
+      "step": 83600
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.381292889256981e-05,
+      "loss": 1.632,
+      "step": 83700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3805534193090395e-05,
+      "loss": 1.6681,
+      "step": 83800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.379813949361098e-05,
+      "loss": 1.647,
+      "step": 83900
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3790744794131567e-05,
+      "loss": 1.6719,
+      "step": 84000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.378335009465215e-05,
+      "loss": 1.6415,
+      "step": 84100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.377595539517274e-05,
+      "loss": 1.6047,
+      "step": 84200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.376856069569333e-05,
+      "loss": 1.6338,
+      "step": 84300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3761165996213916e-05,
+      "loss": 1.6209,
+      "step": 84400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.37537712967345e-05,
+      "loss": 1.6684,
+      "step": 84500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.374637659725509e-05,
+      "loss": 1.6398,
+      "step": 84600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.373898189777568e-05,
+      "loss": 1.636,
+      "step": 84700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3731587198296265e-05,
+      "loss": 1.6495,
+      "step": 84800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.372419249881685e-05,
+      "loss": 1.6281,
+      "step": 84900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3716797799337437e-05,
+      "loss": 1.598,
+      "step": 85000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.370940309985803e-05,
+      "loss": 1.635,
+      "step": 85100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3702008400378615e-05,
+      "loss": 1.6374,
+      "step": 85200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3694613700899193e-05,
+      "loss": 1.6208,
+      "step": 85300
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3687292948414574e-05,
+      "loss": 1.6426,
+      "step": 85400
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3679898248935166e-05,
+      "loss": 1.6094,
+      "step": 85500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367250354945575e-05,
+      "loss": 1.6125,
+      "step": 85600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366510884997634e-05,
+      "loss": 1.6425,
+      "step": 85700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365771415049692e-05,
+      "loss": 1.6346,
+      "step": 85800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3650319451017516e-05,
+      "loss": 1.6576,
+      "step": 85900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.36429247515381e-05,
+      "loss": 1.6152,
+      "step": 86000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.363553005205869e-05,
+      "loss": 1.6333,
+      "step": 86100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.362813535257927e-05,
+      "loss": 1.6195,
+      "step": 86200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3620740653099865e-05,
+      "loss": 1.6354,
+      "step": 86300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361334595362045e-05,
+      "loss": 1.6393,
+      "step": 86400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3605951254141036e-05,
+      "loss": 1.6292,
+      "step": 86500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3598556554661615e-05,
+      "loss": 1.633,
+      "step": 86600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.359116185518221e-05,
+      "loss": 1.6108,
+      "step": 86700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358376715570279e-05,
+      "loss": 1.6389,
+      "step": 86800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.357637245622338e-05,
+      "loss": 1.6415,
+      "step": 86900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3568977756743964e-05,
+      "loss": 1.6349,
+      "step": 87000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.356158305726456e-05,
+      "loss": 1.6338,
+      "step": 87100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.355418835778514e-05,
+      "loss": 1.6118,
+      "step": 87200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.354679365830573e-05,
+      "loss": 1.6309,
+      "step": 87300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3539398958826314e-05,
+      "loss": 1.6465,
+      "step": 87400
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3532004259346906e-05,
+      "loss": 1.6596,
+      "step": 87500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.352460955986749e-05,
+      "loss": 1.6285,
+      "step": 87600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.351721486038808e-05,
+      "loss": 1.6171,
+      "step": 87700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.350982016090866e-05,
+      "loss": 1.6266,
+      "step": 87800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.350242546142925e-05,
+      "loss": 1.6647,
+      "step": 87900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3495030761949834e-05,
+      "loss": 1.6415,
+      "step": 88000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.348763606247042e-05,
+      "loss": 1.6301,
+      "step": 88100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3480241362991006e-05,
+      "loss": 1.6248,
+      "step": 88200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.34728466635116e-05,
+      "loss": 1.6763,
+      "step": 88300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3465451964032184e-05,
+      "loss": 1.6353,
+      "step": 88400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3458131211547564e-05,
+      "loss": 1.6283,
+      "step": 88500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.345073651206815e-05,
+      "loss": 1.6197,
+      "step": 88600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.344334181258874e-05,
+      "loss": 1.6424,
+      "step": 88700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.343594711310933e-05,
+      "loss": 1.6498,
+      "step": 88800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3428552413629913e-05,
+      "loss": 1.6237,
+      "step": 88900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.34211577141505e-05,
+      "loss": 1.6559,
+      "step": 89000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341376301467109e-05,
+      "loss": 1.6505,
+      "step": 89100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.340636831519167e-05,
+      "loss": 1.617,
+      "step": 89200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3398973615712256e-05,
+      "loss": 1.6224,
+      "step": 89300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.339157891623284e-05,
+      "loss": 1.6106,
+      "step": 89400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3384184216753434e-05,
+      "loss": 1.6361,
+      "step": 89500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.337678951727402e-05,
+      "loss": 1.6356,
+      "step": 89600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3369394817794605e-05,
+      "loss": 1.6164,
+      "step": 89700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.336200011831519e-05,
+      "loss": 1.633,
+      "step": 89800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.335460541883578e-05,
+      "loss": 1.6322,
+      "step": 89900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.334721071935637e-05,
+      "loss": 1.6373,
+      "step": 90000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3339816019876955e-05,
+      "loss": 1.6194,
+      "step": 90100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.333242132039754e-05,
+      "loss": 1.659,
+      "step": 90200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.332502662091813e-05,
+      "loss": 1.6306,
+      "step": 90300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.331763192143872e-05,
+      "loss": 1.6437,
+      "step": 90400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.33102372219593e-05,
+      "loss": 1.6229,
+      "step": 90500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.330284252247988e-05,
+      "loss": 1.6333,
+      "step": 90600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3295447823000475e-05,
+      "loss": 1.6032,
+      "step": 90700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.328805312352106e-05,
+      "loss": 1.6351,
+      "step": 90800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.3280658424041647e-05,
+      "loss": 1.6236,
+      "step": 90900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.327326372456223e-05,
+      "loss": 1.6266,
+      "step": 91000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.326594297207762e-05,
+      "loss": 1.6116,
+      "step": 91100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3258548272598205e-05,
+      "loss": 1.6466,
+      "step": 91200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.325115357311879e-05,
+      "loss": 1.6218,
+      "step": 91300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3243758873639376e-05,
+      "loss": 1.6277,
+      "step": 91400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.323636417415997e-05,
+      "loss": 1.6232,
+      "step": 91500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.3228969474680554e-05,
+      "loss": 1.6173,
+      "step": 91600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.322157477520114e-05,
+      "loss": 1.6286,
+      "step": 91700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3214180075721726e-05,
+      "loss": 1.6258,
+      "step": 91800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.320678537624231e-05,
+      "loss": 1.591,
+      "step": 91900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.31993906767629e-05,
+      "loss": 1.5922,
+      "step": 92000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.319199597728348e-05,
+      "loss": 1.6444,
+      "step": 92100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.318460127780407e-05,
+      "loss": 1.6526,
+      "step": 92200
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.317720657832466e-05,
+      "loss": 1.6265,
+      "step": 92300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3169811878845246e-05,
+      "loss": 1.6333,
+      "step": 92400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.316241717936583e-05,
+      "loss": 1.6407,
+      "step": 92500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.315502247988642e-05,
+      "loss": 1.6282,
+      "step": 92600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.314762778040701e-05,
+      "loss": 1.6108,
+      "step": 92700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3140233080927596e-05,
+      "loss": 1.6281,
+      "step": 92800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.313283838144818e-05,
+      "loss": 1.639,
+      "step": 92900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.312544368196877e-05,
+      "loss": 1.6396,
+      "step": 93000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.311804898248936e-05,
+      "loss": 1.6426,
+      "step": 93100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.311065428300994e-05,
+      "loss": 1.6028,
+      "step": 93200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.310333353052532e-05,
+      "loss": 1.6297,
+      "step": 93300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3095938831045904e-05,
+      "loss": 1.6002,
+      "step": 93400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.30885441315665e-05,
+      "loss": 1.6076,
+      "step": 93500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.308114943208708e-05,
+      "loss": 1.6142,
+      "step": 93600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.307375473260767e-05,
+      "loss": 1.6157,
+      "step": 93700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3066360033128254e-05,
+      "loss": 1.6288,
+      "step": 93800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3058965333648846e-05,
+      "loss": 1.6287,
+      "step": 93900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.305157063416943e-05,
+      "loss": 1.6418,
+      "step": 94000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.304417593469002e-05,
+      "loss": 1.6308,
+      "step": 94100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.30367812352106e-05,
+      "loss": 1.6333,
+      "step": 94200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.3029386535731195e-05,
+      "loss": 1.6172,
+      "step": 94300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.302199183625178e-05,
+      "loss": 1.6401,
+      "step": 94400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.301459713677236e-05,
+      "loss": 1.639,
+      "step": 94500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3007202437292945e-05,
+      "loss": 1.6379,
+      "step": 94600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299980773781354e-05,
+      "loss": 1.648,
+      "step": 94700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2992413038334123e-05,
+      "loss": 1.6262,
+      "step": 94800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.298501833885471e-05,
+      "loss": 1.6528,
+      "step": 94900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2977623639375295e-05,
+      "loss": 1.5951,
+      "step": 95000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.297022893989589e-05,
+      "loss": 1.6108,
+      "step": 95100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.296283424041647e-05,
+      "loss": 1.6145,
+      "step": 95200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.295543954093706e-05,
+      "loss": 1.6635,
+      "step": 95300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2948044841457644e-05,
+      "loss": 1.6269,
+      "step": 95400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2940650141978236e-05,
+      "loss": 1.6338,
+      "step": 95500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.293325544249882e-05,
+      "loss": 1.6208,
+      "step": 95600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.292586074301941e-05,
+      "loss": 1.6224,
+      "step": 95700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.291853999053478e-05,
+      "loss": 1.6248,
+      "step": 95800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2911145291055374e-05,
+      "loss": 1.5974,
+      "step": 95900
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.290375059157596e-05,
+      "loss": 1.6579,
+      "step": 96000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2896355892096545e-05,
+      "loss": 1.6457,
+      "step": 96100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.288896119261713e-05,
+      "loss": 1.6079,
+      "step": 96200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.288156649313772e-05,
+      "loss": 1.668,
+      "step": 96300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.287417179365831e-05,
+      "loss": 1.6228,
+      "step": 96400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2866777094178894e-05,
+      "loss": 1.6389,
+      "step": 96500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.285938239469948e-05,
+      "loss": 1.6237,
+      "step": 96600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.285198769522007e-05,
+      "loss": 1.6382,
+      "step": 96700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.284459299574066e-05,
+      "loss": 1.6451,
+      "step": 96800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2837198296261244e-05,
+      "loss": 1.6141,
+      "step": 96900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.282980359678183e-05,
+      "loss": 1.6399,
+      "step": 97000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.282240889730242e-05,
+      "loss": 1.596,
+      "step": 97100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2815014197823e-05,
+      "loss": 1.6428,
+      "step": 97200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2807619498343586e-05,
+      "loss": 1.6249,
+      "step": 97300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.280022479886417e-05,
+      "loss": 1.6077,
+      "step": 97400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2792830099384764e-05,
+      "loss": 1.635,
+      "step": 97500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.278543539990535e-05,
+      "loss": 1.6368,
+      "step": 97600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.2778040700425936e-05,
+      "loss": 1.5985,
+      "step": 97700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2770719947941316e-05,
+      "loss": 1.6455,
+      "step": 97800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.27633252484619e-05,
+      "loss": 1.6137,
+      "step": 97900
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2755930548982494e-05,
+      "loss": 1.6158,
+      "step": 98000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.274853584950308e-05,
+      "loss": 1.6094,
+      "step": 98100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.2741141150023665e-05,
+      "loss": 1.6193,
+      "step": 98200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.273374645054425e-05,
+      "loss": 1.6304,
+      "step": 98300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2726351751064843e-05,
+      "loss": 1.6345,
+      "step": 98400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271895705158542e-05,
+      "loss": 1.6217,
+      "step": 98500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.271156235210601e-05,
+      "loss": 1.6285,
+      "step": 98600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.27041676526266e-05,
+      "loss": 1.5898,
+      "step": 98700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2696772953147186e-05,
+      "loss": 1.6426,
+      "step": 98800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.268937825366777e-05,
+      "loss": 1.5872,
+      "step": 98900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.268198355418836e-05,
+      "loss": 1.6302,
+      "step": 99000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.267458885470895e-05,
+      "loss": 1.6212,
+      "step": 99100
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2667194155229535e-05,
+      "loss": 1.6052,
+      "step": 99200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.265979945575012e-05,
+      "loss": 1.6149,
+      "step": 99300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.265240475627071e-05,
+      "loss": 1.6224,
+      "step": 99400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.26450100567913e-05,
+      "loss": 1.5973,
+      "step": 99500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.2637615357311885e-05,
+      "loss": 1.6388,
+      "step": 99600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.263022065783247e-05,
+      "loss": 1.6465,
+      "step": 99700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.262282595835305e-05,
+      "loss": 1.6219,
+      "step": 99800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2615505205868436e-05,
+      "loss": 1.6281,
+      "step": 99900
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.260811050638902e-05,
+      "loss": 1.6182,
+      "step": 100000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.260071580690961e-05,
+      "loss": 1.6147,
+      "step": 100100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.259332110743019e-05,
+      "loss": 1.5992,
+      "step": 100200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.2585926407950786e-05,
+      "loss": 1.6439,
+      "step": 100300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.257853170847137e-05,
+      "loss": 1.6332,
+      "step": 100400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.257113700899196e-05,
+      "loss": 1.6309,
+      "step": 100500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.256374230951254e-05,
+      "loss": 1.6421,
+      "step": 100600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.255634761003313e-05,
+      "loss": 1.6317,
+      "step": 100700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.254895291055372e-05,
+      "loss": 1.6046,
+      "step": 100800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.2541558211074306e-05,
+      "loss": 1.5878,
+      "step": 100900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.253416351159489e-05,
+      "loss": 1.6006,
+      "step": 101000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252676881211548e-05,
+      "loss": 1.6034,
+      "step": 101100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251937411263606e-05,
+      "loss": 1.6061,
+      "step": 101200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.251197941315665e-05,
+      "loss": 1.6392,
+      "step": 101300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2504584713677235e-05,
+      "loss": 1.64,
+      "step": 101400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.249719001419783e-05,
+      "loss": 1.6063,
+      "step": 101500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.248979531471841e-05,
+      "loss": 1.6245,
+      "step": 101600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.2482400615239e-05,
+      "loss": 1.6393,
+      "step": 101700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2475005915759584e-05,
+      "loss": 1.6135,
+      "step": 101800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2467611216280176e-05,
+      "loss": 1.5999,
+      "step": 101900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.246021651680076e-05,
+      "loss": 1.6203,
+      "step": 102000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.245282181732135e-05,
+      "loss": 1.6123,
+      "step": 102100
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.244542711784193e-05,
+      "loss": 1.6135,
+      "step": 102200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.243803241836252e-05,
+      "loss": 1.6087,
+      "step": 102300
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.2430637718883104e-05,
+      "loss": 1.5978,
+      "step": 102400
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.242324301940369e-05,
+      "loss": 1.6235,
+      "step": 102500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2415848319924276e-05,
+      "loss": 1.6231,
+      "step": 102600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.240845362044487e-05,
+      "loss": 1.6129,
+      "step": 102700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2401058920965454e-05,
+      "loss": 1.6083,
+      "step": 102800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2393738168480834e-05,
+      "loss": 1.6398,
+      "step": 102900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.238634346900142e-05,
+      "loss": 1.6559,
+      "step": 103000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.2378948769522006e-05,
+      "loss": 1.6594,
+      "step": 103100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.23715540700426e-05,
+      "loss": 1.6223,
+      "step": 103200
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2364159370563184e-05,
+      "loss": 1.6015,
+      "step": 103300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.235676467108377e-05,
+      "loss": 1.6017,
+      "step": 103400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2349369971604355e-05,
+      "loss": 1.591,
+      "step": 103500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.234197527212495e-05,
+      "loss": 1.6354,
+      "step": 103600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.233458057264553e-05,
+      "loss": 1.6536,
+      "step": 103700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.232718587316611e-05,
+      "loss": 1.6496,
+      "step": 103800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2319791173686704e-05,
+      "loss": 1.6224,
+      "step": 103900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231239647420729e-05,
+      "loss": 1.6042,
+      "step": 104000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2305001774727875e-05,
+      "loss": 1.6002,
+      "step": 104100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.229760707524846e-05,
+      "loss": 1.6119,
+      "step": 104200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2290212375769053e-05,
+      "loss": 1.6331,
+      "step": 104300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.228281767628964e-05,
+      "loss": 1.5944,
+      "step": 104400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2275422976810225e-05,
+      "loss": 1.6276,
+      "step": 104500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.226802827733081e-05,
+      "loss": 1.6114,
+      "step": 104600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.22606335778514e-05,
+      "loss": 1.6055,
+      "step": 104700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.225323887837199e-05,
+      "loss": 1.6407,
+      "step": 104800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2245844178892574e-05,
+      "loss": 1.6003,
+      "step": 104900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.223844947941316e-05,
+      "loss": 1.5993,
+      "step": 105000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2231054779933745e-05,
+      "loss": 1.6179,
+      "step": 105100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2223734027449126e-05,
+      "loss": 1.6285,
+      "step": 105200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.221633932796971e-05,
+      "loss": 1.6104,
+      "step": 105300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.22089446284903e-05,
+      "loss": 1.6088,
+      "step": 105400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.220154992901089e-05,
+      "loss": 1.5981,
+      "step": 105500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2194155229531475e-05,
+      "loss": 1.6151,
+      "step": 105600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.218676053005206e-05,
+      "loss": 1.6077,
+      "step": 105700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2179365830572646e-05,
+      "loss": 1.5926,
+      "step": 105800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.217197113109323e-05,
+      "loss": 1.625,
+      "step": 105900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2164576431613824e-05,
+      "loss": 1.6096,
+      "step": 106000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.215718173213441e-05,
+      "loss": 1.6453,
+      "step": 106100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2149787032654996e-05,
+      "loss": 1.6129,
+      "step": 106200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.214239233317558e-05,
+      "loss": 1.6151,
+      "step": 106300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.213499763369617e-05,
+      "loss": 1.6382,
+      "step": 106400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212760293421675e-05,
+      "loss": 1.6186,
+      "step": 106500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.212020823473734e-05,
+      "loss": 1.6173,
+      "step": 106600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.211281353525793e-05,
+      "loss": 1.6034,
+      "step": 106700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2105418835778516e-05,
+      "loss": 1.601,
+      "step": 106800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.20980241362991e-05,
+      "loss": 1.573,
+      "step": 106900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.209062943681969e-05,
+      "loss": 1.6017,
+      "step": 107000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.208323473734028e-05,
+      "loss": 1.6469,
+      "step": 107100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2075840037860866e-05,
+      "loss": 1.6156,
+      "step": 107200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206844533838145e-05,
+      "loss": 1.6119,
+      "step": 107300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.206105063890204e-05,
+      "loss": 1.5963,
+      "step": 107400
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.205365593942262e-05,
+      "loss": 1.601,
+      "step": 107500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2046261239943215e-05,
+      "loss": 1.6264,
+      "step": 107600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2038866540463794e-05,
+      "loss": 1.6199,
+      "step": 107700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.203147184098438e-05,
+      "loss": 1.5711,
+      "step": 107800
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.202407714150497e-05,
+      "loss": 1.5916,
+      "step": 107900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.201668244202556e-05,
+      "loss": 1.6458,
+      "step": 108000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.200928774254614e-05,
+      "loss": 1.6154,
+      "step": 108100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.200189304306673e-05,
+      "loss": 1.6247,
+      "step": 108200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.199449834358732e-05,
+      "loss": 1.6042,
+      "step": 108300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.198710364410791e-05,
+      "loss": 1.607,
+      "step": 108400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.197970894462849e-05,
+      "loss": 1.63,
+      "step": 108500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.197231424514908e-05,
+      "loss": 1.6201,
+      "step": 108600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.196491954566967e-05,
+      "loss": 1.6012,
+      "step": 108700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.195759879318505e-05,
+      "loss": 1.6356,
+      "step": 108800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.195020409370564e-05,
+      "loss": 1.6124,
+      "step": 108900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1942809394226216e-05,
+      "loss": 1.6356,
+      "step": 109000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.193541469474681e-05,
+      "loss": 1.6144,
+      "step": 109100
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.1928019995267394e-05,
+      "loss": 1.6298,
+      "step": 109200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.192062529578798e-05,
+      "loss": 1.6183,
+      "step": 109300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1913230596308565e-05,
+      "loss": 1.5844,
+      "step": 109400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.190583589682916e-05,
+      "loss": 1.6193,
+      "step": 109500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.189844119734974e-05,
+      "loss": 1.6049,
+      "step": 109600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.189104649787033e-05,
+      "loss": 1.6099,
+      "step": 109700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1883651798390914e-05,
+      "loss": 1.6278,
+      "step": 109800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1876257098911507e-05,
+      "loss": 1.6018,
+      "step": 109900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186886239943209e-05,
+      "loss": 1.6577,
+      "step": 110000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.186146769995268e-05,
+      "loss": 1.6181,
+      "step": 110100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1854073000473264e-05,
+      "loss": 1.5939,
+      "step": 110200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.184667830099385e-05,
+      "loss": 1.6358,
+      "step": 110300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.1839283601514435e-05,
+      "loss": 1.6107,
+      "step": 110400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.183188890203502e-05,
+      "loss": 1.6319,
+      "step": 110500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1824494202555606e-05,
+      "loss": 1.6506,
+      "step": 110600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.18170995030762e-05,
+      "loss": 1.6076,
+      "step": 110700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1809704803596784e-05,
+      "loss": 1.6188,
+      "step": 110800
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.180231010411737e-05,
+      "loss": 1.6039,
+      "step": 110900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1794915404637955e-05,
+      "loss": 1.6352,
+      "step": 111000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.178752070515855e-05,
+      "loss": 1.6277,
+      "step": 111100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1780126005679133e-05,
+      "loss": 1.6297,
+      "step": 111200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.177273130619972e-05,
+      "loss": 1.6275,
+      "step": 111300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1765336606720305e-05,
+      "loss": 1.5865,
+      "step": 111400
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.17579419072409e-05,
+      "loss": 1.6496,
+      "step": 111500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.175054720776148e-05,
+      "loss": 1.6375,
+      "step": 111600
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.174315250828206e-05,
+      "loss": 1.6219,
+      "step": 111700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.173575780880265e-05,
+      "loss": 1.6049,
+      "step": 111800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.172836310932324e-05,
+      "loss": 1.6053,
+      "step": 111900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1720968409843825e-05,
+      "loss": 1.5958,
+      "step": 112000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.171357371036441e-05,
+      "loss": 1.6279,
+      "step": 112100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.170625295787979e-05,
+      "loss": 1.5954,
+      "step": 112200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1698858258400384e-05,
+      "loss": 1.6297,
+      "step": 112300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.169146355892097e-05,
+      "loss": 1.5992,
+      "step": 112400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.1684068859441555e-05,
+      "loss": 1.6119,
+      "step": 112500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.167667415996214e-05,
+      "loss": 1.6081,
+      "step": 112600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1669279460482726e-05,
+      "loss": 1.5996,
+      "step": 112700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.166188476100332e-05,
+      "loss": 1.6175,
+      "step": 112800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1654490061523904e-05,
+      "loss": 1.5972,
+      "step": 112900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.164709536204448e-05,
+      "loss": 1.6118,
+      "step": 113000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.1639700662565076e-05,
+      "loss": 1.5981,
+      "step": 113100
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.163230596308566e-05,
+      "loss": 1.6277,
+      "step": 113200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.162491126360625e-05,
+      "loss": 1.6026,
+      "step": 113300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.161751656412683e-05,
+      "loss": 1.618,
+      "step": 113400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1610121864647425e-05,
+      "loss": 1.6391,
+      "step": 113500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.160272716516801e-05,
+      "loss": 1.6182,
+      "step": 113600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1595332465688596e-05,
+      "loss": 1.6186,
+      "step": 113700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.158793776620918e-05,
+      "loss": 1.6004,
+      "step": 113800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1580543066729774e-05,
+      "loss": 1.6235,
+      "step": 113900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.157314836725036e-05,
+      "loss": 1.6156,
+      "step": 114000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.1565753667770946e-05,
+      "loss": 1.6116,
+      "step": 114100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.155835896829153e-05,
+      "loss": 1.6172,
+      "step": 114200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.155103821580691e-05,
+      "loss": 1.6125,
+      "step": 114300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.15436435163275e-05,
+      "loss": 1.63,
+      "step": 114400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.153624881684808e-05,
+      "loss": 1.6347,
+      "step": 114500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.152885411736867e-05,
+      "loss": 1.5803,
+      "step": 114600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.152145941788926e-05,
+      "loss": 1.6091,
+      "step": 114700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.151406471840985e-05,
+      "loss": 1.5889,
+      "step": 114800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.150667001893043e-05,
+      "loss": 1.6024,
+      "step": 114900
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.149927531945102e-05,
+      "loss": 1.6139,
+      "step": 115000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.149188061997161e-05,
+      "loss": 1.6277,
+      "step": 115100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.1484485920492196e-05,
+      "loss": 1.6008,
+      "step": 115200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.147709122101278e-05,
+      "loss": 1.6125,
+      "step": 115300
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146969652153337e-05,
+      "loss": 1.6353,
+      "step": 115400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.146230182205395e-05,
+      "loss": 1.6403,
+      "step": 115500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.145490712257454e-05,
+      "loss": 1.605,
+      "step": 115600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1447512423095124e-05,
+      "loss": 1.6449,
+      "step": 115700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.144011772361571e-05,
+      "loss": 1.5805,
+      "step": 115800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.14327230241363e-05,
+      "loss": 1.603,
+      "step": 115900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.142532832465689e-05,
+      "loss": 1.5695,
+      "step": 116000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1417933625177474e-05,
+      "loss": 1.6149,
+      "step": 116100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.141053892569806e-05,
+      "loss": 1.6009,
+      "step": 116200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.140314422621865e-05,
+      "loss": 1.6232,
+      "step": 116300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.139582347373403e-05,
+      "loss": 1.6138,
+      "step": 116400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.138842877425462e-05,
+      "loss": 1.6044,
+      "step": 116500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.13810340747752e-05,
+      "loss": 1.6301,
+      "step": 116600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.137363937529579e-05,
+      "loss": 1.6139,
+      "step": 116700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.136624467581638e-05,
+      "loss": 1.6057,
+      "step": 116800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.135884997633697e-05,
+      "loss": 1.6101,
+      "step": 116900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1351455276857546e-05,
+      "loss": 1.6063,
+      "step": 117000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.134406057737814e-05,
+      "loss": 1.5845,
+      "step": 117100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1336665877898724e-05,
+      "loss": 1.6047,
+      "step": 117200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.132927117841931e-05,
+      "loss": 1.6,
+      "step": 117300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1321876478939895e-05,
+      "loss": 1.6186,
+      "step": 117400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.131448177946049e-05,
+      "loss": 1.591,
+      "step": 117500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.130708707998107e-05,
+      "loss": 1.5918,
+      "step": 117600
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.129969238050166e-05,
+      "loss": 1.6115,
+      "step": 117700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1292297681022245e-05,
+      "loss": 1.6066,
+      "step": 117800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.128490298154283e-05,
+      "loss": 1.6162,
+      "step": 117900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127750828206342e-05,
+      "loss": 1.5887,
+      "step": 118000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.127011358258401e-05,
+      "loss": 1.5795,
+      "step": 118100
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1262718883104594e-05,
+      "loss": 1.6066,
+      "step": 118200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.125532418362518e-05,
+      "loss": 1.6255,
+      "step": 118300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1247929484145765e-05,
+      "loss": 1.6086,
+      "step": 118400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1240608731661146e-05,
+      "loss": 1.6209,
+      "step": 118500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.123321403218173e-05,
+      "loss": 1.6299,
+      "step": 118600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.122581933270232e-05,
+      "loss": 1.6455,
+      "step": 118700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121842463322291e-05,
+      "loss": 1.5975,
+      "step": 118800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1211029933743495e-05,
+      "loss": 1.5845,
+      "step": 118900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.120363523426408e-05,
+      "loss": 1.598,
+      "step": 119000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1196240534784666e-05,
+      "loss": 1.6076,
+      "step": 119100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.118884583530526e-05,
+      "loss": 1.6015,
+      "step": 119200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1181451135825844e-05,
+      "loss": 1.6004,
+      "step": 119300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.117405643634643e-05,
+      "loss": 1.6358,
+      "step": 119400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1166661736867015e-05,
+      "loss": 1.6061,
+      "step": 119500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.11592670373876e-05,
+      "loss": 1.5997,
+      "step": 119600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.115187233790819e-05,
+      "loss": 1.5813,
+      "step": 119700
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.114447763842877e-05,
+      "loss": 1.5993,
+      "step": 119800
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1137082938949365e-05,
+      "loss": 1.5982,
+      "step": 119900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.112968823946995e-05,
+      "loss": 1.5928,
+      "step": 120000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1122293539990536e-05,
+      "loss": 1.6137,
+      "step": 120100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.111489884051112e-05,
+      "loss": 1.6005,
+      "step": 120200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1107504141031714e-05,
+      "loss": 1.5991,
+      "step": 120300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.11001094415523e-05,
+      "loss": 1.5959,
+      "step": 120400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1092714742072885e-05,
+      "loss": 1.6012,
+      "step": 120500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1085393989588266e-05,
+      "loss": 1.5821,
+      "step": 120600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.107799929010885e-05,
+      "loss": 1.6066,
+      "step": 120700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1070604590629444e-05,
+      "loss": 1.6028,
+      "step": 120800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.106320989115002e-05,
+      "loss": 1.5734,
+      "step": 120900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.105581519167061e-05,
+      "loss": 1.5935,
+      "step": 121000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1048420492191194e-05,
+      "loss": 1.6123,
+      "step": 121100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1041025792711786e-05,
+      "loss": 1.5882,
+      "step": 121200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.103363109323237e-05,
+      "loss": 1.583,
+      "step": 121300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.102623639375296e-05,
+      "loss": 1.6131,
+      "step": 121400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1018841694273543e-05,
+      "loss": 1.5695,
+      "step": 121500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1011446994794136e-05,
+      "loss": 1.6164,
+      "step": 121600
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.100405229531472e-05,
+      "loss": 1.6036,
+      "step": 121700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.099665759583531e-05,
+      "loss": 1.6041,
+      "step": 121800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.098926289635589e-05,
+      "loss": 1.5975,
+      "step": 121900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0981868196876485e-05,
+      "loss": 1.6005,
+      "step": 122000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.097447349739707e-05,
+      "loss": 1.6059,
+      "step": 122100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0967078797917656e-05,
+      "loss": 1.5917,
+      "step": 122200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.095968409843824e-05,
+      "loss": 1.5738,
+      "step": 122300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.095228939895883e-05,
+      "loss": 1.5641,
+      "step": 122400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.094489469947941e-05,
+      "loss": 1.5712,
+      "step": 122500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09375e-05,
+      "loss": 1.5988,
+      "step": 122600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.093010530052059e-05,
+      "loss": 1.5891,
+      "step": 122700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.092278454803597e-05,
+      "loss": 1.6058,
+      "step": 122800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.091538984855656e-05,
+      "loss": 1.6177,
+      "step": 122900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.090799514907714e-05,
+      "loss": 1.6241,
+      "step": 123000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.090060044959773e-05,
+      "loss": 1.5881,
+      "step": 123100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.089320575011832e-05,
+      "loss": 1.5933,
+      "step": 123200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.088581105063891e-05,
+      "loss": 1.5864,
+      "step": 123300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.087841635115949e-05,
+      "loss": 1.6276,
+      "step": 123400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.087102165168008e-05,
+      "loss": 1.6104,
+      "step": 123500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0863626952200664e-05,
+      "loss": 1.6178,
+      "step": 123600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.085623225272125e-05,
+      "loss": 1.6164,
+      "step": 123700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.0848837553241835e-05,
+      "loss": 1.6137,
+      "step": 123800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.084144285376242e-05,
+      "loss": 1.6,
+      "step": 123900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.083404815428301e-05,
+      "loss": 1.6108,
+      "step": 124000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.08266534548036e-05,
+      "loss": 1.62,
+      "step": 124100
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0819258755324184e-05,
+      "loss": 1.5989,
+      "step": 124200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.081186405584477e-05,
+      "loss": 1.5963,
+      "step": 124300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.080446935636536e-05,
+      "loss": 1.5965,
+      "step": 124400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.079707465688595e-05,
+      "loss": 1.6012,
+      "step": 124500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0789679957406534e-05,
+      "loss": 1.6007,
+      "step": 124600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.078228525792712e-05,
+      "loss": 1.5601,
+      "step": 124700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0774964505442507e-05,
+      "loss": 1.6455,
+      "step": 124800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0767569805963085e-05,
+      "loss": 1.5978,
+      "step": 124900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.076017510648367e-05,
+      "loss": 1.6046,
+      "step": 125000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.075278040700426e-05,
+      "loss": 1.5961,
+      "step": 125100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.074538570752485e-05,
+      "loss": 1.6216,
+      "step": 125200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0737991008045435e-05,
+      "loss": 1.615,
+      "step": 125300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.073059630856602e-05,
+      "loss": 1.5908,
+      "step": 125400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0723201609086606e-05,
+      "loss": 1.5896,
+      "step": 125500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.07158069096072e-05,
+      "loss": 1.6085,
+      "step": 125600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0708412210127784e-05,
+      "loss": 1.5954,
+      "step": 125700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.070101751064837e-05,
+      "loss": 1.5981,
+      "step": 125800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0693622811168955e-05,
+      "loss": 1.599,
+      "step": 125900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.068622811168955e-05,
+      "loss": 1.5742,
+      "step": 126000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.067883341221013e-05,
+      "loss": 1.5918,
+      "step": 126100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.067143871273071e-05,
+      "loss": 1.6162,
+      "step": 126200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06640440132513e-05,
+      "loss": 1.5793,
+      "step": 126300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.065664931377189e-05,
+      "loss": 1.5984,
+      "step": 126400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.0649254614292476e-05,
+      "loss": 1.5852,
+      "step": 126500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.064185991481306e-05,
+      "loss": 1.5996,
+      "step": 126600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.063446521533365e-05,
+      "loss": 1.5966,
+      "step": 126700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.062707051585424e-05,
+      "loss": 1.593,
+      "step": 126800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0619675816374825e-05,
+      "loss": 1.5904,
+      "step": 126900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.061228111689541e-05,
+      "loss": 1.5995,
+      "step": 127000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0604886417415996e-05,
+      "loss": 1.5983,
+      "step": 127100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.059749171793659e-05,
+      "loss": 1.6027,
+      "step": 127200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.0590097018457175e-05,
+      "loss": 1.5768,
+      "step": 127300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.058270231897776e-05,
+      "loss": 1.6083,
+      "step": 127400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0575307619498346e-05,
+      "loss": 1.6082,
+      "step": 127500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0567986867013726e-05,
+      "loss": 1.5841,
+      "step": 127600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.056059216753431e-05,
+      "loss": 1.617,
+      "step": 127700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.05531974680549e-05,
+      "loss": 1.6136,
+      "step": 127800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.054580276857548e-05,
+      "loss": 1.6089,
+      "step": 127900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0538408069096076e-05,
+      "loss": 1.6092,
+      "step": 128000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.053101336961666e-05,
+      "loss": 1.6015,
+      "step": 128100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.052361867013725e-05,
+      "loss": 1.5771,
+      "step": 128200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.051622397065783e-05,
+      "loss": 1.5755,
+      "step": 128300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0508829271178425e-05,
+      "loss": 1.5881,
+      "step": 128400
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.050143457169901e-05,
+      "loss": 1.5942,
+      "step": 128500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0494039872219596e-05,
+      "loss": 1.608,
+      "step": 128600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.048664517274018e-05,
+      "loss": 1.5872,
+      "step": 128700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0479250473260774e-05,
+      "loss": 1.5956,
+      "step": 128800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.047185577378135e-05,
+      "loss": 1.5904,
+      "step": 128900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.046446107430194e-05,
+      "loss": 1.597,
+      "step": 129000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0457066374822524e-05,
+      "loss": 1.5936,
+      "step": 129100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.044967167534312e-05,
+      "loss": 1.5797,
+      "step": 129200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.04422769758637e-05,
+      "loss": 1.6017,
+      "step": 129300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.043488227638429e-05,
+      "loss": 1.5854,
+      "step": 129400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0427487576904874e-05,
+      "loss": 1.6013,
+      "step": 129500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0420092877425466e-05,
+      "loss": 1.5885,
+      "step": 129600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.041269817794605e-05,
+      "loss": 1.58,
+      "step": 129700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.040530347846664e-05,
+      "loss": 1.5903,
+      "step": 129800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.039798272598202e-05,
+      "loss": 1.5957,
+      "step": 129900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.039058802650261e-05,
+      "loss": 1.6014,
+      "step": 130000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0383193327023196e-05,
+      "loss": 1.5769,
+      "step": 130100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0375798627543775e-05,
+      "loss": 1.5868,
+      "step": 130200
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036840392806436e-05,
+      "loss": 1.5793,
+      "step": 130300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.036100922858495e-05,
+      "loss": 1.5808,
+      "step": 130400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.035361452910554e-05,
+      "loss": 1.6049,
+      "step": 130500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0346219829626124e-05,
+      "loss": 1.5859,
+      "step": 130600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.033882513014671e-05,
+      "loss": 1.5937,
+      "step": 130700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.03314304306673e-05,
+      "loss": 1.5681,
+      "step": 130800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032403573118789e-05,
+      "loss": 1.5643,
+      "step": 130900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0316641031708473e-05,
+      "loss": 1.5766,
+      "step": 131000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.030924633222906e-05,
+      "loss": 1.5856,
+      "step": 131100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.030185163274965e-05,
+      "loss": 1.5886,
+      "step": 131200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.029445693327024e-05,
+      "loss": 1.5961,
+      "step": 131300
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.028706223379082e-05,
+      "loss": 1.5748,
+      "step": 131400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.02796675343114e-05,
+      "loss": 1.5699,
+      "step": 131500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0272272834831994e-05,
+      "loss": 1.5609,
+      "step": 131600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.026487813535258e-05,
+      "loss": 1.5749,
+      "step": 131700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0257483435873165e-05,
+      "loss": 1.6034,
+      "step": 131800
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.025008873639375e-05,
+      "loss": 1.5878,
+      "step": 131900
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.024276798390914e-05,
+      "loss": 1.578,
+      "step": 132000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0235373284429724e-05,
+      "loss": 1.5826,
+      "step": 132100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.022797858495031e-05,
+      "loss": 1.583,
+      "step": 132200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0220583885470895e-05,
+      "loss": 1.5857,
+      "step": 132300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.021318918599149e-05,
+      "loss": 1.5949,
+      "step": 132400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.020579448651207e-05,
+      "loss": 1.571,
+      "step": 132500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.019839978703266e-05,
+      "loss": 1.5904,
+      "step": 132600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0191005087553244e-05,
+      "loss": 1.6172,
+      "step": 132700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.018361038807383e-05,
+      "loss": 1.6077,
+      "step": 132800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0176215688594416e-05,
+      "loss": 1.5975,
+      "step": 132900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0168820989115e-05,
+      "loss": 1.6193,
+      "step": 133000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.016142628963559e-05,
+      "loss": 1.6009,
+      "step": 133100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.015403159015618e-05,
+      "loss": 1.6021,
+      "step": 133200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0146636890676765e-05,
+      "loss": 1.5894,
+      "step": 133300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.013924219119735e-05,
+      "loss": 1.5904,
+      "step": 133400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0131847491717936e-05,
+      "loss": 1.602,
+      "step": 133500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.012445279223853e-05,
+      "loss": 1.6167,
+      "step": 133600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0117058092759114e-05,
+      "loss": 1.5815,
+      "step": 133700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.01096633932797e-05,
+      "loss": 1.5991,
+      "step": 133800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0102268693800286e-05,
+      "loss": 1.5706,
+      "step": 133900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.009487399432088e-05,
+      "loss": 1.5871,
+      "step": 134000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.008747929484146e-05,
+      "loss": 1.5782,
+      "step": 134100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.008008459536204e-05,
+      "loss": 1.5842,
+      "step": 134200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.007268989588263e-05,
+      "loss": 1.5758,
+      "step": 134300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006529519640322e-05,
+      "loss": 1.6083,
+      "step": 134400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0057900496923806e-05,
+      "loss": 1.5591,
+      "step": 134500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.005050579744439e-05,
+      "loss": 1.5663,
+      "step": 134600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.004318504495977e-05,
+      "loss": 1.6085,
+      "step": 134700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0035790345480365e-05,
+      "loss": 1.5919,
+      "step": 134800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.002839564600095e-05,
+      "loss": 1.5955,
+      "step": 134900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0021000946521536e-05,
+      "loss": 1.5813,
+      "step": 135000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.001360624704212e-05,
+      "loss": 1.5893,
+      "step": 135100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0006211547562714e-05,
+      "loss": 1.5904,
+      "step": 135200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.99988168480833e-05,
+      "loss": 1.5645,
+      "step": 135300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9991422148603885e-05,
+      "loss": 1.5418,
+      "step": 135400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9984027449124464e-05,
+      "loss": 1.5324,
+      "step": 135500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9976632749645057e-05,
+      "loss": 1.5107,
+      "step": 135600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996923805016564e-05,
+      "loss": 1.512,
+      "step": 135700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996184335068623e-05,
+      "loss": 1.5242,
+      "step": 135800
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9954448651206814e-05,
+      "loss": 1.4948,
+      "step": 135900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.9947053951727406e-05,
+      "loss": 1.5368,
+      "step": 136000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.993965925224799e-05,
+      "loss": 1.5353,
+      "step": 136100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.993226455276858e-05,
+      "loss": 1.5298,
+      "step": 136200
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.992486985328916e-05,
+      "loss": 1.5413,
+      "step": 136300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9917475153809755e-05,
+      "loss": 1.498,
+      "step": 136400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.991008045433034e-05,
+      "loss": 1.5026,
+      "step": 136500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9902685754850927e-05,
+      "loss": 1.5059,
+      "step": 136600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.989529105537151e-05,
+      "loss": 1.5477,
+      "step": 136700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.98878963558921e-05,
+      "loss": 1.509,
+      "step": 136800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.9880501656412683e-05,
+      "loss": 1.5292,
+      "step": 136900
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.987310695693327e-05,
+      "loss": 1.5119,
+      "step": 137000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9865712257453855e-05,
+      "loss": 1.5227,
+      "step": 137100
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.985831755797445e-05,
+      "loss": 1.5055,
+      "step": 137200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.985099680548983e-05,
+      "loss": 1.5252,
+      "step": 137300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.984360210601041e-05,
+      "loss": 1.5189,
+      "step": 137400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9836207406531e-05,
+      "loss": 1.5386,
+      "step": 137500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982881270705159e-05,
+      "loss": 1.5112,
+      "step": 137600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.982141800757218e-05,
+      "loss": 1.5152,
+      "step": 137700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.981402330809276e-05,
+      "loss": 1.5212,
+      "step": 137800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.980662860861335e-05,
+      "loss": 1.5273,
+      "step": 137900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.979923390913394e-05,
+      "loss": 1.518,
+      "step": 138000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.979183920965452e-05,
+      "loss": 1.4853,
+      "step": 138100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.9784444510175105e-05,
+      "loss": 1.5302,
+      "step": 138200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.977704981069569e-05,
+      "loss": 1.524,
+      "step": 138300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976965511121628e-05,
+      "loss": 1.4986,
+      "step": 138400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.976226041173687e-05,
+      "loss": 1.5343,
+      "step": 138500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9754865712257454e-05,
+      "loss": 1.5223,
+      "step": 138600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.974747101277804e-05,
+      "loss": 1.5207,
+      "step": 138700
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.974007631329863e-05,
+      "loss": 1.5492,
+      "step": 138800
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.973268161381922e-05,
+      "loss": 1.5124,
+      "step": 138900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9725286914339804e-05,
+      "loss": 1.5268,
+      "step": 139000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971789221486039e-05,
+      "loss": 1.507,
+      "step": 139100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.971049751538098e-05,
+      "loss": 1.5437,
+      "step": 139200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.970310281590157e-05,
+      "loss": 1.5095,
+      "step": 139300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9695708116422146e-05,
+      "loss": 1.5011,
+      "step": 139400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.968838736393753e-05,
+      "loss": 1.5183,
+      "step": 139500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.968099266445812e-05,
+      "loss": 1.5071,
+      "step": 139600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9673597964978705e-05,
+      "loss": 1.5277,
+      "step": 139700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.966620326549929e-05,
+      "loss": 1.5349,
+      "step": 139800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9658808566019876e-05,
+      "loss": 1.5254,
+      "step": 139900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.965141386654047e-05,
+      "loss": 1.5267,
+      "step": 140000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9644019167061054e-05,
+      "loss": 1.508,
+      "step": 140100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.963662446758164e-05,
+      "loss": 1.5463,
+      "step": 140200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9629229768102225e-05,
+      "loss": 1.5292,
+      "step": 140300
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962183506862282e-05,
+      "loss": 1.5332,
+      "step": 140400
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9614440369143403e-05,
+      "loss": 1.5044,
+      "step": 140500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.960704566966399e-05,
+      "loss": 1.543,
+      "step": 140600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9599650970184575e-05,
+      "loss": 1.5229,
+      "step": 140700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.959225627070516e-05,
+      "loss": 1.5348,
+      "step": 140800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9584861571225746e-05,
+      "loss": 1.5165,
+      "step": 140900
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957746687174633e-05,
+      "loss": 1.5313,
+      "step": 141000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.957007217226692e-05,
+      "loss": 1.5268,
+      "step": 141100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.956267747278751e-05,
+      "loss": 1.5111,
+      "step": 141200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9555282773308095e-05,
+      "loss": 1.5268,
+      "step": 141300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.954788807382868e-05,
+      "loss": 1.5399,
+      "step": 141400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9540493374349267e-05,
+      "loss": 1.5195,
+      "step": 141500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.953309867486986e-05,
+      "loss": 1.5128,
+      "step": 141600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9525703975390445e-05,
+      "loss": 1.5272,
+      "step": 141700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9518383222905825e-05,
+      "loss": 1.5113,
+      "step": 141800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.951098852342641e-05,
+      "loss": 1.5051,
+      "step": 141900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.9503593823946996e-05,
+      "loss": 1.5356,
+      "step": 142000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.949619912446758e-05,
+      "loss": 1.5489,
+      "step": 142100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.948880442498817e-05,
+      "loss": 1.54,
+      "step": 142200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.948140972550875e-05,
+      "loss": 1.5468,
+      "step": 142300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9474015026029346e-05,
+      "loss": 1.5041,
+      "step": 142400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.946662032654993e-05,
+      "loss": 1.4953,
+      "step": 142500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.945922562707052e-05,
+      "loss": 1.5088,
+      "step": 142600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.94518309275911e-05,
+      "loss": 1.5399,
+      "step": 142700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9444436228111695e-05,
+      "loss": 1.5358,
+      "step": 142800
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.943704152863228e-05,
+      "loss": 1.5313,
+      "step": 142900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9429646829152866e-05,
+      "loss": 1.5396,
+      "step": 143000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.942225212967345e-05,
+      "loss": 1.5393,
+      "step": 143100
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.941485743019404e-05,
+      "loss": 1.5284,
+      "step": 143200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.940746273071463e-05,
+      "loss": 1.5039,
+      "step": 143300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.940006803123521e-05,
+      "loss": 1.5301,
+      "step": 143400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9392673331755795e-05,
+      "loss": 1.5297,
+      "step": 143500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.938527863227639e-05,
+      "loss": 1.534,
+      "step": 143600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937788393279697e-05,
+      "loss": 1.5147,
+      "step": 143700
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.937048923331756e-05,
+      "loss": 1.5064,
+      "step": 143800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9363094533838144e-05,
+      "loss": 1.5557,
+      "step": 143900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9355699834358736e-05,
+      "loss": 1.5135,
+      "step": 144000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.934830513487932e-05,
+      "loss": 1.5297,
+      "step": 144100
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.934091043539991e-05,
+      "loss": 1.5096,
+      "step": 144200
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.933351573592049e-05,
+      "loss": 1.5627,
+      "step": 144300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9326194983435874e-05,
+      "loss": 1.5208,
+      "step": 144400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9318800283956466e-05,
+      "loss": 1.5148,
+      "step": 144500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.931140558447705e-05,
+      "loss": 1.5257,
+      "step": 144600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.930401088499763e-05,
+      "loss": 1.5337,
+      "step": 144700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.929661618551822e-05,
+      "loss": 1.525,
+      "step": 144800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.928922148603881e-05,
+      "loss": 1.4842,
+      "step": 144900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9281826786559394e-05,
+      "loss": 1.5046,
+      "step": 145000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927443208707998e-05,
+      "loss": 1.5138,
+      "step": 145100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926703738760057e-05,
+      "loss": 1.5022,
+      "step": 145200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925964268812116e-05,
+      "loss": 1.5468,
+      "step": 145300
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9252247988641744e-05,
+      "loss": 1.5206,
+      "step": 145400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924485328916233e-05,
+      "loss": 1.5115,
+      "step": 145500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923745858968292e-05,
+      "loss": 1.5336,
+      "step": 145600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923006389020351e-05,
+      "loss": 1.5051,
+      "step": 145700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.922266919072409e-05,
+      "loss": 1.5278,
+      "step": 145800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.921527449124468e-05,
+      "loss": 1.5284,
+      "step": 145900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9207879791765264e-05,
+      "loss": 1.5171,
+      "step": 146000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.920048509228585e-05,
+      "loss": 1.5166,
+      "step": 146100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9193090392806435e-05,
+      "loss": 1.5325,
+      "step": 146200
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.918569569332702e-05,
+      "loss": 1.495,
+      "step": 146300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9178300993847613e-05,
+      "loss": 1.5307,
+      "step": 146400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.91709062943682e-05,
+      "loss": 1.5433,
+      "step": 146500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9163511594888785e-05,
+      "loss": 1.5535,
+      "step": 146600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.915611689540937e-05,
+      "loss": 1.5668,
+      "step": 146700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.914872219592996e-05,
+      "loss": 1.5206,
+      "step": 146800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.914132749645055e-05,
+      "loss": 1.5292,
+      "step": 146900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.9133932796971134e-05,
+      "loss": 1.5257,
+      "step": 147000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.912653809749172e-05,
+      "loss": 1.5159,
+      "step": 147100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.911914339801231e-05,
+      "loss": 1.5149,
+      "step": 147200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.91117486985329e-05,
+      "loss": 1.5172,
+      "step": 147300
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.910435399905348e-05,
+      "loss": 1.516,
+      "step": 147400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.909695929957406e-05,
+      "loss": 1.5206,
+      "step": 147500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9089564600094655e-05,
+      "loss": 1.5205,
+      "step": 147600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.908216990061524e-05,
+      "loss": 1.5496,
+      "step": 147700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.907484914813062e-05,
+      "loss": 1.5248,
+      "step": 147800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9067454448651206e-05,
+      "loss": 1.5232,
+      "step": 147900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.90600597491718e-05,
+      "loss": 1.5097,
+      "step": 148000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9052665049692384e-05,
+      "loss": 1.5178,
+      "step": 148100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.904527035021297e-05,
+      "loss": 1.4961,
+      "step": 148200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9037875650733556e-05,
+      "loss": 1.5114,
+      "step": 148300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.903048095125414e-05,
+      "loss": 1.5442,
+      "step": 148400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9023086251774734e-05,
+      "loss": 1.5203,
+      "step": 148500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.901569155229532e-05,
+      "loss": 1.5245,
+      "step": 148600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.90082968528159e-05,
+      "loss": 1.521,
+      "step": 148700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.900090215333649e-05,
+      "loss": 1.5147,
+      "step": 148800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8993507453857076e-05,
+      "loss": 1.5165,
+      "step": 148900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.898611275437766e-05,
+      "loss": 1.5301,
+      "step": 149000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897871805489825e-05,
+      "loss": 1.5131,
+      "step": 149100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.897132335541884e-05,
+      "loss": 1.5122,
+      "step": 149200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8963928655939426e-05,
+      "loss": 1.5287,
+      "step": 149300
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.895653395646001e-05,
+      "loss": 1.5202,
+      "step": 149400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.89491392569806e-05,
+      "loss": 1.5268,
+      "step": 149500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.894174455750119e-05,
+      "loss": 1.5686,
+      "step": 149600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.8934349858021775e-05,
+      "loss": 1.5052,
+      "step": 149700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8927029105537155e-05,
+      "loss": 1.5105,
+      "step": 149800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891963440605774e-05,
+      "loss": 1.5223,
+      "step": 149900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891223970657833e-05,
+      "loss": 1.521,
+      "step": 150000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.890484500709891e-05,
+      "loss": 1.537,
+      "step": 150100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.88974503076195e-05,
+      "loss": 1.5392,
+      "step": 150200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8890055608140084e-05,
+      "loss": 1.5224,
+      "step": 150300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8882660908660676e-05,
+      "loss": 1.5096,
+      "step": 150400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.887526620918126e-05,
+      "loss": 1.5212,
+      "step": 150500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886787150970185e-05,
+      "loss": 1.5359,
+      "step": 150600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886047681022243e-05,
+      "loss": 1.5101,
+      "step": 150700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.8853082110743025e-05,
+      "loss": 1.5368,
+      "step": 150800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.884568741126361e-05,
+      "loss": 1.5273,
+      "step": 150900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.88382927117842e-05,
+      "loss": 1.5458,
+      "step": 151000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.883089801230478e-05,
+      "loss": 1.5151,
+      "step": 151100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.882350331282537e-05,
+      "loss": 1.5246,
+      "step": 151200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8816108613345954e-05,
+      "loss": 1.5414,
+      "step": 151300
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.880871391386654e-05,
+      "loss": 1.5416,
+      "step": 151400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.8801319214387125e-05,
+      "loss": 1.5138,
+      "step": 151500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.879392451490772e-05,
+      "loss": 1.5274,
+      "step": 151600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.87865298154283e-05,
+      "loss": 1.5137,
+      "step": 151700
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.877913511594889e-05,
+      "loss": 1.5414,
+      "step": 151800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8771740416469474e-05,
+      "loss": 1.5365,
+      "step": 151900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8764345716990067e-05,
+      "loss": 1.5146,
+      "step": 152000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.875695101751065e-05,
+      "loss": 1.548,
+      "step": 152100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.874963026502603e-05,
+      "loss": 1.5208,
+      "step": 152200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.874223556554662e-05,
+      "loss": 1.5034,
+      "step": 152300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8734840866067204e-05,
+      "loss": 1.5216,
+      "step": 152400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8727446166587796e-05,
+      "loss": 1.4953,
+      "step": 152500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8720051467108375e-05,
+      "loss": 1.4898,
+      "step": 152600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.871265676762896e-05,
+      "loss": 1.4953,
+      "step": 152700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.870526206814955e-05,
+      "loss": 1.5367,
+      "step": 152800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.869786736867014e-05,
+      "loss": 1.5486,
+      "step": 152900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.8690472669190725e-05,
+      "loss": 1.5268,
+      "step": 153000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.868307796971131e-05,
+      "loss": 1.5197,
+      "step": 153100
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.86756832702319e-05,
+      "loss": 1.5053,
+      "step": 153200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.866828857075249e-05,
+      "loss": 1.5296,
+      "step": 153300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8660893871273074e-05,
+      "loss": 1.518,
+      "step": 153400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.865349917179366e-05,
+      "loss": 1.5227,
+      "step": 153500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.8646104472314245e-05,
+      "loss": 1.5223,
+      "step": 153600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.863870977283484e-05,
+      "loss": 1.5138,
+      "step": 153700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.863131507335542e-05,
+      "loss": 1.5296,
+      "step": 153800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.862392037387601e-05,
+      "loss": 1.5117,
+      "step": 153900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8616525674396594e-05,
+      "loss": 1.538,
+      "step": 154000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.860913097491718e-05,
+      "loss": 1.5303,
+      "step": 154100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8601736275437766e-05,
+      "loss": 1.5112,
+      "step": 154200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8594415522953146e-05,
+      "loss": 1.5185,
+      "step": 154300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.858702082347373e-05,
+      "loss": 1.5206,
+      "step": 154400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8579626123994324e-05,
+      "loss": 1.5452,
+      "step": 154500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.857223142451491e-05,
+      "loss": 1.5308,
+      "step": 154600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8564836725035496e-05,
+      "loss": 1.5355,
+      "step": 154700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.855744202555608e-05,
+      "loss": 1.5261,
+      "step": 154800
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8550047326076674e-05,
+      "loss": 1.5119,
+      "step": 154900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.854265262659726e-05,
+      "loss": 1.542,
+      "step": 155000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.8535257927117845e-05,
+      "loss": 1.5279,
+      "step": 155100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.852786322763843e-05,
+      "loss": 1.5346,
+      "step": 155200
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8520468528159016e-05,
+      "loss": 1.5291,
+      "step": 155300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.85130738286796e-05,
+      "loss": 1.5604,
+      "step": 155400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.850567912920019e-05,
+      "loss": 1.4958,
+      "step": 155500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.849828442972078e-05,
+      "loss": 1.5123,
+      "step": 155600
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.8490889730241365e-05,
+      "loss": 1.5425,
+      "step": 155700
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.848349503076195e-05,
+      "loss": 1.4989,
+      "step": 155800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.847610033128254e-05,
+      "loss": 1.5123,
+      "step": 155900
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.846870563180313e-05,
+      "loss": 1.5212,
+      "step": 156000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8461310932323715e-05,
+      "loss": 1.5018,
+      "step": 156100
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.84539162328443e-05,
+      "loss": 1.5205,
+      "step": 156200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8446521533364886e-05,
+      "loss": 1.5217,
+      "step": 156300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8439200780880267e-05,
+      "loss": 1.5149,
+      "step": 156400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.843180608140086e-05,
+      "loss": 1.4915,
+      "step": 156500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.842441138192144e-05,
+      "loss": 1.5571,
+      "step": 156600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8417016682442023e-05,
+      "loss": 1.5349,
+      "step": 156700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.8409621982962616e-05,
+      "loss": 1.5304,
+      "step": 156800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.84022272834832e-05,
+      "loss": 1.5405,
+      "step": 156900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.839483258400379e-05,
+      "loss": 1.5182,
+      "step": 157000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.838743788452437e-05,
+      "loss": 1.5217,
+      "step": 157100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.838004318504496e-05,
+      "loss": 1.4984,
+      "step": 157200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.837264848556555e-05,
+      "loss": 1.5241,
+      "step": 157300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8365253786086136e-05,
+      "loss": 1.5191,
+      "step": 157400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.835785908660672e-05,
+      "loss": 1.519,
+      "step": 157500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.835046438712731e-05,
+      "loss": 1.5303,
+      "step": 157600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.83430696876479e-05,
+      "loss": 1.5395,
+      "step": 157700
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.8335674988168486e-05,
+      "loss": 1.5196,
+      "step": 157800
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8328280288689065e-05,
+      "loss": 1.5375,
+      "step": 157900
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.832088558920966e-05,
+      "loss": 1.5428,
+      "step": 158000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.831349088973024e-05,
+      "loss": 1.5359,
+      "step": 158100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.830609619025083e-05,
+      "loss": 1.513,
+      "step": 158200
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8298701490771414e-05,
+      "loss": 1.5272,
+      "step": 158300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8291306791292006e-05,
+      "loss": 1.5364,
+      "step": 158400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.828391209181259e-05,
+      "loss": 1.5082,
+      "step": 158500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.827651739233318e-05,
+      "loss": 1.5272,
+      "step": 158600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826912269285376e-05,
+      "loss": 1.543,
+      "step": 158700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.826172799337435e-05,
+      "loss": 1.5267,
+      "step": 158800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.825433329389494e-05,
+      "loss": 1.5242,
+      "step": 158900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.824693859441553e-05,
+      "loss": 1.5197,
+      "step": 159000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.823954389493611e-05,
+      "loss": 1.5113,
+      "step": 159100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.82321491954567e-05,
+      "loss": 1.5295,
+      "step": 159200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.822482844297208e-05,
+      "loss": 1.5319,
+      "step": 159300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8217433743492664e-05,
+      "loss": 1.5453,
+      "step": 159400
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.821003904401325e-05,
+      "loss": 1.5206,
+      "step": 159500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8202644344533836e-05,
+      "loss": 1.5244,
+      "step": 159600
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.819524964505443e-05,
+      "loss": 1.514,
+      "step": 159700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8187854945575014e-05,
+      "loss": 1.522,
+      "step": 159800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.81804602460956e-05,
+      "loss": 1.5148,
+      "step": 159900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8173065546616185e-05,
+      "loss": 1.5149,
+      "step": 160000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.816567084713678e-05,
+      "loss": 1.4791,
+      "step": 160100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.815827614765736e-05,
+      "loss": 1.5172,
+      "step": 160200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.815088144817795e-05,
+      "loss": 1.5129,
+      "step": 160300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8143486748698534e-05,
+      "loss": 1.5584,
+      "step": 160400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.813609204921913e-05,
+      "loss": 1.5532,
+      "step": 160500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8128697349739706e-05,
+      "loss": 1.5191,
+      "step": 160600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.812130265026029e-05,
+      "loss": 1.5165,
+      "step": 160700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8113907950780884e-05,
+      "loss": 1.5104,
+      "step": 160800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.810651325130147e-05,
+      "loss": 1.5127,
+      "step": 160900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8099118551822055e-05,
+      "loss": 1.5256,
+      "step": 161000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.809172385234264e-05,
+      "loss": 1.5308,
+      "step": 161100
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.808432915286323e-05,
+      "loss": 1.5015,
+      "step": 161200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.807693445338382e-05,
+      "loss": 1.5248,
+      "step": 161300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8069539753904404e-05,
+      "loss": 1.5092,
+      "step": 161400
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.806214505442499e-05,
+      "loss": 1.5393,
+      "step": 161500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8054750354945575e-05,
+      "loss": 1.5294,
+      "step": 161600
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.804735565546617e-05,
+      "loss": 1.5432,
+      "step": 161700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8039960955986754e-05,
+      "loss": 1.5007,
+      "step": 161800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.803256625650733e-05,
+      "loss": 1.5168,
+      "step": 161900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8025171557027925e-05,
+      "loss": 1.5084,
+      "step": 162000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.801777685754851e-05,
+      "loss": 1.5377,
+      "step": 162100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8010382158069096e-05,
+      "loss": 1.5564,
+      "step": 162200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.8003061405584477e-05,
+      "loss": 1.5204,
+      "step": 162300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.799566670610506e-05,
+      "loss": 1.5378,
+      "step": 162400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.7988272006625655e-05,
+      "loss": 1.5489,
+      "step": 162500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.798087730714624e-05,
+      "loss": 1.5433,
+      "step": 162600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7973482607666826e-05,
+      "loss": 1.4955,
+      "step": 162700
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.796608790818741e-05,
+      "loss": 1.5404,
+      "step": 162800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7958693208708004e-05,
+      "loss": 1.5299,
+      "step": 162900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.795129850922859e-05,
+      "loss": 1.5257,
+      "step": 163000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7943903809749175e-05,
+      "loss": 1.5343,
+      "step": 163100
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.793650911026976e-05,
+      "loss": 1.5286,
+      "step": 163200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7929114410790346e-05,
+      "loss": 1.4956,
+      "step": 163300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792171971131093e-05,
+      "loss": 1.541,
+      "step": 163400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.791432501183152e-05,
+      "loss": 1.5287,
+      "step": 163500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.790693031235211e-05,
+      "loss": 1.5048,
+      "step": 163600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.7899535612872696e-05,
+      "loss": 1.5468,
+      "step": 163700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.789214091339328e-05,
+      "loss": 1.5572,
+      "step": 163800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.788474621391387e-05,
+      "loss": 1.5488,
+      "step": 163900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.787735151443445e-05,
+      "loss": 1.5341,
+      "step": 164000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7869956814955045e-05,
+      "loss": 1.5196,
+      "step": 164100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786256211547563e-05,
+      "loss": 1.551,
+      "step": 164200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.7855167415996216e-05,
+      "loss": 1.5046,
+      "step": 164300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.78477727165168e-05,
+      "loss": 1.5463,
+      "step": 164400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.784037801703739e-05,
+      "loss": 1.5447,
+      "step": 164500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.783298331755797e-05,
+      "loss": 1.5415,
+      "step": 164600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.782558861807856e-05,
+      "loss": 1.5306,
+      "step": 164700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781819391859915e-05,
+      "loss": 1.5344,
+      "step": 164800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781079921911974e-05,
+      "loss": 1.5478,
+      "step": 164900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.780347846663512e-05,
+      "loss": 1.5138,
+      "step": 165000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.77960837671557e-05,
+      "loss": 1.538,
+      "step": 165100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.778868906767629e-05,
+      "loss": 1.509,
+      "step": 165200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.778129436819688e-05,
+      "loss": 1.5328,
+      "step": 165300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.777389966871747e-05,
+      "loss": 1.5146,
+      "step": 165400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776650496923805e-05,
+      "loss": 1.5116,
+      "step": 165500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775911026975864e-05,
+      "loss": 1.5542,
+      "step": 165600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775171557027923e-05,
+      "loss": 1.5504,
+      "step": 165700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7744320870799816e-05,
+      "loss": 1.5533,
+      "step": 165800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7736926171320395e-05,
+      "loss": 1.5306,
+      "step": 165900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.772953147184099e-05,
+      "loss": 1.5183,
+      "step": 166000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.772213677236157e-05,
+      "loss": 1.5319,
+      "step": 166100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771474207288216e-05,
+      "loss": 1.5312,
+      "step": 166200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7707347373402744e-05,
+      "loss": 1.498,
+      "step": 166300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.769995267392334e-05,
+      "loss": 1.5548,
+      "step": 166400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.769255797444392e-05,
+      "loss": 1.5129,
+      "step": 166500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.768516327496451e-05,
+      "loss": 1.5138,
+      "step": 166600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7677768575485094e-05,
+      "loss": 1.5326,
+      "step": 166700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.767037387600568e-05,
+      "loss": 1.5412,
+      "step": 166800
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766297917652627e-05,
+      "loss": 1.5079,
+      "step": 166900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.765558447704686e-05,
+      "loss": 1.5241,
+      "step": 167000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.764818977756744e-05,
+      "loss": 1.5191,
+      "step": 167100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.764079507808803e-05,
+      "loss": 1.5298,
+      "step": 167200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.7633400378608614e-05,
+      "loss": 1.5229,
+      "step": 167300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7626079626123995e-05,
+      "loss": 1.5194,
+      "step": 167400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.761868492664458e-05,
+      "loss": 1.5241,
+      "step": 167500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7611290227165166e-05,
+      "loss": 1.5364,
+      "step": 167600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.760389552768576e-05,
+      "loss": 1.5252,
+      "step": 167700
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7596500828206344e-05,
+      "loss": 1.5377,
+      "step": 167800
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.758910612872693e-05,
+      "loss": 1.5146,
+      "step": 167900
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.7581711429247515e-05,
+      "loss": 1.5325,
+      "step": 168000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.757431672976811e-05,
+      "loss": 1.5062,
+      "step": 168100
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.756692203028869e-05,
+      "loss": 1.5162,
+      "step": 168200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.755952733080928e-05,
+      "loss": 1.5347,
+      "step": 168300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7552132631329865e-05,
+      "loss": 1.5173,
+      "step": 168400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.754473793185045e-05,
+      "loss": 1.5102,
+      "step": 168500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.7537343232371036e-05,
+      "loss": 1.5276,
+      "step": 168600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.752994853289162e-05,
+      "loss": 1.5202,
+      "step": 168700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7522553833412214e-05,
+      "loss": 1.5366,
+      "step": 168800
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.75151591339328e-05,
+      "loss": 1.5176,
+      "step": 168900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7507764434453385e-05,
+      "loss": 1.5426,
+      "step": 169000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.750036973497397e-05,
+      "loss": 1.532,
+      "step": 169100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7492975035494556e-05,
+      "loss": 1.5048,
+      "step": 169200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.748558033601515e-05,
+      "loss": 1.5477,
+      "step": 169300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.747825958353053e-05,
+      "loss": 1.5361,
+      "step": 169400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7470864884051115e-05,
+      "loss": 1.5399,
+      "step": 169500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.74634701845717e-05,
+      "loss": 1.5545,
+      "step": 169600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.745607548509229e-05,
+      "loss": 1.5011,
+      "step": 169700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.744868078561287e-05,
+      "loss": 1.5148,
+      "step": 169800
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.744128608613346e-05,
+      "loss": 1.5231,
+      "step": 169900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.743389138665404e-05,
+      "loss": 1.5292,
+      "step": 170000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7426496687174636e-05,
+      "loss": 1.4809,
+      "step": 170100
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741910198769522e-05,
+      "loss": 1.5251,
+      "step": 170200
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.741170728821581e-05,
+      "loss": 1.5232,
+      "step": 170300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.740431258873639e-05,
+      "loss": 1.5087,
+      "step": 170400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7396917889256985e-05,
+      "loss": 1.5524,
+      "step": 170500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.738952318977757e-05,
+      "loss": 1.5114,
+      "step": 170600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7382128490298156e-05,
+      "loss": 1.5244,
+      "step": 170700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.737473379081874e-05,
+      "loss": 1.4936,
+      "step": 170800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7367339091339334e-05,
+      "loss": 1.5364,
+      "step": 170900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.735994439185992e-05,
+      "loss": 1.5315,
+      "step": 171000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.73525496923805e-05,
+      "loss": 1.5272,
+      "step": 171100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.734515499290109e-05,
+      "loss": 1.5493,
+      "step": 171200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.733776029342168e-05,
+      "loss": 1.5101,
+      "step": 171300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.733036559394226e-05,
+      "loss": 1.5398,
+      "step": 171400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.732297089446285e-05,
+      "loss": 1.5201,
+      "step": 171500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7315576194983434e-05,
+      "loss": 1.4922,
+      "step": 171600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.7308181495504026e-05,
+      "loss": 1.5116,
+      "step": 171700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.730078679602461e-05,
+      "loss": 1.5347,
+      "step": 171800
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.72933920965452e-05,
+      "loss": 1.5198,
+      "step": 171900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.728599739706578e-05,
+      "loss": 1.4854,
+      "step": 172000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7278602697586375e-05,
+      "loss": 1.5242,
+      "step": 172100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.727120799810696e-05,
+      "loss": 1.5208,
+      "step": 172200
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.726381329862755e-05,
+      "loss": 1.4922,
+      "step": 172300
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.725649254614293e-05,
+      "loss": 1.5441,
+      "step": 172400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.724909784666351e-05,
+      "loss": 1.553,
+      "step": 172500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.72417031471841e-05,
+      "loss": 1.5018,
+      "step": 172600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7234308447704684e-05,
+      "loss": 1.5003,
+      "step": 172700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.722691374822527e-05,
+      "loss": 1.5171,
+      "step": 172800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721951904874586e-05,
+      "loss": 1.5237,
+      "step": 172900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.721212434926645e-05,
+      "loss": 1.5277,
+      "step": 173000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7204729649787033e-05,
+      "loss": 1.5279,
+      "step": 173100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.719733495030762e-05,
+      "loss": 1.5025,
+      "step": 173200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.718994025082821e-05,
+      "loss": 1.5385,
+      "step": 173300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.71825455513488e-05,
+      "loss": 1.5167,
+      "step": 173400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.717515085186938e-05,
+      "loss": 1.5155,
+      "step": 173500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.716775615238997e-05,
+      "loss": 1.5142,
+      "step": 173600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.716036145291056e-05,
+      "loss": 1.5074,
+      "step": 173700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.715296675343114e-05,
+      "loss": 1.5261,
+      "step": 173800
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7145572053951725e-05,
+      "loss": 1.5034,
+      "step": 173900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.713817735447232e-05,
+      "loss": 1.5219,
+      "step": 174000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.71307826549929e-05,
+      "loss": 1.5377,
+      "step": 174100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.712338795551349e-05,
+      "loss": 1.5329,
+      "step": 174200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7115993256034075e-05,
+      "loss": 1.4909,
+      "step": 174300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7108672503549455e-05,
+      "loss": 1.5052,
+      "step": 174400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.710127780407005e-05,
+      "loss": 1.5397,
+      "step": 174500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.709388310459063e-05,
+      "loss": 1.5459,
+      "step": 174600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.708648840511122e-05,
+      "loss": 1.5564,
+      "step": 174700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7079093705631804e-05,
+      "loss": 1.529,
+      "step": 174800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.70716990061524e-05,
+      "loss": 1.5226,
+      "step": 174900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.706430430667298e-05,
+      "loss": 1.5175,
+      "step": 175000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.705690960719356e-05,
+      "loss": 1.5519,
+      "step": 175100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.704951490771415e-05,
+      "loss": 1.5236,
+      "step": 175200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.704212020823474e-05,
+      "loss": 1.5137,
+      "step": 175300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7034725508755325e-05,
+      "loss": 1.4916,
+      "step": 175400
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.702733080927591e-05,
+      "loss": 1.5355,
+      "step": 175500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7019936109796496e-05,
+      "loss": 1.5257,
+      "step": 175600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.701254141031709e-05,
+      "loss": 1.5348,
+      "step": 175700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7005146710837674e-05,
+      "loss": 1.5064,
+      "step": 175800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.699775201135826e-05,
+      "loss": 1.5246,
+      "step": 175900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.6990357311878846e-05,
+      "loss": 1.5271,
+      "step": 176000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.698296261239944e-05,
+      "loss": 1.4834,
+      "step": 176100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6975567912920024e-05,
+      "loss": 1.4941,
+      "step": 176200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.696817321344061e-05,
+      "loss": 1.5328,
+      "step": 176300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6960778513961195e-05,
+      "loss": 1.541,
+      "step": 176400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.695338381448178e-05,
+      "loss": 1.5115,
+      "step": 176500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.6945989115002366e-05,
+      "loss": 1.5309,
+      "step": 176600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.693859441552295e-05,
+      "loss": 1.5556,
+      "step": 176700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.693119971604354e-05,
+      "loss": 1.5194,
+      "step": 176800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.692380501656413e-05,
+      "loss": 1.523,
+      "step": 176900
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.691648426407951e-05,
+      "loss": 1.5487,
+      "step": 177000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6909089564600096e-05,
+      "loss": 1.5289,
+      "step": 177100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.690169486512068e-05,
+      "loss": 1.5006,
+      "step": 177200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.6894300165641274e-05,
+      "loss": 1.5327,
+      "step": 177300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.688690546616186e-05,
+      "loss": 1.5296,
+      "step": 177400
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6879510766682445e-05,
+      "loss": 1.5279,
+      "step": 177500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.687211606720303e-05,
+      "loss": 1.5324,
+      "step": 177600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6864721367723617e-05,
+      "loss": 1.5256,
+      "step": 177700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.68573266682442e-05,
+      "loss": 1.5204,
+      "step": 177800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.684993196876479e-05,
+      "loss": 1.5193,
+      "step": 177900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6842537269285373e-05,
+      "loss": 1.5009,
+      "step": 178000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6835142569805966e-05,
+      "loss": 1.5057,
+      "step": 178100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682774787032655e-05,
+      "loss": 1.519,
+      "step": 178200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682035317084714e-05,
+      "loss": 1.53,
+      "step": 178300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.681295847136772e-05,
+      "loss": 1.5103,
+      "step": 178400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6805563771888315e-05,
+      "loss": 1.4949,
+      "step": 178500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.67981690724089e-05,
+      "loss": 1.4811,
+      "step": 178600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6790774372929486e-05,
+      "loss": 1.517,
+      "step": 178700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.678337967345007e-05,
+      "loss": 1.5269,
+      "step": 178800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6775984973970665e-05,
+      "loss": 1.5268,
+      "step": 178900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.676859027449125e-05,
+      "loss": 1.499,
+      "step": 179000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.6761269522006624e-05,
+      "loss": 1.5117,
+      "step": 179100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.675387482252721e-05,
+      "loss": 1.5198,
+      "step": 179200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.67464801230478e-05,
+      "loss": 1.5191,
+      "step": 179300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.673908542356839e-05,
+      "loss": 1.5095,
+      "step": 179400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.673169072408897e-05,
+      "loss": 1.5299,
+      "step": 179500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.672429602460956e-05,
+      "loss": 1.5283,
+      "step": 179600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.671690132513015e-05,
+      "loss": 1.5228,
+      "step": 179700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.670950662565074e-05,
+      "loss": 1.5254,
+      "step": 179800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.670211192617132e-05,
+      "loss": 1.5281,
+      "step": 179900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.669471722669191e-05,
+      "loss": 1.5111,
+      "step": 180000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.66873225272125e-05,
+      "loss": 1.5066,
+      "step": 180100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6679927827733086e-05,
+      "loss": 1.5302,
+      "step": 180200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.667253312825367e-05,
+      "loss": 1.5206,
+      "step": 180300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666513842877425e-05,
+      "loss": 1.5028,
+      "step": 180400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.665774372929484e-05,
+      "loss": 1.5163,
+      "step": 180500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.665034902981543e-05,
+      "loss": 1.5158,
+      "step": 180600
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6642954330336014e-05,
+      "loss": 1.4898,
+      "step": 180700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.66355596308566e-05,
+      "loss": 1.5197,
+      "step": 180800
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662816493137719e-05,
+      "loss": 1.5166,
+      "step": 180900
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.662077023189778e-05,
+      "loss": 1.5093,
+      "step": 181000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6613375532418364e-05,
+      "loss": 1.5269,
+      "step": 181100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.660598083293895e-05,
+      "loss": 1.5286,
+      "step": 181200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.659858613345954e-05,
+      "loss": 1.5275,
+      "step": 181300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.659119143398013e-05,
+      "loss": 1.5296,
+      "step": 181400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.658379673450071e-05,
+      "loss": 1.5133,
+      "step": 181500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.65764020350213e-05,
+      "loss": 1.516,
+      "step": 181600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6569007335541884e-05,
+      "loss": 1.4879,
+      "step": 181700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.656161263606247e-05,
+      "loss": 1.5291,
+      "step": 181800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6554217936583056e-05,
+      "loss": 1.5436,
+      "step": 181900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.654682323710364e-05,
+      "loss": 1.5087,
+      "step": 182000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.653950248461903e-05,
+      "loss": 1.5057,
+      "step": 182100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6532107785139614e-05,
+      "loss": 1.5076,
+      "step": 182200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.65247130856602e-05,
+      "loss": 1.5163,
+      "step": 182300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6517318386180785e-05,
+      "loss": 1.5357,
+      "step": 182400
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.650992368670138e-05,
+      "loss": 1.5286,
+      "step": 182500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6502528987221963e-05,
+      "loss": 1.5326,
+      "step": 182600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.649513428774255e-05,
+      "loss": 1.5369,
+      "step": 182700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6487739588263135e-05,
+      "loss": 1.5321,
+      "step": 182800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.648034488878373e-05,
+      "loss": 1.5332,
+      "step": 182900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6472950189304306e-05,
+      "loss": 1.5201,
+      "step": 183000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.646555548982489e-05,
+      "loss": 1.5396,
+      "step": 183100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.645816079034548e-05,
+      "loss": 1.5077,
+      "step": 183200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.645076609086607e-05,
+      "loss": 1.5171,
+      "step": 183300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6443371391386655e-05,
+      "loss": 1.4933,
+      "step": 183400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.643597669190724e-05,
+      "loss": 1.5208,
+      "step": 183500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6428581992427827e-05,
+      "loss": 1.5156,
+      "step": 183600
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.642118729294842e-05,
+      "loss": 1.5185,
+      "step": 183700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6413792593469005e-05,
+      "loss": 1.5241,
+      "step": 183800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.640639789398959e-05,
+      "loss": 1.5073,
+      "step": 183900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6399003194510176e-05,
+      "loss": 1.5314,
+      "step": 184000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.639160849503077e-05,
+      "loss": 1.5269,
+      "step": 184100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6384213795551354e-05,
+      "loss": 1.5071,
+      "step": 184200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.637681909607193e-05,
+      "loss": 1.5156,
+      "step": 184300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.636949834358731e-05,
+      "loss": 1.5187,
+      "step": 184400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6362103644107906e-05,
+      "loss": 1.5487,
+      "step": 184500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.635470894462849e-05,
+      "loss": 1.5219,
+      "step": 184600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.634731424514908e-05,
+      "loss": 1.5236,
+      "step": 184700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.633991954566966e-05,
+      "loss": 1.5314,
+      "step": 184800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6332524846190255e-05,
+      "loss": 1.5143,
+      "step": 184900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.632513014671084e-05,
+      "loss": 1.5217,
+      "step": 185000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6317735447231426e-05,
+      "loss": 1.495,
+      "step": 185100
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.631034074775201e-05,
+      "loss": 1.5373,
+      "step": 185200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6302946048272604e-05,
+      "loss": 1.5249,
+      "step": 185300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.629555134879319e-05,
+      "loss": 1.5114,
+      "step": 185400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6288156649313776e-05,
+      "loss": 1.5211,
+      "step": 185500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.628076194983436e-05,
+      "loss": 1.5106,
+      "step": 185600
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.627336725035495e-05,
+      "loss": 1.5191,
+      "step": 185700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.626597255087553e-05,
+      "loss": 1.4975,
+      "step": 185800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.625857785139612e-05,
+      "loss": 1.5076,
+      "step": 185900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6251183151916704e-05,
+      "loss": 1.5015,
+      "step": 186000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6243788452437296e-05,
+      "loss": 1.4949,
+      "step": 186100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.623639375295788e-05,
+      "loss": 1.5212,
+      "step": 186200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.622899905347847e-05,
+      "loss": 1.5231,
+      "step": 186300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.622167830099385e-05,
+      "loss": 1.5121,
+      "step": 186400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.621428360151444e-05,
+      "loss": 1.5301,
+      "step": 186500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6206888902035026e-05,
+      "loss": 1.526,
+      "step": 186600
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.619949420255561e-05,
+      "loss": 1.5143,
+      "step": 186700
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.61920995030762e-05,
+      "loss": 1.5298,
+      "step": 186800
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.618470480359678e-05,
+      "loss": 1.5281,
+      "step": 186900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.617731010411737e-05,
+      "loss": 1.5238,
+      "step": 187000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6169915404637954e-05,
+      "loss": 1.5398,
+      "step": 187100
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.616252070515854e-05,
+      "loss": 1.5104,
+      "step": 187200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.615512600567913e-05,
+      "loss": 1.4956,
+      "step": 187300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.614773130619972e-05,
+      "loss": 1.5285,
+      "step": 187400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6140336606720304e-05,
+      "loss": 1.5209,
+      "step": 187500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.613294190724089e-05,
+      "loss": 1.5156,
+      "step": 187600
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.612554720776148e-05,
+      "loss": 1.5481,
+      "step": 187700
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611815250828207e-05,
+      "loss": 1.5176,
+      "step": 187800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.611075780880265e-05,
+      "loss": 1.5333,
+      "step": 187900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.610336310932324e-05,
+      "loss": 1.5093,
+      "step": 188000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.609596840984383e-05,
+      "loss": 1.5141,
+      "step": 188100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6088573710364417e-05,
+      "loss": 1.489,
+      "step": 188200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6081179010884995e-05,
+      "loss": 1.5148,
+      "step": 188300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.607378431140558e-05,
+      "loss": 1.5215,
+      "step": 188400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6066389611926173e-05,
+      "loss": 1.5124,
+      "step": 188500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.605899491244676e-05,
+      "loss": 1.5182,
+      "step": 188600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6051600212967345e-05,
+      "loss": 1.5096,
+      "step": 188700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.604420551348793e-05,
+      "loss": 1.4897,
+      "step": 188800
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.603681081400852e-05,
+      "loss": 1.5105,
+      "step": 188900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602941611452911e-05,
+      "loss": 1.4967,
+      "step": 189000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.602209536204449e-05,
+      "loss": 1.518,
+      "step": 189100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6014700662565074e-05,
+      "loss": 1.4998,
+      "step": 189200
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.600730596308566e-05,
+      "loss": 1.4873,
+      "step": 189300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.599991126360625e-05,
+      "loss": 1.5366,
+      "step": 189400
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.599251656412684e-05,
+      "loss": 1.5384,
+      "step": 189500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.598512186464742e-05,
+      "loss": 1.5302,
+      "step": 189600
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.597772716516801e-05,
+      "loss": 1.5117,
+      "step": 189700
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5970332465688595e-05,
+      "loss": 1.5345,
+      "step": 189800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.596293776620918e-05,
+      "loss": 1.4953,
+      "step": 189900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5955543066729766e-05,
+      "loss": 1.518,
+      "step": 190000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.594814836725036e-05,
+      "loss": 1.5336,
+      "step": 190100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.5940753667770944e-05,
+      "loss": 1.5194,
+      "step": 190200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.593335896829153e-05,
+      "loss": 1.532,
+      "step": 190300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5925964268812116e-05,
+      "loss": 1.5191,
+      "step": 190400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.591856956933271e-05,
+      "loss": 1.4929,
+      "step": 190500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5911174869853294e-05,
+      "loss": 1.5123,
+      "step": 190600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.590378017037388e-05,
+      "loss": 1.5103,
+      "step": 190700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5896385470894465e-05,
+      "loss": 1.5328,
+      "step": 190800
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.588899077141506e-05,
+      "loss": 1.5155,
+      "step": 190900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.5881596071935636e-05,
+      "loss": 1.5347,
+      "step": 191000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.587420137245622e-05,
+      "loss": 1.5276,
+      "step": 191100
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.586680667297681e-05,
+      "loss": 1.5183,
+      "step": 191200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.58594119734974e-05,
+      "loss": 1.484,
+      "step": 191300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5852017274017986e-05,
+      "loss": 1.547,
+      "step": 191400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5844696521533366e-05,
+      "loss": 1.5224,
+      "step": 191500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.583730182205395e-05,
+      "loss": 1.5068,
+      "step": 191600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5829907122574544e-05,
+      "loss": 1.4842,
+      "step": 191700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.582251242309513e-05,
+      "loss": 1.505,
+      "step": 191800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5815117723615715e-05,
+      "loss": 1.514,
+      "step": 191900
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.58077230241363e-05,
+      "loss": 1.5378,
+      "step": 192000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.580032832465689e-05,
+      "loss": 1.5025,
+      "step": 192100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.579293362517748e-05,
+      "loss": 1.5358,
+      "step": 192200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.578553892569806e-05,
+      "loss": 1.5185,
+      "step": 192300
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5778144226218644e-05,
+      "loss": 1.4998,
+      "step": 192400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5770749526739236e-05,
+      "loss": 1.511,
+      "step": 192500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.576335482725982e-05,
+      "loss": 1.4953,
+      "step": 192600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.575596012778041e-05,
+      "loss": 1.4966,
+      "step": 192700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.574856542830099e-05,
+      "loss": 1.5428,
+      "step": 192800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5741170728821585e-05,
+      "loss": 1.4629,
+      "step": 192900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.573377602934217e-05,
+      "loss": 1.4836,
+      "step": 193000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5726381329862757e-05,
+      "loss": 1.5118,
+      "step": 193100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571898663038334e-05,
+      "loss": 1.5125,
+      "step": 193200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5711591930903935e-05,
+      "loss": 1.5186,
+      "step": 193300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.570419723142452e-05,
+      "loss": 1.5334,
+      "step": 193400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5696802531945106e-05,
+      "loss": 1.5433,
+      "step": 193500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5689407832465685e-05,
+      "loss": 1.4849,
+      "step": 193600
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.568201313298628e-05,
+      "loss": 1.5385,
+      "step": 193700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567461843350686e-05,
+      "loss": 1.5001,
+      "step": 193800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.566722373402745e-05,
+      "loss": 1.5286,
+      "step": 193900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5659829034548034e-05,
+      "loss": 1.5162,
+      "step": 194000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5652434335068627e-05,
+      "loss": 1.4958,
+      "step": 194100
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.564503963558921e-05,
+      "loss": 1.5164,
+      "step": 194200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.563771888310459e-05,
+      "loss": 1.5182,
+      "step": 194300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.563032418362518e-05,
+      "loss": 1.5071,
+      "step": 194400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5622929484145764e-05,
+      "loss": 1.518,
+      "step": 194500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5615534784666356e-05,
+      "loss": 1.5085,
+      "step": 194600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.560814008518694e-05,
+      "loss": 1.5154,
+      "step": 194700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.560074538570753e-05,
+      "loss": 1.515,
+      "step": 194800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.559335068622811e-05,
+      "loss": 1.5017,
+      "step": 194900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.55859559867487e-05,
+      "loss": 1.4901,
+      "step": 195000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5578561287269285e-05,
+      "loss": 1.4999,
+      "step": 195100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.557116658778987e-05,
+      "loss": 1.5216,
+      "step": 195200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.556377188831046e-05,
+      "loss": 1.5138,
+      "step": 195300
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.555637718883105e-05,
+      "loss": 1.4859,
+      "step": 195400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5548982489351634e-05,
+      "loss": 1.554,
+      "step": 195500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.554158778987222e-05,
+      "loss": 1.5064,
+      "step": 195600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.553419309039281e-05,
+      "loss": 1.5136,
+      "step": 195700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.55267983909134e-05,
+      "loss": 1.5297,
+      "step": 195800
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551940369143398e-05,
+      "loss": 1.5249,
+      "step": 195900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551200899195457e-05,
+      "loss": 1.5049,
+      "step": 196000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.550461429247516e-05,
+      "loss": 1.4953,
+      "step": 196100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.549721959299574e-05,
+      "loss": 1.5036,
+      "step": 196200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5489824893516326e-05,
+      "loss": 1.5086,
+      "step": 196300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.548243019403691e-05,
+      "loss": 1.4954,
+      "step": 196400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5475035494557504e-05,
+      "loss": 1.5042,
+      "step": 196500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.546764079507809e-05,
+      "loss": 1.5187,
+      "step": 196600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5460246095598675e-05,
+      "loss": 1.5182,
+      "step": 196700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5452925343114055e-05,
+      "loss": 1.5204,
+      "step": 196800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.544553064363465e-05,
+      "loss": 1.5163,
+      "step": 196900
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5438135944155234e-05,
+      "loss": 1.5131,
+      "step": 197000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.543074124467582e-05,
+      "loss": 1.4985,
+      "step": 197100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5423346545196405e-05,
+      "loss": 1.5153,
+      "step": 197200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.541595184571699e-05,
+      "loss": 1.5113,
+      "step": 197300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.540855714623758e-05,
+      "loss": 1.5174,
+      "step": 197400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.540116244675817e-05,
+      "loss": 1.5142,
+      "step": 197500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.539376774727875e-05,
+      "loss": 1.5156,
+      "step": 197600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.538637304779934e-05,
+      "loss": 1.5239,
+      "step": 197700
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5378978348319925e-05,
+      "loss": 1.5239,
+      "step": 197800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.537158364884051e-05,
+      "loss": 1.5112,
+      "step": 197900
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.53641889493611e-05,
+      "loss": 1.5009,
+      "step": 198000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.535679424988169e-05,
+      "loss": 1.5268,
+      "step": 198100
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5349399550402275e-05,
+      "loss": 1.5139,
+      "step": 198200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.534200485092286e-05,
+      "loss": 1.5154,
+      "step": 198300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5334610151443446e-05,
+      "loss": 1.5108,
+      "step": 198400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.532721545196404e-05,
+      "loss": 1.5173,
+      "step": 198500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5319820752484624e-05,
+      "loss": 1.5252,
+      "step": 198600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.531242605300521e-05,
+      "loss": 1.4895,
+      "step": 198700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5305031353525795e-05,
+      "loss": 1.5209,
+      "step": 198800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.529763665404638e-05,
+      "loss": 1.5039,
+      "step": 198900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.529024195456697e-05,
+      "loss": 1.536,
+      "step": 199000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.528284725508755e-05,
+      "loss": 1.4951,
+      "step": 199100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.527545255560814e-05,
+      "loss": 1.5265,
+      "step": 199200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526805785612873e-05,
+      "loss": 1.4864,
+      "step": 199300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.526073710364411e-05,
+      "loss": 1.5132,
+      "step": 199400
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5253342404164696e-05,
+      "loss": 1.5211,
+      "step": 199500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.524594770468528e-05,
+      "loss": 1.5048,
+      "step": 199600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.523855300520587e-05,
+      "loss": 1.48,
+      "step": 199700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.523115830572646e-05,
+      "loss": 1.4973,
+      "step": 199800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5223763606247046e-05,
+      "loss": 1.5294,
+      "step": 199900
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.521636890676763e-05,
+      "loss": 1.4917,
+      "step": 200000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.520897420728822e-05,
+      "loss": 1.5269,
+      "step": 200100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.52015795078088e-05,
+      "loss": 1.5321,
+      "step": 200200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.519418480832939e-05,
+      "loss": 1.5146,
+      "step": 200300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5186790108849974e-05,
+      "loss": 1.5024,
+      "step": 200400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5179395409370566e-05,
+      "loss": 1.4993,
+      "step": 200500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.517200070989115e-05,
+      "loss": 1.5319,
+      "step": 200600
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.516460601041174e-05,
+      "loss": 1.5143,
+      "step": 200700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.515721131093232e-05,
+      "loss": 1.5179,
+      "step": 200800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.5149816611452916e-05,
+      "loss": 1.5357,
+      "step": 200900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.51424219119735e-05,
+      "loss": 1.5288,
+      "step": 201000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.513502721249409e-05,
+      "loss": 1.4975,
+      "step": 201100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.512763251301467e-05,
+      "loss": 1.5186,
+      "step": 201200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5120237813535265e-05,
+      "loss": 1.5258,
+      "step": 201300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.511284311405585e-05,
+      "loss": 1.515,
+      "step": 201400
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510544841457643e-05,
+      "loss": 1.5121,
+      "step": 201500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.509812766209181e-05,
+      "loss": 1.4926,
+      "step": 201600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.50907329626124e-05,
+      "loss": 1.5254,
+      "step": 201700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.508333826313299e-05,
+      "loss": 1.5122,
+      "step": 201800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5075943563653574e-05,
+      "loss": 1.5524,
+      "step": 201900
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506854886417416e-05,
+      "loss": 1.4987,
+      "step": 202000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.506115416469475e-05,
+      "loss": 1.5117,
+      "step": 202100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.505375946521534e-05,
+      "loss": 1.5141,
+      "step": 202200
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.504636476573592e-05,
+      "loss": 1.5227,
+      "step": 202300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.503897006625651e-05,
+      "loss": 1.5017,
+      "step": 202400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5031575366777094e-05,
+      "loss": 1.5095,
+      "step": 202500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.502418066729769e-05,
+      "loss": 1.4859,
+      "step": 202600
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.501678596781827e-05,
+      "loss": 1.5046,
+      "step": 202700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.500939126833885e-05,
+      "loss": 1.4923,
+      "step": 202800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5001996568859444e-05,
+      "loss": 1.476,
+      "step": 202900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.499460186938003e-05,
+      "loss": 1.4202,
+      "step": 203000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4987207169900615e-05,
+      "loss": 1.4452,
+      "step": 203100
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.49798124704212e-05,
+      "loss": 1.4498,
+      "step": 203200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.497241777094179e-05,
+      "loss": 1.4632,
+      "step": 203300
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.496502307146238e-05,
+      "loss": 1.4482,
+      "step": 203400
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.4957628371982964e-05,
+      "loss": 1.433,
+      "step": 203500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.495023367250355e-05,
+      "loss": 1.4313,
+      "step": 203600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.494283897302414e-05,
+      "loss": 1.4511,
+      "step": 203700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.493551822053952e-05,
+      "loss": 1.448,
+      "step": 203800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.492812352106011e-05,
+      "loss": 1.4216,
+      "step": 203900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4920728821580694e-05,
+      "loss": 1.4277,
+      "step": 204000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.491333412210128e-05,
+      "loss": 1.45,
+      "step": 204100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4905939422621865e-05,
+      "loss": 1.4198,
+      "step": 204200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.489854472314245e-05,
+      "loss": 1.4095,
+      "step": 204300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.4891150023663036e-05,
+      "loss": 1.4597,
+      "step": 204400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488375532418363e-05,
+      "loss": 1.444,
+      "step": 204500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4876360624704215e-05,
+      "loss": 1.4615,
+      "step": 204600
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.48689659252248e-05,
+      "loss": 1.4499,
+      "step": 204700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4861571225745386e-05,
+      "loss": 1.4282,
+      "step": 204800
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.485417652626597e-05,
+      "loss": 1.4389,
+      "step": 204900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4846781826786564e-05,
+      "loss": 1.4435,
+      "step": 205000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.483938712730715e-05,
+      "loss": 1.4389,
+      "step": 205100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4831992427827735e-05,
+      "loss": 1.4442,
+      "step": 205200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.482459772834832e-05,
+      "loss": 1.4303,
+      "step": 205300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.481720302886891e-05,
+      "loss": 1.4535,
+      "step": 205400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.480980832938949e-05,
+      "loss": 1.4368,
+      "step": 205500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.480241362991008e-05,
+      "loss": 1.4427,
+      "step": 205600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.479501893043067e-05,
+      "loss": 1.4436,
+      "step": 205700
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4787624230951256e-05,
+      "loss": 1.4624,
+      "step": 205800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.478022953147184e-05,
+      "loss": 1.4673,
+      "step": 205900
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.477283483199243e-05,
+      "loss": 1.4762,
+      "step": 206000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.476544013251302e-05,
+      "loss": 1.4508,
+      "step": 206100
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4758045433033605e-05,
+      "loss": 1.4374,
+      "step": 206200
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.475065073355419e-05,
+      "loss": 1.4383,
+      "step": 206300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4743256034074776e-05,
+      "loss": 1.4746,
+      "step": 206400
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.473586133459537e-05,
+      "loss": 1.4509,
+      "step": 206500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4728466635115954e-05,
+      "loss": 1.4372,
+      "step": 206600
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.472107193563654e-05,
+      "loss": 1.4318,
+      "step": 206700
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.471367723615712e-05,
+      "loss": 1.4513,
+      "step": 206800
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.470628253667771e-05,
+      "loss": 1.449,
+      "step": 206900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.46988878371983e-05,
+      "loss": 1.4339,
+      "step": 207000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.469149313771888e-05,
+      "loss": 1.451,
+      "step": 207100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.468409843823947e-05,
+      "loss": 1.4445,
+      "step": 207200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.467670373876006e-05,
+      "loss": 1.4159,
+      "step": 207300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466938298627544e-05,
+      "loss": 1.4671,
+      "step": 207400
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.466198828679603e-05,
+      "loss": 1.4663,
+      "step": 207500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.465459358731661e-05,
+      "loss": 1.4643,
+      "step": 207600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.46471988878372e-05,
+      "loss": 1.4536,
+      "step": 207700
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.463980418835779e-05,
+      "loss": 1.4667,
+      "step": 207800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4632409488878376e-05,
+      "loss": 1.4345,
+      "step": 207900
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.462501478939896e-05,
+      "loss": 1.4282,
+      "step": 208000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461762008991955e-05,
+      "loss": 1.4354,
+      "step": 208100
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.461022539044013e-05,
+      "loss": 1.4437,
+      "step": 208200
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.460283069096072e-05,
+      "loss": 1.4407,
+      "step": 208300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4595435991481304e-05,
+      "loss": 1.4703,
+      "step": 208400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.45880412920019e-05,
+      "loss": 1.4376,
+      "step": 208500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.458064659252248e-05,
+      "loss": 1.4579,
+      "step": 208600
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.457325189304307e-05,
+      "loss": 1.4484,
+      "step": 208700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4565857193563654e-05,
+      "loss": 1.4395,
+      "step": 208800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4558462494084246e-05,
+      "loss": 1.4397,
+      "step": 208900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.455106779460483e-05,
+      "loss": 1.4588,
+      "step": 209000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.454367309512542e-05,
+      "loss": 1.4647,
+      "step": 209100
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.4536278395646e-05,
+      "loss": 1.4295,
+      "step": 209200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.452888369616659e-05,
+      "loss": 1.4432,
+      "step": 209300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.452148899668718e-05,
+      "loss": 1.4296,
+      "step": 209400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.451409429720776e-05,
+      "loss": 1.4767,
+      "step": 209500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4506699597728345e-05,
+      "loss": 1.4432,
+      "step": 209600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.449930489824894e-05,
+      "loss": 1.4422,
+      "step": 209700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4491910198769523e-05,
+      "loss": 1.4468,
+      "step": 209800
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.448451549929011e-05,
+      "loss": 1.461,
+      "step": 209900
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4477120799810695e-05,
+      "loss": 1.445,
+      "step": 210000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4469800047326075e-05,
+      "loss": 1.4507,
+      "step": 210100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.446240534784667e-05,
+      "loss": 1.4299,
+      "step": 210200
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.445501064836725e-05,
+      "loss": 1.4292,
+      "step": 210300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.444761594888784e-05,
+      "loss": 1.4423,
+      "step": 210400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4440221249408425e-05,
+      "loss": 1.4787,
+      "step": 210500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.443282654992902e-05,
+      "loss": 1.4368,
+      "step": 210600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.44254318504496e-05,
+      "loss": 1.4669,
+      "step": 210700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441803715097018e-05,
+      "loss": 1.4524,
+      "step": 210800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4410642451490774e-05,
+      "loss": 1.4455,
+      "step": 210900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.440324775201136e-05,
+      "loss": 1.4502,
+      "step": 211000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.4395853052531945e-05,
+      "loss": 1.4656,
+      "step": 211100
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.438845835305253e-05,
+      "loss": 1.4462,
+      "step": 211200
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.438106365357312e-05,
+      "loss": 1.4765,
+      "step": 211300
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.437366895409371e-05,
+      "loss": 1.4761,
+      "step": 211400
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4366274254614294e-05,
+      "loss": 1.4469,
+      "step": 211500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.435887955513488e-05,
+      "loss": 1.4518,
+      "step": 211600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.435148485565547e-05,
+      "loss": 1.4462,
+      "step": 211700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.434409015617606e-05,
+      "loss": 1.4291,
+      "step": 211800
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4336695456696644e-05,
+      "loss": 1.4782,
+      "step": 211900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.432930075721723e-05,
+      "loss": 1.4346,
+      "step": 212000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4321906057737815e-05,
+      "loss": 1.4632,
+      "step": 212100
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.43145113582584e-05,
+      "loss": 1.4507,
+      "step": 212200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4307116658778986e-05,
+      "loss": 1.4575,
+      "step": 212300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.429972195929957e-05,
+      "loss": 1.4961,
+      "step": 212400
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4292327259820164e-05,
+      "loss": 1.4423,
+      "step": 212500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.428493256034075e-05,
+      "loss": 1.4741,
+      "step": 212600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4277537860861336e-05,
+      "loss": 1.4779,
+      "step": 212700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4270217108376716e-05,
+      "loss": 1.4514,
+      "step": 212800
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.42628224088973e-05,
+      "loss": 1.4456,
+      "step": 212900
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4255427709417894e-05,
+      "loss": 1.4587,
+      "step": 213000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.424803300993848e-05,
+      "loss": 1.4517,
+      "step": 213100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.4240638310459065e-05,
+      "loss": 1.4655,
+      "step": 213200
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.423324361097965e-05,
+      "loss": 1.4575,
+      "step": 213300
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.422584891150024e-05,
+      "loss": 1.4428,
+      "step": 213400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421845421202082e-05,
+      "loss": 1.4334,
+      "step": 213500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.421105951254141e-05,
+      "loss": 1.436,
+      "step": 213600
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4203664813062e-05,
+      "loss": 1.4577,
+      "step": 213700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.4196270113582586e-05,
+      "loss": 1.4476,
+      "step": 213800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.418887541410317e-05,
+      "loss": 1.4709,
+      "step": 213900
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.418148071462376e-05,
+      "loss": 1.4479,
+      "step": 214000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.417408601514435e-05,
+      "loss": 1.4473,
+      "step": 214100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4166691315664935e-05,
+      "loss": 1.4334,
+      "step": 214200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.415929661618552e-05,
+      "loss": 1.4306,
+      "step": 214300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.415190191670611e-05,
+      "loss": 1.4555,
+      "step": 214400
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.414450721722669e-05,
+      "loss": 1.4437,
+      "step": 214500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.4137112517747285e-05,
+      "loss": 1.4559,
+      "step": 214600
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4129717818267864e-05,
+      "loss": 1.4304,
+      "step": 214700
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.412232311878845e-05,
+      "loss": 1.4627,
+      "step": 214800
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.4115002366303836e-05,
+      "loss": 1.4788,
+      "step": 214900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.410760766682442e-05,
+      "loss": 1.4392,
+      "step": 215000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.410021296734501e-05,
+      "loss": 1.4518,
+      "step": 215100
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.409281826786559e-05,
+      "loss": 1.427,
+      "step": 215200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.408542356838618e-05,
+      "loss": 1.4791,
+      "step": 215300
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.407802886890677e-05,
+      "loss": 1.4585,
+      "step": 215400
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.407063416942736e-05,
+      "loss": 1.4765,
+      "step": 215500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.406323946994794e-05,
+      "loss": 1.4367,
+      "step": 215600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.405584477046853e-05,
+      "loss": 1.4522,
+      "step": 215700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.404845007098912e-05,
+      "loss": 1.4528,
+      "step": 215800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.4041055371509706e-05,
+      "loss": 1.4597,
+      "step": 215900
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.403366067203029e-05,
+      "loss": 1.4231,
+      "step": 216000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.402626597255088e-05,
+      "loss": 1.4376,
+      "step": 216100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401887127307146e-05,
+      "loss": 1.4908,
+      "step": 216200
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.401147657359205e-05,
+      "loss": 1.4361,
+      "step": 216300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.4004081874112635e-05,
+      "loss": 1.4421,
+      "step": 216400
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.399668717463323e-05,
+      "loss": 1.4523,
+      "step": 216500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.398929247515381e-05,
+      "loss": 1.457,
+      "step": 216600
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.39818977756744e-05,
+      "loss": 1.4546,
+      "step": 216700
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3974503076194984e-05,
+      "loss": 1.4732,
+      "step": 216800
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.396710837671557e-05,
+      "loss": 1.4533,
+      "step": 216900
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395978762423096e-05,
+      "loss": 1.429,
+      "step": 217000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.395239292475154e-05,
+      "loss": 1.4433,
+      "step": 217100
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.394499822527213e-05,
+      "loss": 1.4395,
+      "step": 217200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.3937603525792714e-05,
+      "loss": 1.4702,
+      "step": 217300
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.39302088263133e-05,
+      "loss": 1.4479,
+      "step": 217400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3922814126833885e-05,
+      "loss": 1.4529,
+      "step": 217500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.391541942735447e-05,
+      "loss": 1.4576,
+      "step": 217600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3908024727875056e-05,
+      "loss": 1.4356,
+      "step": 217700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.390063002839565e-05,
+      "loss": 1.4582,
+      "step": 217800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.3893235328916234e-05,
+      "loss": 1.4645,
+      "step": 217900
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.388584062943682e-05,
+      "loss": 1.4593,
+      "step": 218000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3878445929957406e-05,
+      "loss": 1.4457,
+      "step": 218100
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3871051230478e-05,
+      "loss": 1.4242,
+      "step": 218200
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3863656530998584e-05,
+      "loss": 1.4487,
+      "step": 218300
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.385626183151917e-05,
+      "loss": 1.4566,
+      "step": 218400
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3848867132039755e-05,
+      "loss": 1.436,
+      "step": 218500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.384147243256035e-05,
+      "loss": 1.4592,
+      "step": 218600
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3834077733080926e-05,
+      "loss": 1.4626,
+      "step": 218700
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.382668303360151e-05,
+      "loss": 1.4797,
+      "step": 218800
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3819288334122104e-05,
+      "loss": 1.4678,
+      "step": 218900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.381189363464269e-05,
+      "loss": 1.4504,
+      "step": 219000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3804498935163275e-05,
+      "loss": 1.4582,
+      "step": 219100
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.379710423568386e-05,
+      "loss": 1.4346,
+      "step": 219200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.3789709536204454e-05,
+      "loss": 1.4601,
+      "step": 219300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.378231483672504e-05,
+      "loss": 1.4311,
+      "step": 219400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3774920137245625e-05,
+      "loss": 1.4747,
+      "step": 219500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.376752543776621e-05,
+      "loss": 1.4642,
+      "step": 219600
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3760130738286796e-05,
+      "loss": 1.4645,
+      "step": 219700
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.375273603880739e-05,
+      "loss": 1.4475,
+      "step": 219800
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3745341339327974e-05,
+      "loss": 1.4827,
+      "step": 219900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.373794663984855e-05,
+      "loss": 1.4653,
+      "step": 220000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.373062588736394e-05,
+      "loss": 1.4656,
+      "step": 220100
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3723231187884526e-05,
+      "loss": 1.4609,
+      "step": 220200
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.371583648840511e-05,
+      "loss": 1.4321,
+      "step": 220300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.37084417889257e-05,
+      "loss": 1.4321,
+      "step": 220400
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.370104708944628e-05,
+      "loss": 1.4848,
+      "step": 220500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3693652389966875e-05,
+      "loss": 1.4539,
+      "step": 220600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.368625769048746e-05,
+      "loss": 1.4658,
+      "step": 220700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3678862991008046e-05,
+      "loss": 1.4603,
+      "step": 220800
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.367146829152863e-05,
+      "loss": 1.451,
+      "step": 220900
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3664073592049225e-05,
+      "loss": 1.4548,
+      "step": 221000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.365667889256981e-05,
+      "loss": 1.4472,
+      "step": 221100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.3649284193090396e-05,
+      "loss": 1.4557,
+      "step": 221200
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.364188949361098e-05,
+      "loss": 1.4471,
+      "step": 221300
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.363449479413157e-05,
+      "loss": 1.452,
+      "step": 221400
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.362710009465215e-05,
+      "loss": 1.4725,
+      "step": 221500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361970539517274e-05,
+      "loss": 1.4553,
+      "step": 221600
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.361231069569333e-05,
+      "loss": 1.453,
+      "step": 221700
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.3604915996213916e-05,
+      "loss": 1.4278,
+      "step": 221800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.35975212967345e-05,
+      "loss": 1.4487,
+      "step": 221900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.359012659725509e-05,
+      "loss": 1.4678,
+      "step": 222000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.358273189777567e-05,
+      "loss": 1.458,
+      "step": 222100
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3575337198296266e-05,
+      "loss": 1.4512,
+      "step": 222200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356794249881685e-05,
+      "loss": 1.4444,
+      "step": 222300
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.356054779933744e-05,
+      "loss": 1.5072,
+      "step": 222400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.355315309985802e-05,
+      "loss": 1.4419,
+      "step": 222500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3545758400378615e-05,
+      "loss": 1.4642,
+      "step": 222600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3538363700899194e-05,
+      "loss": 1.4721,
+      "step": 222700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.353096900141978e-05,
+      "loss": 1.4793,
+      "step": 222800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.352357430194037e-05,
+      "loss": 1.4563,
+      "step": 222900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.351617960246096e-05,
+      "loss": 1.4351,
+      "step": 223000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.350878490298154e-05,
+      "loss": 1.4408,
+      "step": 223100
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.350139020350213e-05,
+      "loss": 1.4192,
+      "step": 223200
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.349399550402272e-05,
+      "loss": 1.4325,
+      "step": 223300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.348660080454331e-05,
+      "loss": 1.4408,
+      "step": 223400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347928005205869e-05,
+      "loss": 1.446,
+      "step": 223500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.347188535257927e-05,
+      "loss": 1.4552,
+      "step": 223600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.346449065309986e-05,
+      "loss": 1.4486,
+      "step": 223700
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.345709595362045e-05,
+      "loss": 1.4542,
+      "step": 223800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.344970125414104e-05,
+      "loss": 1.452,
+      "step": 223900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3442306554661616e-05,
+      "loss": 1.4718,
+      "step": 224000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.343491185518221e-05,
+      "loss": 1.48,
+      "step": 224100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3427517155702794e-05,
+      "loss": 1.4444,
+      "step": 224200
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.342012245622338e-05,
+      "loss": 1.449,
+      "step": 224300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3412727756743965e-05,
+      "loss": 1.4548,
+      "step": 224400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.340533305726456e-05,
+      "loss": 1.452,
+      "step": 224500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.339793835778514e-05,
+      "loss": 1.4536,
+      "step": 224600
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.339054365830573e-05,
+      "loss": 1.4558,
+      "step": 224700
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.3383148958826314e-05,
+      "loss": 1.4351,
+      "step": 224800
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.33757542593469e-05,
+      "loss": 1.44,
+      "step": 224900
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336835955986749e-05,
+      "loss": 1.4473,
+      "step": 225000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.336096486038808e-05,
+      "loss": 1.449,
+      "step": 225100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3353570160908664e-05,
+      "loss": 1.4554,
+      "step": 225200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.334617546142925e-05,
+      "loss": 1.4322,
+      "step": 225300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3338780761949835e-05,
+      "loss": 1.4404,
+      "step": 225400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.333138606247042e-05,
+      "loss": 1.4728,
+      "step": 225500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3323991362991006e-05,
+      "loss": 1.4521,
+      "step": 225600
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.33165966635116e-05,
+      "loss": 1.4484,
+      "step": 225700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.330927591102698e-05,
+      "loss": 1.4182,
+      "step": 225800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3301881211547565e-05,
+      "loss": 1.4491,
+      "step": 225900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.329448651206815e-05,
+      "loss": 1.4651,
+      "step": 226000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.3287091812588736e-05,
+      "loss": 1.4812,
+      "step": 226100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.327969711310933e-05,
+      "loss": 1.449,
+      "step": 226200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3272302413629914e-05,
+      "loss": 1.4565,
+      "step": 226300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.32649077141505e-05,
+      "loss": 1.4206,
+      "step": 226400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3257513014671085e-05,
+      "loss": 1.4458,
+      "step": 226500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.325011831519167e-05,
+      "loss": 1.5017,
+      "step": 226600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.3242723615712256e-05,
+      "loss": 1.4658,
+      "step": 226700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.323532891623284e-05,
+      "loss": 1.483,
+      "step": 226800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3227934216753435e-05,
+      "loss": 1.4776,
+      "step": 226900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.322053951727402e-05,
+      "loss": 1.4439,
+      "step": 227000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3213144817794606e-05,
+      "loss": 1.4507,
+      "step": 227100
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.320575011831519e-05,
+      "loss": 1.4624,
+      "step": 227200
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.319835541883578e-05,
+      "loss": 1.4533,
+      "step": 227300
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.319096071935637e-05,
+      "loss": 1.4531,
+      "step": 227400
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.3183566019876955e-05,
+      "loss": 1.4582,
+      "step": 227500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.317617132039754e-05,
+      "loss": 1.4636,
+      "step": 227600
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3168776620918126e-05,
+      "loss": 1.4405,
+      "step": 227700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316138192143872e-05,
+      "loss": 1.4343,
+      "step": 227800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.31539872219593e-05,
+      "loss": 1.458,
+      "step": 227900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.314659252247988e-05,
+      "loss": 1.4641,
+      "step": 228000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.3139197823000476e-05,
+      "loss": 1.4405,
+      "step": 228100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.313180312352106e-05,
+      "loss": 1.4493,
+      "step": 228200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.312448237103644e-05,
+      "loss": 1.4373,
+      "step": 228300
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311708767155703e-05,
+      "loss": 1.4467,
+      "step": 228400
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.310969297207761e-05,
+      "loss": 1.4587,
+      "step": 228500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.3102298272598206e-05,
+      "loss": 1.4464,
+      "step": 228600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.309490357311879e-05,
+      "loss": 1.4684,
+      "step": 228700
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.308750887363938e-05,
+      "loss": 1.4847,
+      "step": 228800
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.308011417415996e-05,
+      "loss": 1.4437,
+      "step": 228900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3072719474680555e-05,
+      "loss": 1.4265,
+      "step": 229000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.306532477520114e-05,
+      "loss": 1.4563,
+      "step": 229100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.3057930075721726e-05,
+      "loss": 1.4276,
+      "step": 229200
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305053537624231e-05,
+      "loss": 1.4528,
+      "step": 229300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.30431406767629e-05,
+      "loss": 1.4737,
+      "step": 229400
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.303574597728348e-05,
+      "loss": 1.4328,
+      "step": 229500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302835127780407e-05,
+      "loss": 1.4384,
+      "step": 229600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.302095657832466e-05,
+      "loss": 1.4221,
+      "step": 229700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.301356187884525e-05,
+      "loss": 1.447,
+      "step": 229800
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.300616717936583e-05,
+      "loss": 1.4384,
+      "step": 229900
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.299877247988642e-05,
+      "loss": 1.4497,
+      "step": 230000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2991377780407004e-05,
+      "loss": 1.4625,
+      "step": 230100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.2983983080927596e-05,
+      "loss": 1.4362,
+      "step": 230200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.297658838144818e-05,
+      "loss": 1.4556,
+      "step": 230300
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296919368196877e-05,
+      "loss": 1.4745,
+      "step": 230400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.296179898248935e-05,
+      "loss": 1.4879,
+      "step": 230500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.295440428300994e-05,
+      "loss": 1.4561,
+      "step": 230600
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.2947009583530524e-05,
+      "loss": 1.4837,
+      "step": 230700
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.293961488405111e-05,
+      "loss": 1.4721,
+      "step": 230800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.29322201845717e-05,
+      "loss": 1.4633,
+      "step": 230900
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.292482548509229e-05,
+      "loss": 1.4543,
+      "step": 231000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.291750473260767e-05,
+      "loss": 1.4759,
+      "step": 231100
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.2910110033128254e-05,
+      "loss": 1.4526,
+      "step": 231200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290271533364884e-05,
+      "loss": 1.4815,
+      "step": 231300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.289532063416943e-05,
+      "loss": 1.4567,
+      "step": 231400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.288792593469002e-05,
+      "loss": 1.4574,
+      "step": 231500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.28805312352106e-05,
+      "loss": 1.4739,
+      "step": 231600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.287313653573119e-05,
+      "loss": 1.4784,
+      "step": 231700
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.286574183625178e-05,
+      "loss": 1.4821,
+      "step": 231800
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285834713677236e-05,
+      "loss": 1.4552,
+      "step": 231900
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2850952437292946e-05,
+      "loss": 1.4453,
+      "step": 232000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.284355773781354e-05,
+      "loss": 1.4604,
+      "step": 232100
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.2836163038334124e-05,
+      "loss": 1.4495,
+      "step": 232200
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.282876833885471e-05,
+      "loss": 1.4648,
+      "step": 232300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2821373639375295e-05,
+      "loss": 1.4429,
+      "step": 232400
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.281397893989588e-05,
+      "loss": 1.4455,
+      "step": 232500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.280658424041647e-05,
+      "loss": 1.4326,
+      "step": 232600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.279918954093706e-05,
+      "loss": 1.4309,
+      "step": 232700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2791794841457645e-05,
+      "loss": 1.4444,
+      "step": 232800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.278440014197823e-05,
+      "loss": 1.4739,
+      "step": 232900
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.277700544249882e-05,
+      "loss": 1.4576,
+      "step": 233000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276961074301941e-05,
+      "loss": 1.4311,
+      "step": 233100
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.276228999053478e-05,
+      "loss": 1.466,
+      "step": 233200
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.275489529105537e-05,
+      "loss": 1.4605,
+      "step": 233300
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.274750059157596e-05,
+      "loss": 1.4525,
+      "step": 233400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2740105892096546e-05,
+      "loss": 1.4771,
+      "step": 233500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.273271119261713e-05,
+      "loss": 1.4862,
+      "step": 233600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.272531649313772e-05,
+      "loss": 1.4223,
+      "step": 233700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.271792179365831e-05,
+      "loss": 1.4843,
+      "step": 233800
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2710527094178895e-05,
+      "loss": 1.4683,
+      "step": 233900
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.270313239469948e-05,
+      "loss": 1.4354,
+      "step": 234000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.2695737695220066e-05,
+      "loss": 1.4617,
+      "step": 234100
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.268834299574066e-05,
+      "loss": 1.4489,
+      "step": 234200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2680948296261244e-05,
+      "loss": 1.4502,
+      "step": 234300
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.267355359678183e-05,
+      "loss": 1.4543,
+      "step": 234400
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2666158897302416e-05,
+      "loss": 1.4637,
+      "step": 234500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2658764197823e-05,
+      "loss": 1.4581,
+      "step": 234600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.265136949834359e-05,
+      "loss": 1.4234,
+      "step": 234700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.264397479886417e-05,
+      "loss": 1.4624,
+      "step": 234800
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.2636580099384765e-05,
+      "loss": 1.4669,
+      "step": 234900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.262918539990535e-05,
+      "loss": 1.4352,
+      "step": 235000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2621790700425936e-05,
+      "loss": 1.4327,
+      "step": 235100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2614469947941317e-05,
+      "loss": 1.4422,
+      "step": 235200
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.26070752484619e-05,
+      "loss": 1.4746,
+      "step": 235300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2599680548982495e-05,
+      "loss": 1.4763,
+      "step": 235400
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.259228584950308e-05,
+      "loss": 1.4368,
+      "step": 235500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2584891150023666e-05,
+      "loss": 1.4727,
+      "step": 235600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.257749645054425e-05,
+      "loss": 1.4363,
+      "step": 235700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2570101751064844e-05,
+      "loss": 1.4573,
+      "step": 235800
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.256270705158542e-05,
+      "loss": 1.4631,
+      "step": 235900
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.255531235210601e-05,
+      "loss": 1.4711,
+      "step": 236000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2547917652626594e-05,
+      "loss": 1.4499,
+      "step": 236100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.2540522953147187e-05,
+      "loss": 1.4454,
+      "step": 236200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.253312825366777e-05,
+      "loss": 1.449,
+      "step": 236300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.252573355418836e-05,
+      "loss": 1.4243,
+      "step": 236400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2518338854708943e-05,
+      "loss": 1.4591,
+      "step": 236500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.2510944155229536e-05,
+      "loss": 1.4638,
+      "step": 236600
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.250354945575012e-05,
+      "loss": 1.4424,
+      "step": 236700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.249615475627071e-05,
+      "loss": 1.4517,
+      "step": 236800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.248876005679129e-05,
+      "loss": 1.4548,
+      "step": 236900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2481365357311885e-05,
+      "loss": 1.4472,
+      "step": 237000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.247397065783247e-05,
+      "loss": 1.4397,
+      "step": 237100
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2466649905347844e-05,
+      "loss": 1.4491,
+      "step": 237200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.245925520586843e-05,
+      "loss": 1.4537,
+      "step": 237300
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.245186050638902e-05,
+      "loss": 1.4472,
+      "step": 237400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.244446580690961e-05,
+      "loss": 1.4218,
+      "step": 237500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2437071107430194e-05,
+      "loss": 1.475,
+      "step": 237600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242967640795078e-05,
+      "loss": 1.4464,
+      "step": 237700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.242228170847137e-05,
+      "loss": 1.4725,
+      "step": 237800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.241488700899196e-05,
+      "loss": 1.4732,
+      "step": 237900
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.240749230951254e-05,
+      "loss": 1.4584,
+      "step": 238000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.240009761003313e-05,
+      "loss": 1.4746,
+      "step": 238100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.239270291055372e-05,
+      "loss": 1.4642,
+      "step": 238200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.238530821107431e-05,
+      "loss": 1.4508,
+      "step": 238300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237791351159489e-05,
+      "loss": 1.4541,
+      "step": 238400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.237051881211547e-05,
+      "loss": 1.4451,
+      "step": 238500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2363124112636064e-05,
+      "loss": 1.4644,
+      "step": 238600
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.235572941315665e-05,
+      "loss": 1.4558,
+      "step": 238700
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2348334713677235e-05,
+      "loss": 1.417,
+      "step": 238800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.234094001419782e-05,
+      "loss": 1.4587,
+      "step": 238900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.233354531471841e-05,
+      "loss": 1.4327,
+      "step": 239000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2326150615239e-05,
+      "loss": 1.4856,
+      "step": 239100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2318755915759584e-05,
+      "loss": 1.4343,
+      "step": 239200
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.231136121628017e-05,
+      "loss": 1.4598,
+      "step": 239300
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.230396651680076e-05,
+      "loss": 1.4681,
+      "step": 239400
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.229657181732135e-05,
+      "loss": 1.4706,
+      "step": 239500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.2289177117841934e-05,
+      "loss": 1.446,
+      "step": 239600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.228178241836252e-05,
+      "loss": 1.4683,
+      "step": 239700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2274387718883105e-05,
+      "loss": 1.455,
+      "step": 239800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.226699301940369e-05,
+      "loss": 1.4581,
+      "step": 239900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.225967226691907e-05,
+      "loss": 1.4718,
+      "step": 240000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.225227756743966e-05,
+      "loss": 1.4662,
+      "step": 240100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.224488286796025e-05,
+      "loss": 1.4319,
+      "step": 240200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.2237488168480835e-05,
+      "loss": 1.4447,
+      "step": 240300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.223009346900142e-05,
+      "loss": 1.472,
+      "step": 240400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2222698769522006e-05,
+      "loss": 1.4484,
+      "step": 240500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.22153040700426e-05,
+      "loss": 1.4396,
+      "step": 240600
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2207909370563184e-05,
+      "loss": 1.4661,
+      "step": 240700
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.220051467108377e-05,
+      "loss": 1.4479,
+      "step": 240800
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.2193119971604355e-05,
+      "loss": 1.4598,
+      "step": 240900
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.218572527212495e-05,
+      "loss": 1.4468,
+      "step": 241000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.217833057264553e-05,
+      "loss": 1.4472,
+      "step": 241100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.217093587316611e-05,
+      "loss": 1.4582,
+      "step": 241200
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.21635411736867e-05,
+      "loss": 1.4578,
+      "step": 241300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.215614647420729e-05,
+      "loss": 1.467,
+      "step": 241400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.2148751774727876e-05,
+      "loss": 1.4569,
+      "step": 241500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.214135707524846e-05,
+      "loss": 1.4629,
+      "step": 241600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.213396237576905e-05,
+      "loss": 1.468,
+      "step": 241700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.212656767628964e-05,
+      "loss": 1.4412,
+      "step": 241800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2119172976810225e-05,
+      "loss": 1.4575,
+      "step": 241900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.211177827733081e-05,
+      "loss": 1.4483,
+      "step": 242000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2104383577851397e-05,
+      "loss": 1.4663,
+      "step": 242100
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.209698887837199e-05,
+      "loss": 1.4624,
+      "step": 242200
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2089594178892575e-05,
+      "loss": 1.4506,
+      "step": 242300
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.208219947941316e-05,
+      "loss": 1.4508,
+      "step": 242400
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2074878726928534e-05,
+      "loss": 1.4211,
+      "step": 242500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2067484027449126e-05,
+      "loss": 1.4594,
+      "step": 242600
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.206008932796971e-05,
+      "loss": 1.4498,
+      "step": 242700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.20526946284903e-05,
+      "loss": 1.4589,
+      "step": 242800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.204529992901088e-05,
+      "loss": 1.452,
+      "step": 242900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.2037905229531476e-05,
+      "loss": 1.4555,
+      "step": 243000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.203051053005206e-05,
+      "loss": 1.4408,
+      "step": 243100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.202311583057265e-05,
+      "loss": 1.4347,
+      "step": 243200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201572113109323e-05,
+      "loss": 1.425,
+      "step": 243300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.2008326431613825e-05,
+      "loss": 1.4706,
+      "step": 243400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.200093173213441e-05,
+      "loss": 1.4657,
+      "step": 243500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.1993537032654996e-05,
+      "loss": 1.4482,
+      "step": 243600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.198614233317558e-05,
+      "loss": 1.4613,
+      "step": 243700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.197874763369617e-05,
+      "loss": 1.4527,
+      "step": 243800
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.197135293421675e-05,
+      "loss": 1.4539,
+      "step": 243900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.196395823473734e-05,
+      "loss": 1.4557,
+      "step": 244000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1956563535257924e-05,
+      "loss": 1.4675,
+      "step": 244100
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.194916883577852e-05,
+      "loss": 1.4284,
+      "step": 244200
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.19417741362991e-05,
+      "loss": 1.4536,
+      "step": 244300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.193437943681969e-05,
+      "loss": 1.4614,
+      "step": 244400
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1926984737340274e-05,
+      "loss": 1.4691,
+      "step": 244500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1919590037860866e-05,
+      "loss": 1.4686,
+      "step": 244600
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.191219533838145e-05,
+      "loss": 1.4255,
+      "step": 244700
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.190480063890204e-05,
+      "loss": 1.4494,
+      "step": 244800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.189747988641742e-05,
+      "loss": 1.4466,
+      "step": 244900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.1890085186938004e-05,
+      "loss": 1.4677,
+      "step": 245000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.188269048745859e-05,
+      "loss": 1.4468,
+      "step": 245100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1875295787979175e-05,
+      "loss": 1.4638,
+      "step": 245200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186790108849976e-05,
+      "loss": 1.4574,
+      "step": 245300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.186050638902035e-05,
+      "loss": 1.4255,
+      "step": 245400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.185311168954094e-05,
+      "loss": 1.4608,
+      "step": 245500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1845716990061524e-05,
+      "loss": 1.4493,
+      "step": 245600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.183832229058211e-05,
+      "loss": 1.4608,
+      "step": 245700
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.18309275911027e-05,
+      "loss": 1.4477,
+      "step": 245800
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.182353289162329e-05,
+      "loss": 1.4867,
+      "step": 245900
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1816138192143873e-05,
+      "loss": 1.4555,
+      "step": 246000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.180874349266446e-05,
+      "loss": 1.4719,
+      "step": 246100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.180134879318505e-05,
+      "loss": 1.4805,
+      "step": 246200
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.179395409370564e-05,
+      "loss": 1.4482,
+      "step": 246300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1786559394226216e-05,
+      "loss": 1.4435,
+      "step": 246400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.17791646947468e-05,
+      "loss": 1.4361,
+      "step": 246500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1771769995267394e-05,
+      "loss": 1.4462,
+      "step": 246600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.176437529578798e-05,
+      "loss": 1.492,
+      "step": 246700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1756980596308565e-05,
+      "loss": 1.4231,
+      "step": 246800
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.174958589682915e-05,
+      "loss": 1.4501,
+      "step": 246900
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.174219119734974e-05,
+      "loss": 1.4567,
+      "step": 247000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.173479649787033e-05,
+      "loss": 1.4564,
+      "step": 247100
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1727401798390915e-05,
+      "loss": 1.4374,
+      "step": 247200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.17200070989115e-05,
+      "loss": 1.4556,
+      "step": 247300
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.171268634642689e-05,
+      "loss": 1.4478,
+      "step": 247400
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.170529164694747e-05,
+      "loss": 1.4674,
+      "step": 247500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.169789694746806e-05,
+      "loss": 1.4562,
+      "step": 247600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1690502247988644e-05,
+      "loss": 1.4535,
+      "step": 247700
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.168310754850923e-05,
+      "loss": 1.4544,
+      "step": 247800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1675712849029816e-05,
+      "loss": 1.4627,
+      "step": 247900
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.16683181495504e-05,
+      "loss": 1.4373,
+      "step": 248000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.166092345007099e-05,
+      "loss": 1.4433,
+      "step": 248100
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.165352875059158e-05,
+      "loss": 1.4748,
+      "step": 248200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1646134051112165e-05,
+      "loss": 1.4577,
+      "step": 248300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.163873935163275e-05,
+      "loss": 1.4591,
+      "step": 248400
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1631344652153336e-05,
+      "loss": 1.4396,
+      "step": 248500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.162394995267393e-05,
+      "loss": 1.4501,
+      "step": 248600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1616555253194514e-05,
+      "loss": 1.4706,
+      "step": 248700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.16091605537151e-05,
+      "loss": 1.4643,
+      "step": 248800
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.1601765854235686e-05,
+      "loss": 1.4331,
+      "step": 248900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.159437115475628e-05,
+      "loss": 1.4583,
+      "step": 249000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.158697645527686e-05,
+      "loss": 1.4725,
+      "step": 249100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157958175579744e-05,
+      "loss": 1.4618,
+      "step": 249200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.157218705631803e-05,
+      "loss": 1.477,
+      "step": 249300
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.156479235683862e-05,
+      "loss": 1.4867,
+      "step": 249400
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1557471604354e-05,
+      "loss": 1.4771,
+      "step": 249500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.155007690487459e-05,
+      "loss": 1.4643,
+      "step": 249600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.154268220539517e-05,
+      "loss": 1.4616,
+      "step": 249700
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1535287505915765e-05,
+      "loss": 1.4612,
+      "step": 249800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.152789280643635e-05,
+      "loss": 1.458,
+      "step": 249900
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1520498106956936e-05,
+      "loss": 1.4418,
+      "step": 250000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.151310340747752e-05,
+      "loss": 1.4659,
+      "step": 250100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.150570870799811e-05,
+      "loss": 1.4273,
+      "step": 250200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.14983140085187e-05,
+      "loss": 1.4438,
+      "step": 250300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.149091930903928e-05,
+      "loss": 1.4741,
+      "step": 250400
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1483524609559864e-05,
+      "loss": 1.468,
+      "step": 250500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.147612991008046e-05,
+      "loss": 1.4647,
+      "step": 250600
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.146873521060104e-05,
+      "loss": 1.4563,
+      "step": 250700
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.146134051112163e-05,
+      "loss": 1.459,
+      "step": 250800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1453945811642214e-05,
+      "loss": 1.4524,
+      "step": 250900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.1446551112162806e-05,
+      "loss": 1.4834,
+      "step": 251000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.143915641268339e-05,
+      "loss": 1.4779,
+      "step": 251100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.143176171320398e-05,
+      "loss": 1.4791,
+      "step": 251200
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.142436701372456e-05,
+      "loss": 1.4597,
+      "step": 251300
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1416972314245155e-05,
+      "loss": 1.4616,
+      "step": 251400
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.140957761476574e-05,
+      "loss": 1.4634,
+      "step": 251500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.1402182915286327e-05,
+      "loss": 1.4781,
+      "step": 251600
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.13948621628017e-05,
+      "loss": 1.467,
+      "step": 251700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.138746746332229e-05,
+      "loss": 1.4414,
+      "step": 251800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.138007276384288e-05,
+      "loss": 1.4603,
+      "step": 251900
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1372678064363464e-05,
+      "loss": 1.47,
+      "step": 252000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.136528336488405e-05,
+      "loss": 1.4566,
+      "step": 252100
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.135788866540464e-05,
+      "loss": 1.455,
+      "step": 252200
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.135049396592523e-05,
+      "loss": 1.4718,
+      "step": 252300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.134309926644581e-05,
+      "loss": 1.4804,
+      "step": 252400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.13357045669664e-05,
+      "loss": 1.4586,
+      "step": 252500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132830986748699e-05,
+      "loss": 1.4705,
+      "step": 252600
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132091516800758e-05,
+      "loss": 1.4189,
+      "step": 252700
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.131352046852816e-05,
+      "loss": 1.4423,
+      "step": 252800
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.130612576904875e-05,
+      "loss": 1.4676,
+      "step": 252900
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1298731069569334e-05,
+      "loss": 1.4623,
+      "step": 253000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.129133637008992e-05,
+      "loss": 1.4569,
+      "step": 253100
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1283941670610505e-05,
+      "loss": 1.4338,
+      "step": 253200
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.127654697113109e-05,
+      "loss": 1.4857,
+      "step": 253300
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126915227165168e-05,
+      "loss": 1.4386,
+      "step": 253400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.126175757217227e-05,
+      "loss": 1.4579,
+      "step": 253500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1254362872692854e-05,
+      "loss": 1.4322,
+      "step": 253600
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.124696817321344e-05,
+      "loss": 1.466,
+      "step": 253700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.123957347373403e-05,
+      "loss": 1.4423,
+      "step": 253800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.123217877425462e-05,
+      "loss": 1.4366,
+      "step": 253900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1224784074775204e-05,
+      "loss": 1.4432,
+      "step": 254000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.121738937529579e-05,
+      "loss": 1.47,
+      "step": 254100
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.121006862281117e-05,
+      "loss": 1.4665,
+      "step": 254200
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.120267392333176e-05,
+      "loss": 1.4624,
+      "step": 254300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.119527922385234e-05,
+      "loss": 1.4807,
+      "step": 254400
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.118788452437293e-05,
+      "loss": 1.4502,
+      "step": 254500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.118048982489352e-05,
+      "loss": 1.4431,
+      "step": 254600
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1173095125414105e-05,
+      "loss": 1.424,
+      "step": 254700
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.116570042593469e-05,
+      "loss": 1.4472,
+      "step": 254800
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1158305726455276e-05,
+      "loss": 1.4356,
+      "step": 254900
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.115091102697587e-05,
+      "loss": 1.4435,
+      "step": 255000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1143516327496454e-05,
+      "loss": 1.4551,
+      "step": 255100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.113612162801704e-05,
+      "loss": 1.4553,
+      "step": 255200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1128726928537625e-05,
+      "loss": 1.4468,
+      "step": 255300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.112133222905821e-05,
+      "loss": 1.4626,
+      "step": 255400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1113937529578803e-05,
+      "loss": 1.4547,
+      "step": 255500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.110654283009939e-05,
+      "loss": 1.433,
+      "step": 255600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.109914813061997e-05,
+      "loss": 1.4489,
+      "step": 255700
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.109175343114056e-05,
+      "loss": 1.4201,
+      "step": 255800
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1084358731661146e-05,
+      "loss": 1.4341,
+      "step": 255900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.107696403218173e-05,
+      "loss": 1.4415,
+      "step": 256000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.106956933270232e-05,
+      "loss": 1.4422,
+      "step": 256100
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.106217463322291e-05,
+      "loss": 1.4599,
+      "step": 256200
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.1054779933743495e-05,
+      "loss": 1.4481,
+      "step": 256300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.104738523426408e-05,
+      "loss": 1.4249,
+      "step": 256400
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.104006448177946e-05,
+      "loss": 1.4642,
+      "step": 256500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.103266978230005e-05,
+      "loss": 1.4423,
+      "step": 256600
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.102527508282064e-05,
+      "loss": 1.4772,
+      "step": 256700
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1017880383341225e-05,
+      "loss": 1.4786,
+      "step": 256800
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.101048568386181e-05,
+      "loss": 1.4647,
+      "step": 256900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1003090984382396e-05,
+      "loss": 1.4688,
+      "step": 257000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.099569628490298e-05,
+      "loss": 1.4537,
+      "step": 257100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.098830158542357e-05,
+      "loss": 1.4441,
+      "step": 257200
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.098090688594415e-05,
+      "loss": 1.4517,
+      "step": 257300
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.0973512186464746e-05,
+      "loss": 1.4403,
+      "step": 257400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.096611748698533e-05,
+      "loss": 1.4369,
+      "step": 257500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.095872278750592e-05,
+      "loss": 1.4753,
+      "step": 257600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.09513280880265e-05,
+      "loss": 1.4565,
+      "step": 257700
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.094393338854709e-05,
+      "loss": 1.4734,
+      "step": 257800
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.093653868906768e-05,
+      "loss": 1.4578,
+      "step": 257900
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0929143989588266e-05,
+      "loss": 1.4291,
+      "step": 258000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.092174929010885e-05,
+      "loss": 1.4683,
+      "step": 258100
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091435459062944e-05,
+      "loss": 1.4355,
+      "step": 258200
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.090695989115002e-05,
+      "loss": 1.4812,
+      "step": 258300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.089956519167061e-05,
+      "loss": 1.4621,
+      "step": 258400
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.0892170492191195e-05,
+      "loss": 1.4205,
+      "step": 258500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.088477579271179e-05,
+      "loss": 1.4575,
+      "step": 258600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.087738109323237e-05,
+      "loss": 1.4566,
+      "step": 258700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.086998639375296e-05,
+      "loss": 1.4783,
+      "step": 258800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0862591694273544e-05,
+      "loss": 1.4467,
+      "step": 258900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0855196994794136e-05,
+      "loss": 1.4777,
+      "step": 259000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.084780229531472e-05,
+      "loss": 1.4653,
+      "step": 259100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.08404815428301e-05,
+      "loss": 1.436,
+      "step": 259200
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.083308684335069e-05,
+      "loss": 1.4384,
+      "step": 259300
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0825692143871274e-05,
+      "loss": 1.4586,
+      "step": 259400
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.0818297444391866e-05,
+      "loss": 1.4328,
+      "step": 259500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.081090274491245e-05,
+      "loss": 1.4634,
+      "step": 259600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.080350804543303e-05,
+      "loss": 1.4164,
+      "step": 259700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.079611334595362e-05,
+      "loss": 1.4378,
+      "step": 259800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.078871864647421e-05,
+      "loss": 1.4221,
+      "step": 259900
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0781323946994794e-05,
+      "loss": 1.4478,
+      "step": 260000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.077392924751538e-05,
+      "loss": 1.4634,
+      "step": 260100
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.076653454803597e-05,
+      "loss": 1.4531,
+      "step": 260200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.075913984855656e-05,
+      "loss": 1.4501,
+      "step": 260300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0751745149077144e-05,
+      "loss": 1.46,
+      "step": 260400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.074435044959773e-05,
+      "loss": 1.443,
+      "step": 260500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0736955750118315e-05,
+      "loss": 1.4446,
+      "step": 260600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072956105063891e-05,
+      "loss": 1.4435,
+      "step": 260700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.072216635115949e-05,
+      "loss": 1.453,
+      "step": 260800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.071477165168008e-05,
+      "loss": 1.4593,
+      "step": 260900
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0707376952200664e-05,
+      "loss": 1.4824,
+      "step": 261000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.069998225272125e-05,
+      "loss": 1.444,
+      "step": 261100
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0692587553241835e-05,
+      "loss": 1.4434,
+      "step": 261200
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.068519285376242e-05,
+      "loss": 1.4484,
+      "step": 261300
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.06778721012778e-05,
+      "loss": 1.4546,
+      "step": 261400
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0670477401798394e-05,
+      "loss": 1.4435,
+      "step": 261500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.066308270231898e-05,
+      "loss": 1.4472,
+      "step": 261600
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0655688002839565e-05,
+      "loss": 1.4501,
+      "step": 261700
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.064829330336015e-05,
+      "loss": 1.4376,
+      "step": 261800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.064089860388074e-05,
+      "loss": 1.4524,
+      "step": 261900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.063350390440133e-05,
+      "loss": 1.474,
+      "step": 262000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0626109204921915e-05,
+      "loss": 1.4342,
+      "step": 262100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.06187145054425e-05,
+      "loss": 1.4517,
+      "step": 262200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.0611319805963086e-05,
+      "loss": 1.4701,
+      "step": 262300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.060392510648367e-05,
+      "loss": 1.4433,
+      "step": 262400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.059653040700426e-05,
+      "loss": 1.4313,
+      "step": 262500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.058913570752485e-05,
+      "loss": 1.462,
+      "step": 262600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0581741008045435e-05,
+      "loss": 1.4654,
+      "step": 262700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.057434630856602e-05,
+      "loss": 1.437,
+      "step": 262800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0566951609086606e-05,
+      "loss": 1.4649,
+      "step": 262900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.055955690960719e-05,
+      "loss": 1.4598,
+      "step": 263000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0552162210127784e-05,
+      "loss": 1.4531,
+      "step": 263100
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.054476751064837e-05,
+      "loss": 1.4248,
+      "step": 263200
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0537372811168956e-05,
+      "loss": 1.4671,
+      "step": 263300
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0530052058684336e-05,
+      "loss": 1.4596,
+      "step": 263400
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.052265735920493e-05,
+      "loss": 1.4571,
+      "step": 263500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0515262659725507e-05,
+      "loss": 1.449,
+      "step": 263600
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0507867960246097e-05,
+      "loss": 1.4454,
+      "step": 263700
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0500473260766682e-05,
+      "loss": 1.4522,
+      "step": 263800
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0493078561287268e-05,
+      "loss": 1.4814,
+      "step": 263900
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0485683861807857e-05,
+      "loss": 1.4452,
+      "step": 264000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0478289162328442e-05,
+      "loss": 1.4403,
+      "step": 264100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.047089446284903e-05,
+      "loss": 1.4635,
+      "step": 264200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0463499763369617e-05,
+      "loss": 1.4392,
+      "step": 264300
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0456105063890206e-05,
+      "loss": 1.4775,
+      "step": 264400
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0448710364410792e-05,
+      "loss": 1.4486,
+      "step": 264500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.044131566493138e-05,
+      "loss": 1.4507,
+      "step": 264600
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0433920965451966e-05,
+      "loss": 1.4551,
+      "step": 264700
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0426526265972555e-05,
+      "loss": 1.4425,
+      "step": 264800
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0419131566493138e-05,
+      "loss": 1.4591,
+      "step": 264900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0411736867013723e-05,
+      "loss": 1.4626,
+      "step": 265000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0404342167534312e-05,
+      "loss": 1.439,
+      "step": 265100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0396947468054898e-05,
+      "loss": 1.4599,
+      "step": 265200
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.0389552768575487e-05,
+      "loss": 1.4376,
+      "step": 265300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0382158069096073e-05,
+      "loss": 1.4509,
+      "step": 265400
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0374837316611453e-05,
+      "loss": 1.4408,
+      "step": 265500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0367442617132042e-05,
+      "loss": 1.4709,
+      "step": 265600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0360047917652628e-05,
+      "loss": 1.4705,
+      "step": 265700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0352653218173217e-05,
+      "loss": 1.4483,
+      "step": 265800
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0345258518693802e-05,
+      "loss": 1.4379,
+      "step": 265900
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.033786381921439e-05,
+      "loss": 1.4769,
+      "step": 266000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0330469119734977e-05,
+      "loss": 1.4412,
+      "step": 266100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0323074420255566e-05,
+      "loss": 1.4616,
+      "step": 266200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.031567972077615e-05,
+      "loss": 1.4841,
+      "step": 266300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0308285021296734e-05,
+      "loss": 1.439,
+      "step": 266400
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.030089032181732e-05,
+      "loss": 1.424,
+      "step": 266500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.029349562233791e-05,
+      "loss": 1.4557,
+      "step": 266600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0286100922858494e-05,
+      "loss": 1.4481,
+      "step": 266700
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0278706223379083e-05,
+      "loss": 1.4483,
+      "step": 266800
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.027131152389967e-05,
+      "loss": 1.4332,
+      "step": 266900
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0263916824420258e-05,
+      "loss": 1.4615,
+      "step": 267000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0256522124940844e-05,
+      "loss": 1.4224,
+      "step": 267100
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0249127425461433e-05,
+      "loss": 1.4723,
+      "step": 267200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.024173272598202e-05,
+      "loss": 1.467,
+      "step": 267300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0234338026502607e-05,
+      "loss": 1.4352,
+      "step": 267400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0226943327023193e-05,
+      "loss": 1.4885,
+      "step": 267500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0219548627543775e-05,
+      "loss": 1.4652,
+      "step": 267600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0212153928064364e-05,
+      "loss": 1.4473,
+      "step": 267700
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.020475922858495e-05,
+      "loss": 1.4456,
+      "step": 267800
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.019736452910554e-05,
+      "loss": 1.4611,
+      "step": 267900
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0189969829626125e-05,
+      "loss": 1.4293,
+      "step": 268000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0182575130146714e-05,
+      "loss": 1.4638,
+      "step": 268100
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.01751804306673e-05,
+      "loss": 1.4595,
+      "step": 268200
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0167785731187885e-05,
+      "loss": 1.4483,
+      "step": 268300
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0160391031708474e-05,
+      "loss": 1.4867,
+      "step": 268400
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.015299633222906e-05,
+      "loss": 1.4749,
+      "step": 268500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.014560163274965e-05,
+      "loss": 1.4623,
+      "step": 268600
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.013828088026503e-05,
+      "loss": 1.441,
+      "step": 268700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0130886180785618e-05,
+      "loss": 1.4679,
+      "step": 268800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0123491481306197e-05,
+      "loss": 1.453,
+      "step": 268900
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0116096781826786e-05,
+      "loss": 1.4646,
+      "step": 269000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.010870208234737e-05,
+      "loss": 1.4265,
+      "step": 269100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.010130738286796e-05,
+      "loss": 1.4553,
+      "step": 269200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0093912683388546e-05,
+      "loss": 1.4309,
+      "step": 269300
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0086517983909135e-05,
+      "loss": 1.4472,
+      "step": 269400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007912328442972e-05,
+      "loss": 1.4453,
+      "step": 269500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.007172858495031e-05,
+      "loss": 1.4412,
+      "step": 269600
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0064333885470896e-05,
+      "loss": 1.4683,
+      "step": 269700
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0056939185991485e-05,
+      "loss": 1.4261,
+      "step": 269800
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.004954448651207e-05,
+      "loss": 1.4568,
+      "step": 269900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.004214978703266e-05,
+      "loss": 1.4615,
+      "step": 270000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0034755087553245e-05,
+      "loss": 1.4484,
+      "step": 270100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0027360388073827e-05,
+      "loss": 1.4745,
+      "step": 270200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0019965688594416e-05,
+      "loss": 1.4542,
+      "step": 270300
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0012570989115002e-05,
+      "loss": 1.444,
+      "step": 270400
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.000517628963559e-05,
+      "loss": 1.4219,
+      "step": 270500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9997781590156176e-05,
+      "loss": 1.3698,
+      "step": 270600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9990386890676762e-05,
+      "loss": 1.3848,
+      "step": 270700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.998299219119735e-05,
+      "loss": 1.4004,
+      "step": 270800
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9975597491717937e-05,
+      "loss": 1.3691,
+      "step": 270900
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9968202792238526e-05,
+      "loss": 1.4072,
+      "step": 271000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9960882039753906e-05,
+      "loss": 1.3749,
+      "step": 271100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.9953487340274495e-05,
+      "loss": 1.387,
+      "step": 271200
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.994609264079508e-05,
+      "loss": 1.3935,
+      "step": 271300
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.993869794131567e-05,
+      "loss": 1.3942,
+      "step": 271400
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.993130324183625e-05,
+      "loss": 1.3825,
+      "step": 271500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9923908542356838e-05,
+      "loss": 1.3686,
+      "step": 271600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9916513842877423e-05,
+      "loss": 1.3966,
+      "step": 271700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9909119143398012e-05,
+      "loss": 1.394,
+      "step": 271800
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9901724443918598e-05,
+      "loss": 1.3981,
+      "step": 271900
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9894329744439187e-05,
+      "loss": 1.3897,
+      "step": 272000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.9886935044959773e-05,
+      "loss": 1.4052,
+      "step": 272100
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9879540345480362e-05,
+      "loss": 1.4164,
+      "step": 272200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9872145646000947e-05,
+      "loss": 1.3779,
+      "step": 272300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9864750946521536e-05,
+      "loss": 1.392,
+      "step": 272400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9857356247042122e-05,
+      "loss": 1.3841,
+      "step": 272500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.984996154756271e-05,
+      "loss": 1.4046,
+      "step": 272600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9842566848083297e-05,
+      "loss": 1.3781,
+      "step": 272700
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.9835172148603886e-05,
+      "loss": 1.3891,
+      "step": 272800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9827777449124468e-05,
+      "loss": 1.3916,
+      "step": 272900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9820382749645054e-05,
+      "loss": 1.3962,
+      "step": 273000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9812988050165643e-05,
+      "loss": 1.3882,
+      "step": 273100
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.980559335068623e-05,
+      "loss": 1.3682,
+      "step": 273200
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9798198651206814e-05,
+      "loss": 1.3904,
+      "step": 273300
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9790803951727403e-05,
+      "loss": 1.3874,
+      "step": 273400
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.978340925224799e-05,
+      "loss": 1.3798,
+      "step": 273500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9776014552768578e-05,
+      "loss": 1.3843,
+      "step": 273600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9768619853289163e-05,
+      "loss": 1.3884,
+      "step": 273700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9761225153809752e-05,
+      "loss": 1.348,
+      "step": 273800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9753830454330338e-05,
+      "loss": 1.4056,
+      "step": 273900
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9746435754850927e-05,
+      "loss": 1.3648,
+      "step": 274000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9739041055371513e-05,
+      "loss": 1.43,
+      "step": 274100
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.973172030288689e-05,
+      "loss": 1.3886,
+      "step": 274200
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9724325603407475e-05,
+      "loss": 1.3801,
+      "step": 274300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9716930903928064e-05,
+      "loss": 1.3922,
+      "step": 274400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.970953620444865e-05,
+      "loss": 1.4062,
+      "step": 274500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.970214150496924e-05,
+      "loss": 1.402,
+      "step": 274600
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9694746805489825e-05,
+      "loss": 1.39,
+      "step": 274700
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9687352106010414e-05,
+      "loss": 1.3792,
+      "step": 274800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9679957406531e-05,
+      "loss": 1.3935,
+      "step": 274900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.967256270705159e-05,
+      "loss": 1.3846,
+      "step": 275000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9665168007572174e-05,
+      "loss": 1.3881,
+      "step": 275100
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9657773308092763e-05,
+      "loss": 1.4005,
+      "step": 275200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.965037860861335e-05,
+      "loss": 1.3965,
+      "step": 275300
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9642983909133938e-05,
+      "loss": 1.4019,
+      "step": 275400
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.963558920965452e-05,
+      "loss": 1.4251,
+      "step": 275500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9628194510175106e-05,
+      "loss": 1.3955,
+      "step": 275600
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9620799810695695e-05,
+      "loss": 1.3832,
+      "step": 275700
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.961340511121628e-05,
+      "loss": 1.4405,
+      "step": 275800
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9606010411736866e-05,
+      "loss": 1.3938,
+      "step": 275900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9598615712257455e-05,
+      "loss": 1.386,
+      "step": 276000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.959122101277804e-05,
+      "loss": 1.4083,
+      "step": 276100
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.958382631329863e-05,
+      "loss": 1.4335,
+      "step": 276200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9576431613819215e-05,
+      "loss": 1.3892,
+      "step": 276300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9569036914339804e-05,
+      "loss": 1.4128,
+      "step": 276400
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.956164221486039e-05,
+      "loss": 1.4321,
+      "step": 276500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9554321462375774e-05,
+      "loss": 1.3953,
+      "step": 276600
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.954692676289636e-05,
+      "loss": 1.3843,
+      "step": 276700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.953953206341694e-05,
+      "loss": 1.3799,
+      "step": 276800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9532137363937527e-05,
+      "loss": 1.3716,
+      "step": 276900
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9524742664458116e-05,
+      "loss": 1.4048,
+      "step": 277000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9517347964978702e-05,
+      "loss": 1.3871,
+      "step": 277100
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.950995326549929e-05,
+      "loss": 1.4006,
+      "step": 277200
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9502558566019877e-05,
+      "loss": 1.4159,
+      "step": 277300
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9495163866540466e-05,
+      "loss": 1.4032,
+      "step": 277400
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.948776916706105e-05,
+      "loss": 1.4016,
+      "step": 277500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.948037446758164e-05,
+      "loss": 1.4114,
+      "step": 277600
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9472979768102226e-05,
+      "loss": 1.3942,
+      "step": 277700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9465585068622815e-05,
+      "loss": 1.3795,
+      "step": 277800
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.94581903691434e-05,
+      "loss": 1.3886,
+      "step": 277900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.945079566966399e-05,
+      "loss": 1.3886,
+      "step": 278000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9443400970184575e-05,
+      "loss": 1.3941,
+      "step": 278100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.9436006270705157e-05,
+      "loss": 1.4028,
+      "step": 278200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9428611571225746e-05,
+      "loss": 1.3757,
+      "step": 278300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9421216871746332e-05,
+      "loss": 1.4051,
+      "step": 278400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9413822172266918e-05,
+      "loss": 1.4317,
+      "step": 278500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9406427472787507e-05,
+      "loss": 1.3936,
+      "step": 278600
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9399106720302887e-05,
+      "loss": 1.3885,
+      "step": 278700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9391712020823476e-05,
+      "loss": 1.4096,
+      "step": 278800
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9384317321344062e-05,
+      "loss": 1.4038,
+      "step": 278900
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.937692262186465e-05,
+      "loss": 1.3699,
+      "step": 279000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9369527922385237e-05,
+      "loss": 1.3965,
+      "step": 279100
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9362133222905826e-05,
+      "loss": 1.42,
+      "step": 279200
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.935473852342641e-05,
+      "loss": 1.4035,
+      "step": 279300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9347343823947e-05,
+      "loss": 1.3799,
+      "step": 279400
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.933994912446758e-05,
+      "loss": 1.4015,
+      "step": 279500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9332554424988168e-05,
+      "loss": 1.3781,
+      "step": 279600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9325159725508754e-05,
+      "loss": 1.4084,
+      "step": 279700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9317765026029343e-05,
+      "loss": 1.4066,
+      "step": 279800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.931037032654993e-05,
+      "loss": 1.4019,
+      "step": 279900
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9302975627070517e-05,
+      "loss": 1.3928,
+      "step": 280000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9295580927591103e-05,
+      "loss": 1.3932,
+      "step": 280100
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.9288186228111692e-05,
+      "loss": 1.3943,
+      "step": 280200
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9280791528632278e-05,
+      "loss": 1.3998,
+      "step": 280300
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9273396829152867e-05,
+      "loss": 1.4403,
+      "step": 280400
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9266002129673452e-05,
+      "loss": 1.4033,
+      "step": 280500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.925860743019404e-05,
+      "loss": 1.3885,
+      "step": 280600
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9251212730714627e-05,
+      "loss": 1.3953,
+      "step": 280700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.924381803123521e-05,
+      "loss": 1.3653,
+      "step": 280800
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.92364233317558e-05,
+      "loss": 1.3739,
+      "step": 280900
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9229028632276384e-05,
+      "loss": 1.4124,
+      "step": 281000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9221707879791764e-05,
+      "loss": 1.3835,
+      "step": 281100
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9214313180312353e-05,
+      "loss": 1.4057,
+      "step": 281200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.920691848083294e-05,
+      "loss": 1.3904,
+      "step": 281300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9199523781353528e-05,
+      "loss": 1.3636,
+      "step": 281400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9192129081874114e-05,
+      "loss": 1.3902,
+      "step": 281500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9184734382394703e-05,
+      "loss": 1.3879,
+      "step": 281600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.917733968291529e-05,
+      "loss": 1.4237,
+      "step": 281700
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9169944983435877e-05,
+      "loss": 1.4052,
+      "step": 281800
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9162550283956463e-05,
+      "loss": 1.4376,
+      "step": 281900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9155155584477052e-05,
+      "loss": 1.3966,
+      "step": 282000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.914776088499763e-05,
+      "loss": 1.3835,
+      "step": 282100
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.914036618551822e-05,
+      "loss": 1.3788,
+      "step": 282200
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9132971486038806e-05,
+      "loss": 1.4168,
+      "step": 282300
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9125576786559395e-05,
+      "loss": 1.4064,
+      "step": 282400
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911818208707998e-05,
+      "loss": 1.3886,
+      "step": 282500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.911078738760057e-05,
+      "loss": 1.4049,
+      "step": 282600
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9103392688121155e-05,
+      "loss": 1.4418,
+      "step": 282700
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9095997988641744e-05,
+      "loss": 1.3843,
+      "step": 282800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.908860328916233e-05,
+      "loss": 1.3907,
+      "step": 282900
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.908120858968292e-05,
+      "loss": 1.4314,
+      "step": 283000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9073813890203504e-05,
+      "loss": 1.4178,
+      "step": 283100
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9066419190724093e-05,
+      "loss": 1.4006,
+      "step": 283200
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9059098438239474e-05,
+      "loss": 1.3997,
+      "step": 283300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9051703738760056e-05,
+      "loss": 1.4285,
+      "step": 283400
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.904430903928064e-05,
+      "loss": 1.3899,
+      "step": 283500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.903691433980123e-05,
+      "loss": 1.3918,
+      "step": 283600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9029519640321816e-05,
+      "loss": 1.3865,
+      "step": 283700
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.9022124940842405e-05,
+      "loss": 1.3726,
+      "step": 283800
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.901473024136299e-05,
+      "loss": 1.4013,
+      "step": 283900
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.900733554188358e-05,
+      "loss": 1.3871,
+      "step": 284000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.8999940842404166e-05,
+      "loss": 1.4047,
+      "step": 284100
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.8992546142924755e-05,
+      "loss": 1.3992,
+      "step": 284200
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.898515144344534e-05,
+      "loss": 1.3815,
+      "step": 284300
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.897775674396593e-05,
+      "loss": 1.4037,
+      "step": 284400
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8970362044486515e-05,
+      "loss": 1.4115,
+      "step": 284500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8962967345007104e-05,
+      "loss": 1.3929,
+      "step": 284600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.895557264552769e-05,
+      "loss": 1.4171,
+      "step": 284700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8948177946048272e-05,
+      "loss": 1.3957,
+      "step": 284800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8940783246568858e-05,
+      "loss": 1.3951,
+      "step": 284900
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8933388547089447e-05,
+      "loss": 1.3868,
+      "step": 285000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8925993847610032e-05,
+      "loss": 1.401,
+      "step": 285100
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.891859914813062e-05,
+      "loss": 1.39,
+      "step": 285200
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8911204448651207e-05,
+      "loss": 1.4027,
+      "step": 285300
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.8903809749171796e-05,
+      "loss": 1.3884,
+      "step": 285400
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.889641504969238e-05,
+      "loss": 1.3792,
+      "step": 285500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.888902035021297e-05,
+      "loss": 1.4087,
+      "step": 285600
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.888169959772835e-05,
+      "loss": 1.4136,
+      "step": 285700
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.887430489824894e-05,
+      "loss": 1.3675,
+      "step": 285800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8866910198769526e-05,
+      "loss": 1.4047,
+      "step": 285900
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.885951549929011e-05,
+      "loss": 1.3875,
+      "step": 286000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8852120799810694e-05,
+      "loss": 1.3916,
+      "step": 286100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8844726100331283e-05,
+      "loss": 1.3959,
+      "step": 286200
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.8837331400851868e-05,
+      "loss": 1.406,
+      "step": 286300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8829936701372457e-05,
+      "loss": 1.4104,
+      "step": 286400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8822542001893043e-05,
+      "loss": 1.4054,
+      "step": 286500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8815147302413632e-05,
+      "loss": 1.3742,
+      "step": 286600
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8807752602934218e-05,
+      "loss": 1.3788,
+      "step": 286700
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8800357903454807e-05,
+      "loss": 1.4198,
+      "step": 286800
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.8792963203975392e-05,
+      "loss": 1.4164,
+      "step": 286900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.878556850449598e-05,
+      "loss": 1.3812,
+      "step": 287000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8778173805016567e-05,
+      "loss": 1.4123,
+      "step": 287100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8770779105537156e-05,
+      "loss": 1.4089,
+      "step": 287200
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876338440605774e-05,
+      "loss": 1.4031,
+      "step": 287300
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8755989706578324e-05,
+      "loss": 1.3679,
+      "step": 287400
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.874859500709891e-05,
+      "loss": 1.395,
+      "step": 287500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.87412003076195e-05,
+      "loss": 1.3919,
+      "step": 287600
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8733805608140084e-05,
+      "loss": 1.4031,
+      "step": 287700
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8726410908660673e-05,
+      "loss": 1.3972,
+      "step": 287800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.871901620918126e-05,
+      "loss": 1.3781,
+      "step": 287900
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8711695456696643e-05,
+      "loss": 1.3883,
+      "step": 288000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8704300757217228e-05,
+      "loss": 1.3918,
+      "step": 288100
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8696906057737817e-05,
+      "loss": 1.3804,
+      "step": 288200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8689511358258403e-05,
+      "loss": 1.4147,
+      "step": 288300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8682116658778992e-05,
+      "loss": 1.4315,
+      "step": 288400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8674721959299578e-05,
+      "loss": 1.3891,
+      "step": 288500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8667327259820163e-05,
+      "loss": 1.391,
+      "step": 288600
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8659932560340745e-05,
+      "loss": 1.3827,
+      "step": 288700
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8652537860861334e-05,
+      "loss": 1.3936,
+      "step": 288800
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.864514316138192e-05,
+      "loss": 1.39,
+      "step": 288900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.863774846190251e-05,
+      "loss": 1.3901,
+      "step": 289000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8630353762423095e-05,
+      "loss": 1.4108,
+      "step": 289100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8622959062943684e-05,
+      "loss": 1.4048,
+      "step": 289200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.861556436346427e-05,
+      "loss": 1.4003,
+      "step": 289300
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.860816966398486e-05,
+      "loss": 1.3991,
+      "step": 289400
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8600774964505444e-05,
+      "loss": 1.374,
+      "step": 289500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8593380265026033e-05,
+      "loss": 1.4193,
+      "step": 289600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.858598556554662e-05,
+      "loss": 1.4239,
+      "step": 289700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8578590866067208e-05,
+      "loss": 1.4035,
+      "step": 289800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8571196166587793e-05,
+      "loss": 1.4181,
+      "step": 289900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8563801467108376e-05,
+      "loss": 1.4156,
+      "step": 290000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8556480714623756e-05,
+      "loss": 1.4212,
+      "step": 290100
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8549086015144345e-05,
+      "loss": 1.3951,
+      "step": 290200
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.854169131566493e-05,
+      "loss": 1.3955,
+      "step": 290300
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.853429661618552e-05,
+      "loss": 1.3681,
+      "step": 290400
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8526901916706105e-05,
+      "loss": 1.4056,
+      "step": 290500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8519507217226694e-05,
+      "loss": 1.3818,
+      "step": 290600
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.851211251774728e-05,
+      "loss": 1.3988,
+      "step": 290700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.850471781826787e-05,
+      "loss": 1.3988,
+      "step": 290800
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8497323118788455e-05,
+      "loss": 1.4182,
+      "step": 290900
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.848992841930904e-05,
+      "loss": 1.4102,
+      "step": 291000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.848253371982963e-05,
+      "loss": 1.4322,
+      "step": 291100
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8475139020350215e-05,
+      "loss": 1.4221,
+      "step": 291200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8467744320870804e-05,
+      "loss": 1.4201,
+      "step": 291300
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8460349621391386e-05,
+      "loss": 1.4131,
+      "step": 291400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8452954921911972e-05,
+      "loss": 1.4188,
+      "step": 291500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.844556022243256e-05,
+      "loss": 1.3829,
+      "step": 291600
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8438165522953147e-05,
+      "loss": 1.4186,
+      "step": 291700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8430770823473736e-05,
+      "loss": 1.3747,
+      "step": 291800
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.842337612399432e-05,
+      "loss": 1.3808,
+      "step": 291900
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.841598142451491e-05,
+      "loss": 1.3875,
+      "step": 292000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.840866067203029e-05,
+      "loss": 1.4022,
+      "step": 292100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8401265972550876e-05,
+      "loss": 1.4042,
+      "step": 292200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8393871273071465e-05,
+      "loss": 1.4041,
+      "step": 292300
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.838647657359205e-05,
+      "loss": 1.38,
+      "step": 292400
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.837908187411264e-05,
+      "loss": 1.3741,
+      "step": 292500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8371687174633226e-05,
+      "loss": 1.4029,
+      "step": 292600
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8364292475153808e-05,
+      "loss": 1.382,
+      "step": 292700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8356897775674397e-05,
+      "loss": 1.4097,
+      "step": 292800
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8349503076194983e-05,
+      "loss": 1.3897,
+      "step": 292900
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8342108376715572e-05,
+      "loss": 1.3815,
+      "step": 293000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.8334713677236157e-05,
+      "loss": 1.4094,
+      "step": 293100
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8327318977756746e-05,
+      "loss": 1.3908,
+      "step": 293200
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8319924278277332e-05,
+      "loss": 1.4051,
+      "step": 293300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.831252957879792e-05,
+      "loss": 1.3974,
+      "step": 293400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8305134879318507e-05,
+      "loss": 1.4135,
+      "step": 293500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8297740179839092e-05,
+      "loss": 1.3962,
+      "step": 293600
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.829034548035968e-05,
+      "loss": 1.432,
+      "step": 293700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8282950780880267e-05,
+      "loss": 1.3993,
+      "step": 293800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8275556081400856e-05,
+      "loss": 1.399,
+      "step": 293900
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8268161381921438e-05,
+      "loss": 1.3919,
+      "step": 294000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8260766682442024e-05,
+      "loss": 1.397,
+      "step": 294100
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8253445929957408e-05,
+      "loss": 1.4321,
+      "step": 294200
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8246051230477993e-05,
+      "loss": 1.3729,
+      "step": 294300
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.823865653099858e-05,
+      "loss": 1.3853,
+      "step": 294400
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8231261831519168e-05,
+      "loss": 1.396,
+      "step": 294500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8223867132039754e-05,
+      "loss": 1.3994,
+      "step": 294600
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8216472432560343e-05,
+      "loss": 1.4032,
+      "step": 294700
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.820907773308093e-05,
+      "loss": 1.4005,
+      "step": 294800
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8201683033601517e-05,
+      "loss": 1.4302,
+      "step": 294900
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8194288334122103e-05,
+      "loss": 1.3921,
+      "step": 295000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8186893634642692e-05,
+      "loss": 1.4094,
+      "step": 295100
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8179498935163278e-05,
+      "loss": 1.3973,
+      "step": 295200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.817210423568386e-05,
+      "loss": 1.4016,
+      "step": 295300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.816470953620445e-05,
+      "loss": 1.4024,
+      "step": 295400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8157314836725035e-05,
+      "loss": 1.377,
+      "step": 295500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8149920137245624e-05,
+      "loss": 1.404,
+      "step": 295600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.814252543776621e-05,
+      "loss": 1.3917,
+      "step": 295700
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.8135130738286798e-05,
+      "loss": 1.3965,
+      "step": 295800
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8127736038807384e-05,
+      "loss": 1.379,
+      "step": 295900
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8120341339327973e-05,
+      "loss": 1.4034,
+      "step": 296000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.811294663984856e-05,
+      "loss": 1.4126,
+      "step": 296100
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8105551940369144e-05,
+      "loss": 1.4348,
+      "step": 296200
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8098231187884528e-05,
+      "loss": 1.398,
+      "step": 296300
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8090836488405114e-05,
+      "loss": 1.3766,
+      "step": 296400
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8083441788925703e-05,
+      "loss": 1.4034,
+      "step": 296500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8076047089446285e-05,
+      "loss": 1.4128,
+      "step": 296600
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.806865238996687e-05,
+      "loss": 1.3915,
+      "step": 296700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.806125769048746e-05,
+      "loss": 1.4104,
+      "step": 296800
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8053862991008045e-05,
+      "loss": 1.3745,
+      "step": 296900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.804646829152863e-05,
+      "loss": 1.4287,
+      "step": 297000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.803907359204922e-05,
+      "loss": 1.4033,
+      "step": 297100
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8031678892569806e-05,
+      "loss": 1.3933,
+      "step": 297200
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8024284193090395e-05,
+      "loss": 1.411,
+      "step": 297300
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.801688949361098e-05,
+      "loss": 1.3957,
+      "step": 297400
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.800949479413157e-05,
+      "loss": 1.4075,
+      "step": 297500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.8002100094652155e-05,
+      "loss": 1.396,
+      "step": 297600
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.7994705395172744e-05,
+      "loss": 1.3808,
+      "step": 297700
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.798731069569333e-05,
+      "loss": 1.3991,
+      "step": 297800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.797991599621392e-05,
+      "loss": 1.4053,
+      "step": 297900
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.79725212967345e-05,
+      "loss": 1.4183,
+      "step": 298000
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7965126597255086e-05,
+      "loss": 1.3875,
+      "step": 298100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7957731897775675e-05,
+      "loss": 1.3697,
+      "step": 298200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.795033719829626e-05,
+      "loss": 1.4007,
+      "step": 298300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.794294249881685e-05,
+      "loss": 1.4025,
+      "step": 298400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.7935547799337436e-05,
+      "loss": 1.389,
+      "step": 298500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7928153099858025e-05,
+      "loss": 1.3968,
+      "step": 298600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.792075840037861e-05,
+      "loss": 1.4114,
+      "step": 298700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7913363700899196e-05,
+      "loss": 1.4218,
+      "step": 298800
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7905969001419785e-05,
+      "loss": 1.3861,
+      "step": 298900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7898648248935166e-05,
+      "loss": 1.4197,
+      "step": 299000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.7891253549455755e-05,
+      "loss": 1.4049,
+      "step": 299100
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.788385884997634e-05,
+      "loss": 1.4299,
+      "step": 299200
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7876464150496922e-05,
+      "loss": 1.3937,
+      "step": 299300
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.786906945101751e-05,
+      "loss": 1.4192,
+      "step": 299400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7861674751538097e-05,
+      "loss": 1.3999,
+      "step": 299500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7854280052058683e-05,
+      "loss": 1.358,
+      "step": 299600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7846885352579272e-05,
+      "loss": 1.3942,
+      "step": 299700
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7839490653099857e-05,
+      "loss": 1.4014,
+      "step": 299800
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7832095953620446e-05,
+      "loss": 1.392,
+      "step": 299900
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7824701254141032e-05,
+      "loss": 1.3779,
+      "step": 300000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.781730655466162e-05,
+      "loss": 1.3978,
+      "step": 300100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7809911855182207e-05,
+      "loss": 1.3962,
+      "step": 300200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.7802517155702796e-05,
+      "loss": 1.414,
+      "step": 300300
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.779512245622338e-05,
+      "loss": 1.3898,
+      "step": 300400
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.778772775674397e-05,
+      "loss": 1.4142,
+      "step": 300500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7780333057264553e-05,
+      "loss": 1.4096,
+      "step": 300600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.777293835778514e-05,
+      "loss": 1.3696,
+      "step": 300700
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7765543658305727e-05,
+      "loss": 1.4099,
+      "step": 300800
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7758148958826313e-05,
+      "loss": 1.4091,
+      "step": 300900
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7750754259346902e-05,
+      "loss": 1.4059,
+      "step": 301000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7743433506862282e-05,
+      "loss": 1.3945,
+      "step": 301100
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.7736038807382868e-05,
+      "loss": 1.3814,
+      "step": 301200
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7728644107903457e-05,
+      "loss": 1.4168,
+      "step": 301300
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7721249408424043e-05,
+      "loss": 1.3843,
+      "step": 301400
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7713854708944632e-05,
+      "loss": 1.4102,
+      "step": 301500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7706460009465217e-05,
+      "loss": 1.3841,
+      "step": 301600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7699065309985806e-05,
+      "loss": 1.4118,
+      "step": 301700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7691670610506392e-05,
+      "loss": 1.4146,
+      "step": 301800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.7684275911026974e-05,
+      "loss": 1.4064,
+      "step": 301900
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7676881211547563e-05,
+      "loss": 1.4041,
+      "step": 302000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.766948651206815e-05,
+      "loss": 1.3948,
+      "step": 302100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7662091812588735e-05,
+      "loss": 1.4038,
+      "step": 302200
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7654697113109324e-05,
+      "loss": 1.3799,
+      "step": 302300
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.764730241362991e-05,
+      "loss": 1.3944,
+      "step": 302400
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.76399077141505e-05,
+      "loss": 1.3978,
+      "step": 302500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7632513014671084e-05,
+      "loss": 1.4021,
+      "step": 302600
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7625118315191673e-05,
+      "loss": 1.3842,
+      "step": 302700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.761772361571226e-05,
+      "loss": 1.4106,
+      "step": 302800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7610328916232848e-05,
+      "loss": 1.4069,
+      "step": 302900
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7602934216753433e-05,
+      "loss": 1.3812,
+      "step": 303000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7595539517274022e-05,
+      "loss": 1.399,
+      "step": 303100
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7588218764789403e-05,
+      "loss": 1.3976,
+      "step": 303200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7580824065309985e-05,
+      "loss": 1.3913,
+      "step": 303300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.757342936583057e-05,
+      "loss": 1.3909,
+      "step": 303400
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.756603466635116e-05,
+      "loss": 1.401,
+      "step": 303500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7558639966871745e-05,
+      "loss": 1.3997,
+      "step": 303600
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7551245267392334e-05,
+      "loss": 1.4052,
+      "step": 303700
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.754385056791292e-05,
+      "loss": 1.4019,
+      "step": 303800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.753645586843351e-05,
+      "loss": 1.4279,
+      "step": 303900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7529061168954095e-05,
+      "loss": 1.4318,
+      "step": 304000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7521666469474684e-05,
+      "loss": 1.3936,
+      "step": 304100
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.751427176999527e-05,
+      "loss": 1.4057,
+      "step": 304200
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.750687707051586e-05,
+      "loss": 1.4191,
+      "step": 304300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7499482371036444e-05,
+      "loss": 1.3874,
+      "step": 304400
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7492087671557033e-05,
+      "loss": 1.4066,
+      "step": 304500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7484692972077615e-05,
+      "loss": 1.3791,
+      "step": 304600
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.74772982725982e-05,
+      "loss": 1.4139,
+      "step": 304700
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7469903573118787e-05,
+      "loss": 1.3962,
+      "step": 304800
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7462508873639376e-05,
+      "loss": 1.4071,
+      "step": 304900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.745511417415996e-05,
+      "loss": 1.3976,
+      "step": 305000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.744771947468055e-05,
+      "loss": 1.4062,
+      "step": 305100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7440324775201136e-05,
+      "loss": 1.3743,
+      "step": 305200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7432930075721725e-05,
+      "loss": 1.3982,
+      "step": 305300
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.742553537624231e-05,
+      "loss": 1.4134,
+      "step": 305400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7418214623757694e-05,
+      "loss": 1.4157,
+      "step": 305500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.741081992427828e-05,
+      "loss": 1.3945,
+      "step": 305600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.740342522479887e-05,
+      "loss": 1.3865,
+      "step": 305700
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7396030525319455e-05,
+      "loss": 1.419,
+      "step": 305800
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7388635825840037e-05,
+      "loss": 1.3965,
+      "step": 305900
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7381241126360623e-05,
+      "loss": 1.4041,
+      "step": 306000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.737384642688121e-05,
+      "loss": 1.41,
+      "step": 306100
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7366451727401797e-05,
+      "loss": 1.4232,
+      "step": 306200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7359057027922386e-05,
+      "loss": 1.3874,
+      "step": 306300
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7351662328442972e-05,
+      "loss": 1.3956,
+      "step": 306400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.734426762896356e-05,
+      "loss": 1.3779,
+      "step": 306500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.7336872929484147e-05,
+      "loss": 1.3915,
+      "step": 306600
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7329478230004736e-05,
+      "loss": 1.4063,
+      "step": 306700
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.732208353052532e-05,
+      "loss": 1.4044,
+      "step": 306800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.731468883104591e-05,
+      "loss": 1.415,
+      "step": 306900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7307294131566496e-05,
+      "loss": 1.4124,
+      "step": 307000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7299899432087085e-05,
+      "loss": 1.406,
+      "step": 307100
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7292504732607667e-05,
+      "loss": 1.4166,
+      "step": 307200
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.7285110033128253e-05,
+      "loss": 1.4211,
+      "step": 307300
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.727771533364884e-05,
+      "loss": 1.3863,
+      "step": 307400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7270320634169427e-05,
+      "loss": 1.3903,
+      "step": 307500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7262925934690013e-05,
+      "loss": 1.3859,
+      "step": 307600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7255605182205397e-05,
+      "loss": 1.4115,
+      "step": 307700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.7248210482725983e-05,
+      "loss": 1.4098,
+      "step": 307800
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.724081578324657e-05,
+      "loss": 1.3912,
+      "step": 307900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7233421083767157e-05,
+      "loss": 1.395,
+      "step": 308000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7226026384287746e-05,
+      "loss": 1.3902,
+      "step": 308100
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7218631684808332e-05,
+      "loss": 1.4241,
+      "step": 308200
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.721123698532892e-05,
+      "loss": 1.4162,
+      "step": 308300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7203842285849507e-05,
+      "loss": 1.4169,
+      "step": 308400
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.719644758637009e-05,
+      "loss": 1.4073,
+      "step": 308500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.7189052886890674e-05,
+      "loss": 1.4012,
+      "step": 308600
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7181658187411263e-05,
+      "loss": 1.3902,
+      "step": 308700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.717426348793185e-05,
+      "loss": 1.4071,
+      "step": 308800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7166868788452438e-05,
+      "loss": 1.4028,
+      "step": 308900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7159474088973024e-05,
+      "loss": 1.4176,
+      "step": 309000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7152079389493613e-05,
+      "loss": 1.3869,
+      "step": 309100
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.71446846900142e-05,
+      "loss": 1.4362,
+      "step": 309200
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7137289990534787e-05,
+      "loss": 1.3982,
+      "step": 309300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7129895291055373e-05,
+      "loss": 1.3764,
+      "step": 309400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7122500591575962e-05,
+      "loss": 1.4363,
+      "step": 309500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7115105892096548e-05,
+      "loss": 1.397,
+      "step": 309600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7107711192617137e-05,
+      "loss": 1.3899,
+      "step": 309700
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7100316493137722e-05,
+      "loss": 1.4021,
+      "step": 309800
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.70929957406531e-05,
+      "loss": 1.4213,
+      "step": 309900
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7085601041173685e-05,
+      "loss": 1.424,
+      "step": 310000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7078206341694274e-05,
+      "loss": 1.3989,
+      "step": 310100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.707081164221486e-05,
+      "loss": 1.3979,
+      "step": 310200
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.706341694273545e-05,
+      "loss": 1.4068,
+      "step": 310300
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7056022243256034e-05,
+      "loss": 1.3994,
+      "step": 310400
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7048627543776624e-05,
+      "loss": 1.4063,
+      "step": 310500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.704123284429721e-05,
+      "loss": 1.3879,
+      "step": 310600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7033838144817798e-05,
+      "loss": 1.3846,
+      "step": 310700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7026443445338384e-05,
+      "loss": 1.4188,
+      "step": 310800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7019048745858973e-05,
+      "loss": 1.4022,
+      "step": 310900
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.701165404637956e-05,
+      "loss": 1.3922,
+      "step": 311000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7004259346900147e-05,
+      "loss": 1.4035,
+      "step": 311100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.6996864647420726e-05,
+      "loss": 1.3813,
+      "step": 311200
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.6989469947941315e-05,
+      "loss": 1.3837,
+      "step": 311300
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.69820752484619e-05,
+      "loss": 1.3998,
+      "step": 311400
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.697468054898249e-05,
+      "loss": 1.4259,
+      "step": 311500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6967285849503076e-05,
+      "loss": 1.4068,
+      "step": 311600
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6959891150023665e-05,
+      "loss": 1.3909,
+      "step": 311700
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.695249645054425e-05,
+      "loss": 1.4087,
+      "step": 311800
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.694510175106484e-05,
+      "loss": 1.4111,
+      "step": 311900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6937707051585425e-05,
+      "loss": 1.3985,
+      "step": 312000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6930386299100805e-05,
+      "loss": 1.4061,
+      "step": 312100
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6922991599621394e-05,
+      "loss": 1.4065,
+      "step": 312200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.691559690014198e-05,
+      "loss": 1.3924,
+      "step": 312300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.690820220066257e-05,
+      "loss": 1.3844,
+      "step": 312400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.690080750118315e-05,
+      "loss": 1.4003,
+      "step": 312500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6893412801703737e-05,
+      "loss": 1.4288,
+      "step": 312600
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6886018102224326e-05,
+      "loss": 1.4118,
+      "step": 312700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6878623402744912e-05,
+      "loss": 1.3864,
+      "step": 312800
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.68712287032655e-05,
+      "loss": 1.4224,
+      "step": 312900
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6863834003786086e-05,
+      "loss": 1.3975,
+      "step": 313000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6856439304306675e-05,
+      "loss": 1.3982,
+      "step": 313100
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.684904460482726e-05,
+      "loss": 1.4121,
+      "step": 313200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.684164990534785e-05,
+      "loss": 1.3935,
+      "step": 313300
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6834255205868436e-05,
+      "loss": 1.3517,
+      "step": 313400
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6826860506389025e-05,
+      "loss": 1.3912,
+      "step": 313500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.681946580690961e-05,
+      "loss": 1.3977,
+      "step": 313600
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.68120711074302e-05,
+      "loss": 1.4388,
+      "step": 313700
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6804676407950778e-05,
+      "loss": 1.4014,
+      "step": 313800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6797281708471367e-05,
+      "loss": 1.3796,
+      "step": 313900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6789887008991953e-05,
+      "loss": 1.3977,
+      "step": 314000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6782492309512542e-05,
+      "loss": 1.4009,
+      "step": 314100
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6775171557027922e-05,
+      "loss": 1.3908,
+      "step": 314200
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.676777685754851e-05,
+      "loss": 1.4075,
+      "step": 314300
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6760382158069097e-05,
+      "loss": 1.4018,
+      "step": 314400
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6752987458589686e-05,
+      "loss": 1.4072,
+      "step": 314500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6745592759110272e-05,
+      "loss": 1.4175,
+      "step": 314600
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6738198059630857e-05,
+      "loss": 1.4097,
+      "step": 314700
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6730803360151446e-05,
+      "loss": 1.4168,
+      "step": 314800
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6723408660672032e-05,
+      "loss": 1.3969,
+      "step": 314900
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.671601396119262e-05,
+      "loss": 1.389,
+      "step": 315000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6708619261713203e-05,
+      "loss": 1.4078,
+      "step": 315100
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.670122456223379e-05,
+      "loss": 1.4054,
+      "step": 315200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6693829862754378e-05,
+      "loss": 1.4086,
+      "step": 315300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6686435163274964e-05,
+      "loss": 1.4127,
+      "step": 315400
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6679040463795553e-05,
+      "loss": 1.4049,
+      "step": 315500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6671645764316138e-05,
+      "loss": 1.4171,
+      "step": 315600
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6664251064836727e-05,
+      "loss": 1.3779,
+      "step": 315700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6656856365357313e-05,
+      "loss": 1.406,
+      "step": 315800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6649461665877902e-05,
+      "loss": 1.4004,
+      "step": 315900
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6642066966398488e-05,
+      "loss": 1.419,
+      "step": 316000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.6634672266919077e-05,
+      "loss": 1.4127,
+      "step": 316100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6627277567439662e-05,
+      "loss": 1.4013,
+      "step": 316200
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.661988286796025e-05,
+      "loss": 1.418,
+      "step": 316300
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6612488168480837e-05,
+      "loss": 1.4184,
+      "step": 316400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.660509346900142e-05,
+      "loss": 1.3857,
+      "step": 316500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6597698769522005e-05,
+      "loss": 1.3737,
+      "step": 316600
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.659037801703739e-05,
+      "loss": 1.4014,
+      "step": 316700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6582983317557974e-05,
+      "loss": 1.4109,
+      "step": 316800
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6575588618078563e-05,
+      "loss": 1.3923,
+      "step": 316900
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.656819391859915e-05,
+      "loss": 1.3728,
+      "step": 317000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6560799219119738e-05,
+      "loss": 1.4171,
+      "step": 317100
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6553404519640324e-05,
+      "loss": 1.4139,
+      "step": 317200
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.654600982016091e-05,
+      "loss": 1.401,
+      "step": 317300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.6538615120681498e-05,
+      "loss": 1.3988,
+      "step": 317400
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6531220421202084e-05,
+      "loss": 1.4097,
+      "step": 317500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6523825721722673e-05,
+      "loss": 1.4153,
+      "step": 317600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651643102224326e-05,
+      "loss": 1.3922,
+      "step": 317700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.650903632276384e-05,
+      "loss": 1.3918,
+      "step": 317800
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.650164162328443e-05,
+      "loss": 1.4184,
+      "step": 317900
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6494246923805015e-05,
+      "loss": 1.4213,
+      "step": 318000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6486852224325604e-05,
+      "loss": 1.402,
+      "step": 318100
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.647945752484619e-05,
+      "loss": 1.3948,
+      "step": 318200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.647206282536678e-05,
+      "loss": 1.388,
+      "step": 318300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6464668125887365e-05,
+      "loss": 1.3909,
+      "step": 318400
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6457273426407954e-05,
+      "loss": 1.418,
+      "step": 318500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.644987872692854e-05,
+      "loss": 1.4173,
+      "step": 318600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.644248402744913e-05,
+      "loss": 1.4153,
+      "step": 318700
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6435089327969714e-05,
+      "loss": 1.3701,
+      "step": 318800
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6427694628490303e-05,
+      "loss": 1.3928,
+      "step": 318900
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6420373876005684e-05,
+      "loss": 1.4116,
+      "step": 319000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6412979176526266e-05,
+      "loss": 1.4005,
+      "step": 319100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.640558447704685e-05,
+      "loss": 1.3824,
+      "step": 319200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.639818977756744e-05,
+      "loss": 1.4029,
+      "step": 319300
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.6390795078088026e-05,
+      "loss": 1.4155,
+      "step": 319400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6383400378608615e-05,
+      "loss": 1.4083,
+      "step": 319500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.63760056791292e-05,
+      "loss": 1.4158,
+      "step": 319600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.636861097964979e-05,
+      "loss": 1.4122,
+      "step": 319700
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6361216280170375e-05,
+      "loss": 1.4198,
+      "step": 319800
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.635382158069096e-05,
+      "loss": 1.3805,
+      "step": 319900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.634642688121155e-05,
+      "loss": 1.4009,
+      "step": 320000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6339032181732136e-05,
+      "loss": 1.395,
+      "step": 320100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6331637482252725e-05,
+      "loss": 1.431,
+      "step": 320200
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.632424278277331e-05,
+      "loss": 1.4195,
+      "step": 320300
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6316848083293893e-05,
+      "loss": 1.374,
+      "step": 320400
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6309453383814482e-05,
+      "loss": 1.3721,
+      "step": 320500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6302058684335067e-05,
+      "loss": 1.4014,
+      "step": 320600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6294663984855656e-05,
+      "loss": 1.4085,
+      "step": 320700
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6287269285376242e-05,
+      "loss": 1.4139,
+      "step": 320800
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.627987458589683e-05,
+      "loss": 1.4055,
+      "step": 320900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6272479886417417e-05,
+      "loss": 1.4042,
+      "step": 321000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6265085186938006e-05,
+      "loss": 1.4195,
+      "step": 321100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6257764434453386e-05,
+      "loss": 1.3895,
+      "step": 321200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6250369734973972e-05,
+      "loss": 1.4039,
+      "step": 321300
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.624297503549456e-05,
+      "loss": 1.3841,
+      "step": 321400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6235580336015146e-05,
+      "loss": 1.3889,
+      "step": 321500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6228185636535735e-05,
+      "loss": 1.3979,
+      "step": 321600
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.622079093705632e-05,
+      "loss": 1.4202,
+      "step": 321700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6213396237576903e-05,
+      "loss": 1.3931,
+      "step": 321800
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6206001538097492e-05,
+      "loss": 1.3893,
+      "step": 321900
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6198606838618078e-05,
+      "loss": 1.3864,
+      "step": 322000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.6191212139138667e-05,
+      "loss": 1.4198,
+      "step": 322100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6183817439659253e-05,
+      "loss": 1.3964,
+      "step": 322200
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6176422740179842e-05,
+      "loss": 1.4046,
+      "step": 322300
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6169028040700427e-05,
+      "loss": 1.412,
+      "step": 322400
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6161633341221013e-05,
+      "loss": 1.4092,
+      "step": 322500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6154238641741602e-05,
+      "loss": 1.4168,
+      "step": 322600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6146843942262188e-05,
+      "loss": 1.3844,
+      "step": 322700
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6139449242782777e-05,
+      "loss": 1.38,
+      "step": 322800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6132054543303362e-05,
+      "loss": 1.4074,
+      "step": 322900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.612465984382395e-05,
+      "loss": 1.3954,
+      "step": 323000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6117265144344534e-05,
+      "loss": 1.3753,
+      "step": 323100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.610987044486512e-05,
+      "loss": 1.4286,
+      "step": 323200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6102475745385708e-05,
+      "loss": 1.4093,
+      "step": 323300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.609515499290109e-05,
+      "loss": 1.4161,
+      "step": 323400
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6087760293421674e-05,
+      "loss": 1.377,
+      "step": 323500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6080365593942263e-05,
+      "loss": 1.4051,
+      "step": 323600
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.607297089446285e-05,
+      "loss": 1.3867,
+      "step": 323700
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6065576194983438e-05,
+      "loss": 1.3852,
+      "step": 323800
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6058181495504024e-05,
+      "loss": 1.395,
+      "step": 323900
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6050786796024613e-05,
+      "loss": 1.4133,
+      "step": 324000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.60433920965452e-05,
+      "loss": 1.3982,
+      "step": 324100
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6035997397065787e-05,
+      "loss": 1.4248,
+      "step": 324200
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6028602697586373e-05,
+      "loss": 1.4216,
+      "step": 324300
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6021207998106955e-05,
+      "loss": 1.4014,
+      "step": 324400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6013813298627544e-05,
+      "loss": 1.4228,
+      "step": 324500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.600641859914813e-05,
+      "loss": 1.3547,
+      "step": 324600
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.599902389966872e-05,
+      "loss": 1.3968,
+      "step": 324700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.5991629200189305e-05,
+      "loss": 1.394,
+      "step": 324800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5984234500709894e-05,
+      "loss": 1.3957,
+      "step": 324900
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.597683980123048e-05,
+      "loss": 1.4235,
+      "step": 325000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5969445101751065e-05,
+      "loss": 1.4078,
+      "step": 325100
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5962050402271654e-05,
+      "loss": 1.3816,
+      "step": 325200
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.595465570279224e-05,
+      "loss": 1.3976,
+      "step": 325300
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.594726100331283e-05,
+      "loss": 1.402,
+      "step": 325400
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5939866303833414e-05,
+      "loss": 1.3941,
+      "step": 325500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5932545551348798e-05,
+      "loss": 1.4363,
+      "step": 325600
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.592515085186938e-05,
+      "loss": 1.4188,
+      "step": 325700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5917756152389966e-05,
+      "loss": 1.3932,
+      "step": 325800
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.591036145291055e-05,
+      "loss": 1.4292,
+      "step": 325900
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.590296675343114e-05,
+      "loss": 1.412,
+      "step": 326000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5895572053951726e-05,
+      "loss": 1.3943,
+      "step": 326100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5888177354472315e-05,
+      "loss": 1.4092,
+      "step": 326200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.58807826549929e-05,
+      "loss": 1.398,
+      "step": 326300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.587338795551349e-05,
+      "loss": 1.3717,
+      "step": 326400
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5865993256034076e-05,
+      "loss": 1.4091,
+      "step": 326500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5858598556554665e-05,
+      "loss": 1.3953,
+      "step": 326600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.585120385707525e-05,
+      "loss": 1.3906,
+      "step": 326700
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.584380915759584e-05,
+      "loss": 1.4101,
+      "step": 326800
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5836414458116425e-05,
+      "loss": 1.3829,
+      "step": 326900
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5829019758637007e-05,
+      "loss": 1.39,
+      "step": 327000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5821625059157596e-05,
+      "loss": 1.3935,
+      "step": 327100
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5814230359678182e-05,
+      "loss": 1.3865,
+      "step": 327200
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.580683566019877e-05,
+      "loss": 1.4081,
+      "step": 327300
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5799440960719356e-05,
+      "loss": 1.4142,
+      "step": 327400
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5792046261239946e-05,
+      "loss": 1.4058,
+      "step": 327500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.578465156176053e-05,
+      "loss": 1.4113,
+      "step": 327600
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5777256862281117e-05,
+      "loss": 1.424,
+      "step": 327700
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5769862162801706e-05,
+      "loss": 1.4089,
+      "step": 327800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.576246746332229e-05,
+      "loss": 1.3901,
+      "step": 327900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.575507276384288e-05,
+      "loss": 1.4001,
+      "step": 328000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5747678064363466e-05,
+      "loss": 1.3706,
+      "step": 328100
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5740283364884055e-05,
+      "loss": 1.4038,
+      "step": 328200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.573288866540464e-05,
+      "loss": 1.4163,
+      "step": 328300
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5725567912920018e-05,
+      "loss": 1.433,
+      "step": 328400
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5718173213440603e-05,
+      "loss": 1.3997,
+      "step": 328500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5710778513961193e-05,
+      "loss": 1.3934,
+      "step": 328600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5703383814481778e-05,
+      "loss": 1.3886,
+      "step": 328700
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5695989115002367e-05,
+      "loss": 1.4273,
+      "step": 328800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.5688594415522953e-05,
+      "loss": 1.3939,
+      "step": 328900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5681199716043542e-05,
+      "loss": 1.4052,
+      "step": 329000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5673805016564127e-05,
+      "loss": 1.398,
+      "step": 329100
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5666410317084716e-05,
+      "loss": 1.4253,
+      "step": 329200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5659015617605302e-05,
+      "loss": 1.4074,
+      "step": 329300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.565162091812589e-05,
+      "loss": 1.3836,
+      "step": 329400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5644226218646477e-05,
+      "loss": 1.3709,
+      "step": 329500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5636831519167066e-05,
+      "loss": 1.3991,
+      "step": 329600
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5629436819687648e-05,
+      "loss": 1.3991,
+      "step": 329700
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5622042120208234e-05,
+      "loss": 1.3876,
+      "step": 329800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5614647420728823e-05,
+      "loss": 1.4114,
+      "step": 329900
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.560725272124941e-05,
+      "loss": 1.4185,
+      "step": 330000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5599858021769997e-05,
+      "loss": 1.4153,
+      "step": 330100
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5592463322290583e-05,
+      "loss": 1.3831,
+      "step": 330200
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.558506862281117e-05,
+      "loss": 1.4161,
+      "step": 330300
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5577673923331758e-05,
+      "loss": 1.4176,
+      "step": 330400
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5570279223852343e-05,
+      "loss": 1.3946,
+      "step": 330500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5562884524372932e-05,
+      "loss": 1.404,
+      "step": 330600
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5555489824893518e-05,
+      "loss": 1.3935,
+      "step": 330700
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5548095125414107e-05,
+      "loss": 1.4059,
+      "step": 330800
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5540700425934693e-05,
+      "loss": 1.4215,
+      "step": 330900
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5533305726455275e-05,
+      "loss": 1.411,
+      "step": 331000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5525984973970655e-05,
+      "loss": 1.4078,
+      "step": 331100
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5518590274491244e-05,
+      "loss": 1.4236,
+      "step": 331200
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.551119557501183e-05,
+      "loss": 1.3888,
+      "step": 331300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.550380087553242e-05,
+      "loss": 1.3943,
+      "step": 331400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5496406176053005e-05,
+      "loss": 1.3837,
+      "step": 331500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5489011476573594e-05,
+      "loss": 1.3937,
+      "step": 331600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.548161677709418e-05,
+      "loss": 1.3897,
+      "step": 331700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.547422207761477e-05,
+      "loss": 1.4386,
+      "step": 331800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5466827378135354e-05,
+      "loss": 1.3989,
+      "step": 331900
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5459432678655943e-05,
+      "loss": 1.4145,
+      "step": 332000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.545203797917653e-05,
+      "loss": 1.4008,
+      "step": 332100
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5444643279697118e-05,
+      "loss": 1.3886,
+      "step": 332200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.54372485802177e-05,
+      "loss": 1.4076,
+      "step": 332300
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5429853880738286e-05,
+      "loss": 1.408,
+      "step": 332400
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5422459181258875e-05,
+      "loss": 1.3714,
+      "step": 332500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.541506448177946e-05,
+      "loss": 1.3997,
+      "step": 332600
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5407669782300046e-05,
+      "loss": 1.4232,
+      "step": 332700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.5400275082820635e-05,
+      "loss": 1.3737,
+      "step": 332800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.539288038334122e-05,
+      "loss": 1.3903,
+      "step": 332900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.538548568386181e-05,
+      "loss": 1.4049,
+      "step": 333000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5378090984382395e-05,
+      "loss": 1.3803,
+      "step": 333100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5370696284902984e-05,
+      "loss": 1.4067,
+      "step": 333200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.536330158542357e-05,
+      "loss": 1.3726,
+      "step": 333300
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.535590688594416e-05,
+      "loss": 1.4359,
+      "step": 333400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.534858613345954e-05,
+      "loss": 1.4146,
+      "step": 333500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.534119143398012e-05,
+      "loss": 1.4106,
+      "step": 333600
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5333796734500707e-05,
+      "loss": 1.3999,
+      "step": 333700
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5326402035021296e-05,
+      "loss": 1.4005,
+      "step": 333800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5319007335541882e-05,
+      "loss": 1.4153,
+      "step": 333900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.531161263606247e-05,
+      "loss": 1.4184,
+      "step": 334000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5304217936583057e-05,
+      "loss": 1.4164,
+      "step": 334100
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5296823237103646e-05,
+      "loss": 1.3939,
+      "step": 334200
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.528942853762423e-05,
+      "loss": 1.3871,
+      "step": 334300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.528203383814482e-05,
+      "loss": 1.3975,
+      "step": 334400
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5274639138665406e-05,
+      "loss": 1.4158,
+      "step": 334500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5267244439185995e-05,
+      "loss": 1.4101,
+      "step": 334600
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525984973970658e-05,
+      "loss": 1.4026,
+      "step": 334700
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.525245504022717e-05,
+      "loss": 1.3994,
+      "step": 334800
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5245060340747755e-05,
+      "loss": 1.3891,
+      "step": 334900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5237665641268337e-05,
+      "loss": 1.3969,
+      "step": 335000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5230270941788927e-05,
+      "loss": 1.4142,
+      "step": 335100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5222876242309512e-05,
+      "loss": 1.4041,
+      "step": 335200
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5215481542830098e-05,
+      "loss": 1.4263,
+      "step": 335300
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5208086843350687e-05,
+      "loss": 1.4229,
+      "step": 335400
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5200692143871272e-05,
+      "loss": 1.4045,
+      "step": 335500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.519329744439186e-05,
+      "loss": 1.3706,
+      "step": 335600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5185902744912447e-05,
+      "loss": 1.4175,
+      "step": 335700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5178508045433036e-05,
+      "loss": 1.387,
+      "step": 335800
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5171113345953622e-05,
+      "loss": 1.401,
+      "step": 335900
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.516371864647421e-05,
+      "loss": 1.4065,
+      "step": 336000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5156323946994796e-05,
+      "loss": 1.3945,
+      "step": 336100
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5148929247515385e-05,
+      "loss": 1.4092,
+      "step": 336200
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.514160849503076e-05,
+      "loss": 1.4123,
+      "step": 336300
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5134213795551348e-05,
+      "loss": 1.3895,
+      "step": 336400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5126819096071934e-05,
+      "loss": 1.399,
+      "step": 336500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5119424396592523e-05,
+      "loss": 1.3791,
+      "step": 336600
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511202969711311e-05,
+      "loss": 1.4009,
+      "step": 336700
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5104634997633697e-05,
+      "loss": 1.4128,
+      "step": 336800
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5097240298154283e-05,
+      "loss": 1.4035,
+      "step": 336900
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5089845598674872e-05,
+      "loss": 1.4161,
+      "step": 337000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5082450899195458e-05,
+      "loss": 1.4023,
+      "step": 337100
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5075056199716047e-05,
+      "loss": 1.381,
+      "step": 337200
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5067661500236632e-05,
+      "loss": 1.4085,
+      "step": 337300
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.506026680075722e-05,
+      "loss": 1.3829,
+      "step": 337400
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5052872101277807e-05,
+      "loss": 1.4298,
+      "step": 337500
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.504547740179839e-05,
+      "loss": 1.411,
+      "step": 337600
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.503808270231898e-05,
+      "loss": 1.4376,
+      "step": 337700
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5030688002839564e-05,
+      "loss": 1.4147,
+      "step": 337800
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.502329330336015e-05,
+      "loss": 1.3851,
+      "step": 337900
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.501589860388074e-05,
+      "loss": 1.4166,
+      "step": 338000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5008503904401324e-05,
+      "loss": 1.3948,
+      "step": 338100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5001109204921913e-05,
+      "loss": 1.3702,
+      "step": 338200
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.49937145054425e-05,
+      "loss": 1.3485,
+      "step": 338300
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.4986393752957883e-05,
+      "loss": 1.3543,
+      "step": 338400
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.497899905347847e-05,
+      "loss": 1.3329,
+      "step": 338500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4971604353999054e-05,
+      "loss": 1.3373,
+      "step": 338600
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.496420965451964e-05,
+      "loss": 1.3363,
+      "step": 338700
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.495681495504023e-05,
+      "loss": 1.3605,
+      "step": 338800
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4949420255560814e-05,
+      "loss": 1.3547,
+      "step": 338900
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.4942025556081403e-05,
+      "loss": 1.3487,
+      "step": 339000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.493463085660199e-05,
+      "loss": 1.3582,
+      "step": 339100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4927236157122578e-05,
+      "loss": 1.3308,
+      "step": 339200
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.491984145764316e-05,
+      "loss": 1.334,
+      "step": 339300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.491244675816375e-05,
+      "loss": 1.3531,
+      "step": 339400
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4905052058684335e-05,
+      "loss": 1.3437,
+      "step": 339500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.4897657359204924e-05,
+      "loss": 1.3161,
+      "step": 339600
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.489026265972551e-05,
+      "loss": 1.3156,
+      "step": 339700
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.48828679602461e-05,
+      "loss": 1.3701,
+      "step": 339800
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.487547326076668e-05,
+      "loss": 1.3449,
+      "step": 339900
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.486807856128727e-05,
+      "loss": 1.3695,
+      "step": 340000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4860683861807856e-05,
+      "loss": 1.3214,
+      "step": 340100
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4853289162328445e-05,
+      "loss": 1.3218,
+      "step": 340200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.484589446284903e-05,
+      "loss": 1.3362,
+      "step": 340300
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.483849976336962e-05,
+      "loss": 1.3517,
+      "step": 340400
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4831105063890205e-05,
+      "loss": 1.3528,
+      "step": 340500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.482371036441079e-05,
+      "loss": 1.3078,
+      "step": 340600
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.481638961192617e-05,
+      "loss": 1.3507,
+      "step": 340700
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.480899491244676e-05,
+      "loss": 1.3499,
+      "step": 340800
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4801600212967346e-05,
+      "loss": 1.3384,
+      "step": 340900
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4794205513487935e-05,
+      "loss": 1.35,
+      "step": 341000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.478681081400852e-05,
+      "loss": 1.3296,
+      "step": 341100
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4779416114529106e-05,
+      "loss": 1.3198,
+      "step": 341200
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.477202141504969e-05,
+      "loss": 1.332,
+      "step": 341300
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.476462671557028e-05,
+      "loss": 1.3688,
+      "step": 341400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4757232016090866e-05,
+      "loss": 1.3425,
+      "step": 341500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4749837316611455e-05,
+      "loss": 1.3265,
+      "step": 341600
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.474244261713204e-05,
+      "loss": 1.3433,
+      "step": 341700
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.473504791765263e-05,
+      "loss": 1.3219,
+      "step": 341800
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.4727653218173212e-05,
+      "loss": 1.3654,
+      "step": 341900
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.47202585186938e-05,
+      "loss": 1.3468,
+      "step": 342000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.4712863819214387e-05,
+      "loss": 1.3484,
+      "step": 342100
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.4705469119734976e-05,
+      "loss": 1.3366,
+      "step": 342200
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.469807442025556e-05,
+      "loss": 1.3378,
+      "step": 342300
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.469067972077615e-05,
+      "loss": 1.33,
+      "step": 342400
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4683285021296736e-05,
+      "loss": 1.3422,
+      "step": 342500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4675890321817322e-05,
+      "loss": 1.3506,
+      "step": 342600
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4668495622337907e-05,
+      "loss": 1.3379,
+      "step": 342700
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.466117486985329e-05,
+      "loss": 1.3706,
+      "step": 342800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4653780170373877e-05,
+      "loss": 1.3361,
+      "step": 342900
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4646385470894466e-05,
+      "loss": 1.3462,
+      "step": 343000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.463899077141505e-05,
+      "loss": 1.3442,
+      "step": 343100
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4631596071935637e-05,
+      "loss": 1.336,
+      "step": 343200
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4624201372456223e-05,
+      "loss": 1.3334,
+      "step": 343300
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4616806672976812e-05,
+      "loss": 1.3449,
+      "step": 343400
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4609411973497398e-05,
+      "loss": 1.336,
+      "step": 343500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4602017274017987e-05,
+      "loss": 1.3497,
+      "step": 343600
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4594622574538572e-05,
+      "loss": 1.3586,
+      "step": 343700
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.458722787505916e-05,
+      "loss": 1.323,
+      "step": 343800
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4579833175579744e-05,
+      "loss": 1.3518,
+      "step": 343900
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4572438476100333e-05,
+      "loss": 1.3332,
+      "step": 344000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4565043776620918e-05,
+      "loss": 1.3651,
+      "step": 344100
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4557649077141507e-05,
+      "loss": 1.3599,
+      "step": 344200
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4550254377662093e-05,
+      "loss": 1.3434,
+      "step": 344300
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4542859678182682e-05,
+      "loss": 1.3533,
+      "step": 344400
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4535464978703264e-05,
+      "loss": 1.3188,
+      "step": 344500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4528070279223853e-05,
+      "loss": 1.3369,
+      "step": 344600
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.452067557974444e-05,
+      "loss": 1.3393,
+      "step": 344700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4513280880265028e-05,
+      "loss": 1.3307,
+      "step": 344800
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4505960127780408e-05,
+      "loss": 1.3485,
+      "step": 344900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4498565428300997e-05,
+      "loss": 1.3228,
+      "step": 345000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4491170728821583e-05,
+      "loss": 1.3602,
+      "step": 345100
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.448377602934217e-05,
+      "loss": 1.3233,
+      "step": 345200
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4476381329862754e-05,
+      "loss": 1.352,
+      "step": 345300
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4468986630383343e-05,
+      "loss": 1.3616,
+      "step": 345400
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.446159193090393e-05,
+      "loss": 1.333,
+      "step": 345500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4454197231424518e-05,
+      "loss": 1.369,
+      "step": 345600
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4446802531945104e-05,
+      "loss": 1.3199,
+      "step": 345700
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4439407832465693e-05,
+      "loss": 1.3311,
+      "step": 345800
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4432013132986275e-05,
+      "loss": 1.3319,
+      "step": 345900
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4424618433506864e-05,
+      "loss": 1.3702,
+      "step": 346000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.441722373402745e-05,
+      "loss": 1.3414,
+      "step": 346100
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.440982903454804e-05,
+      "loss": 1.3635,
+      "step": 346200
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4402434335068624e-05,
+      "loss": 1.3551,
+      "step": 346300
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4395039635589213e-05,
+      "loss": 1.3301,
+      "step": 346400
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.4387644936109795e-05,
+      "loss": 1.3368,
+      "step": 346500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4380250236630384e-05,
+      "loss": 1.3348,
+      "step": 346600
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.437285553715097e-05,
+      "loss": 1.3608,
+      "step": 346700
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.436546083767156e-05,
+      "loss": 1.3438,
+      "step": 346800
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4358066138192145e-05,
+      "loss": 1.3665,
+      "step": 346900
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4350671438712734e-05,
+      "loss": 1.3661,
+      "step": 347000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.4343350686228114e-05,
+      "loss": 1.3751,
+      "step": 347100
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.43359559867487e-05,
+      "loss": 1.3241,
+      "step": 347200
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4328561287269285e-05,
+      "loss": 1.3514,
+      "step": 347300
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4321166587789875e-05,
+      "loss": 1.35,
+      "step": 347400
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.431377188831046e-05,
+      "loss": 1.3572,
+      "step": 347500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.430637718883105e-05,
+      "loss": 1.3777,
+      "step": 347600
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4298982489351635e-05,
+      "loss": 1.3747,
+      "step": 347700
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.429158778987222e-05,
+      "loss": 1.3402,
+      "step": 347800
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4284193090392806e-05,
+      "loss": 1.372,
+      "step": 347900
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4276798390913395e-05,
+      "loss": 1.3475,
+      "step": 348000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.426940369143398e-05,
+      "loss": 1.3753,
+      "step": 348100
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.426200899195457e-05,
+      "loss": 1.3414,
+      "step": 348200
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4254614292475155e-05,
+      "loss": 1.3287,
+      "step": 348300
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4247219592995744e-05,
+      "loss": 1.36,
+      "step": 348400
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 2.4239824893516327e-05,
+      "loss": 1.3176,
+      "step": 348500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4232430194036916e-05,
+      "loss": 1.3575,
+      "step": 348600
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.42250354945575e-05,
+      "loss": 1.3686,
+      "step": 348700
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.421764079507809e-05,
+      "loss": 1.3468,
+      "step": 348800
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4210246095598676e-05,
+      "loss": 1.3428,
+      "step": 348900
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4202851396119265e-05,
+      "loss": 1.3493,
+      "step": 349000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.419545669663985e-05,
+      "loss": 1.3746,
+      "step": 349100
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4188061997160436e-05,
+      "loss": 1.3618,
+      "step": 349200
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4180667297681022e-05,
+      "loss": 1.3373,
+      "step": 349300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4173346545196406e-05,
+      "loss": 1.3801,
+      "step": 349400
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.416595184571699e-05,
+      "loss": 1.3208,
+      "step": 349500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4158557146237577e-05,
+      "loss": 1.3318,
+      "step": 349600
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4151162446758166e-05,
+      "loss": 1.3546,
+      "step": 349700
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4143767747278752e-05,
+      "loss": 1.3686,
+      "step": 349800
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.4136373047799337e-05,
+      "loss": 1.366,
+      "step": 349900
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4128978348319926e-05,
+      "loss": 1.3592,
+      "step": 350000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4121583648840512e-05,
+      "loss": 1.3487,
+      "step": 350100
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.41141889493611e-05,
+      "loss": 1.3572,
+      "step": 350200
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4106794249881687e-05,
+      "loss": 1.3297,
+      "step": 350300
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4099399550402276e-05,
+      "loss": 1.3562,
+      "step": 350400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4092004850922858e-05,
+      "loss": 1.3757,
+      "step": 350500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4084610151443447e-05,
+      "loss": 1.3353,
+      "step": 350600
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4077215451964033e-05,
+      "loss": 1.3293,
+      "step": 350700
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.406982075248462e-05,
+      "loss": 1.3459,
+      "step": 350800
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4062426053005207e-05,
+      "loss": 1.3724,
+      "step": 350900
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4055031353525796e-05,
+      "loss": 1.348,
+      "step": 351000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.404763665404638e-05,
+      "loss": 1.3536,
+      "step": 351100
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4040241954566968e-05,
+      "loss": 1.3623,
+      "step": 351200
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4032847255087553e-05,
+      "loss": 1.3587,
+      "step": 351300
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4025452555608142e-05,
+      "loss": 1.3258,
+      "step": 351400
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4018057856128728e-05,
+      "loss": 1.3554,
+      "step": 351500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4010663156649317e-05,
+      "loss": 1.3583,
+      "step": 351600
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4003268457169903e-05,
+      "loss": 1.3475,
+      "step": 351700
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.3995873757690488e-05,
+      "loss": 1.3436,
+      "step": 351800
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.398855300520587e-05,
+      "loss": 1.3676,
+      "step": 351900
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3981158305726458e-05,
+      "loss": 1.3588,
+      "step": 352000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3973763606247043e-05,
+      "loss": 1.3522,
+      "step": 352100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.396636890676763e-05,
+      "loss": 1.3447,
+      "step": 352200
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3958974207288218e-05,
+      "loss": 1.3548,
+      "step": 352300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3951579507808804e-05,
+      "loss": 1.3581,
+      "step": 352400
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.394418480832939e-05,
+      "loss": 1.3626,
+      "step": 352500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3936790108849978e-05,
+      "loss": 1.3487,
+      "step": 352600
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3929395409370564e-05,
+      "loss": 1.3624,
+      "step": 352700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3922000709891153e-05,
+      "loss": 1.3553,
+      "step": 352800
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.391460601041174e-05,
+      "loss": 1.3216,
+      "step": 352900
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3907211310932324e-05,
+      "loss": 1.3566,
+      "step": 353000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.389981661145291e-05,
+      "loss": 1.3676,
+      "step": 353100
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.38924219119735e-05,
+      "loss": 1.3445,
+      "step": 353200
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3885027212494085e-05,
+      "loss": 1.3795,
+      "step": 353300
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3877632513014674e-05,
+      "loss": 1.3506,
+      "step": 353400
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.387023781353526e-05,
+      "loss": 1.3655,
+      "step": 353500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3862843114055848e-05,
+      "loss": 1.3697,
+      "step": 353600
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3855448414576434e-05,
+      "loss": 1.3324,
+      "step": 353700
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.384805371509702e-05,
+      "loss": 1.3343,
+      "step": 353800
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3840659015617605e-05,
+      "loss": 1.355,
+      "step": 353900
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3833264316138194e-05,
+      "loss": 1.3639,
+      "step": 354000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.382586961665878e-05,
+      "loss": 1.3654,
+      "step": 354100
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.381854886417416e-05,
+      "loss": 1.3718,
+      "step": 354200
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.381115416469475e-05,
+      "loss": 1.3633,
+      "step": 354300
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3803759465215335e-05,
+      "loss": 1.3697,
+      "step": 354400
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.379636476573592e-05,
+      "loss": 1.3683,
+      "step": 354500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.378897006625651e-05,
+      "loss": 1.3455,
+      "step": 354600
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3781575366777095e-05,
+      "loss": 1.3487,
+      "step": 354700
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.377418066729768e-05,
+      "loss": 1.3375,
+      "step": 354800
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.376678596781827e-05,
+      "loss": 1.3701,
+      "step": 354900
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3759391268338856e-05,
+      "loss": 1.3438,
+      "step": 355000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.375199656885944e-05,
+      "loss": 1.3543,
+      "step": 355100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.374460186938003e-05,
+      "loss": 1.3552,
+      "step": 355200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3737207169900616e-05,
+      "loss": 1.3419,
+      "step": 355300
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3729812470421205e-05,
+      "loss": 1.3706,
+      "step": 355400
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.372241777094179e-05,
+      "loss": 1.3587,
+      "step": 355500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3715023071462376e-05,
+      "loss": 1.348,
+      "step": 355600
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3707628371982965e-05,
+      "loss": 1.3674,
+      "step": 355700
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.370023367250355e-05,
+      "loss": 1.3469,
+      "step": 355800
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3692838973024136e-05,
+      "loss": 1.3095,
+      "step": 355900
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3685444273544725e-05,
+      "loss": 1.3535,
+      "step": 356000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.367804957406531e-05,
+      "loss": 1.3539,
+      "step": 356100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.367072882158069e-05,
+      "loss": 1.3797,
+      "step": 356200
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.366333412210128e-05,
+      "loss": 1.3584,
+      "step": 356300
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3655939422621863e-05,
+      "loss": 1.3396,
+      "step": 356400
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3648544723142452e-05,
+      "loss": 1.3826,
+      "step": 356500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3641150023663037e-05,
+      "loss": 1.3658,
+      "step": 356600
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3633755324183627e-05,
+      "loss": 1.3391,
+      "step": 356700
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3626360624704212e-05,
+      "loss": 1.3616,
+      "step": 356800
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.36189659252248e-05,
+      "loss": 1.3618,
+      "step": 356900
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3611571225745387e-05,
+      "loss": 1.3438,
+      "step": 357000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3604176526265972e-05,
+      "loss": 1.336,
+      "step": 357100
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.359678182678656e-05,
+      "loss": 1.3394,
+      "step": 357200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.3589387127307147e-05,
+      "loss": 1.3923,
+      "step": 357300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3581992427827733e-05,
+      "loss": 1.3293,
+      "step": 357400
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3574597728348322e-05,
+      "loss": 1.3411,
+      "step": 357500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3567203028868907e-05,
+      "loss": 1.3703,
+      "step": 357600
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3559808329389496e-05,
+      "loss": 1.3448,
+      "step": 357700
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3552413629910082e-05,
+      "loss": 1.3374,
+      "step": 357800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3545018930430668e-05,
+      "loss": 1.3796,
+      "step": 357900
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3537624230951257e-05,
+      "loss": 1.3564,
+      "step": 358000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3530229531471842e-05,
+      "loss": 1.3552,
+      "step": 358100
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3522908778987223e-05,
+      "loss": 1.3746,
+      "step": 358200
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3515514079507812e-05,
+      "loss": 1.3688,
+      "step": 358300
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3508119380028394e-05,
+      "loss": 1.3477,
+      "step": 358400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3500724680548983e-05,
+      "loss": 1.3635,
+      "step": 358500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.349332998106957e-05,
+      "loss": 1.3558,
+      "step": 358600
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3485935281590158e-05,
+      "loss": 1.355,
+      "step": 358700
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3478540582110743e-05,
+      "loss": 1.3598,
+      "step": 358800
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3471145882631332e-05,
+      "loss": 1.3429,
+      "step": 358900
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3463751183151918e-05,
+      "loss": 1.3768,
+      "step": 359000
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.3456356483672504e-05,
+      "loss": 1.3594,
+      "step": 359100
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.344896178419309e-05,
+      "loss": 1.3476,
+      "step": 359200
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.344156708471368e-05,
+      "loss": 1.3589,
+      "step": 359300
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3434172385234264e-05,
+      "loss": 1.326,
+      "step": 359400
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3426777685754853e-05,
+      "loss": 1.3501,
+      "step": 359500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.341938298627544e-05,
+      "loss": 1.3557,
+      "step": 359600
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3411988286796024e-05,
+      "loss": 1.324,
+      "step": 359700
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3404593587316613e-05,
+      "loss": 1.3608,
+      "step": 359800
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.33971988878372e-05,
+      "loss": 1.3564,
+      "step": 359900
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.3389804188357785e-05,
+      "loss": 1.3756,
+      "step": 360000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.338248343587317e-05,
+      "loss": 1.3827,
+      "step": 360100
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3375088736393754e-05,
+      "loss": 1.3527,
+      "step": 360200
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3367694036914343e-05,
+      "loss": 1.3564,
+      "step": 360300
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3360299337434925e-05,
+      "loss": 1.3469,
+      "step": 360400
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3352904637955514e-05,
+      "loss": 1.3817,
+      "step": 360500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.33455099384761e-05,
+      "loss": 1.3679,
+      "step": 360600
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.333811523899669e-05,
+      "loss": 1.3551,
+      "step": 360700
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3330720539517275e-05,
+      "loss": 1.3651,
+      "step": 360800
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3323325840037864e-05,
+      "loss": 1.3574,
+      "step": 360900
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.331593114055845e-05,
+      "loss": 1.3482,
+      "step": 361000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3308536441079035e-05,
+      "loss": 1.3568,
+      "step": 361100
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.330114174159962e-05,
+      "loss": 1.3438,
+      "step": 361200
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.329374704212021e-05,
+      "loss": 1.3549,
+      "step": 361300
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3286352342640795e-05,
+      "loss": 1.3632,
+      "step": 361400
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3278957643161384e-05,
+      "loss": 1.3705,
+      "step": 361500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.327156294368197e-05,
+      "loss": 1.3236,
+      "step": 361600
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3264168244202556e-05,
+      "loss": 1.3419,
+      "step": 361700
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.325677354472314e-05,
+      "loss": 1.3449,
+      "step": 361800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.324937884524373e-05,
+      "loss": 1.3646,
+      "step": 361900
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3241984145764316e-05,
+      "loss": 1.3447,
+      "step": 362000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3234589446284905e-05,
+      "loss": 1.375,
+      "step": 362100
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.322719474680549e-05,
+      "loss": 1.3972,
+      "step": 362200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.321980004732608e-05,
+      "loss": 1.3478,
+      "step": 362300
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3212405347846662e-05,
+      "loss": 1.3451,
+      "step": 362400
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.320501064836725e-05,
+      "loss": 1.3443,
+      "step": 362500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3197615948887837e-05,
+      "loss": 1.3716,
+      "step": 362600
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3190221249408426e-05,
+      "loss": 1.36,
+      "step": 362700
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.318282654992901e-05,
+      "loss": 1.3498,
+      "step": 362800
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.31754318504496e-05,
+      "loss": 1.3447,
+      "step": 362900
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3168037150970186e-05,
+      "loss": 1.3486,
+      "step": 363000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.316064245149077e-05,
+      "loss": 1.3642,
+      "step": 363100
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.315324775201136e-05,
+      "loss": 1.3316,
+      "step": 363200
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3145853052531946e-05,
+      "loss": 1.3485,
+      "step": 363300
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3138458353052532e-05,
+      "loss": 1.3408,
+      "step": 363400
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.313106365357312e-05,
+      "loss": 1.3639,
+      "step": 363500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3123668954093706e-05,
+      "loss": 1.3714,
+      "step": 363600
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3116274254614292e-05,
+      "loss": 1.3556,
+      "step": 363700
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.310887955513488e-05,
+      "loss": 1.3674,
+      "step": 363800
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3101484855655467e-05,
+      "loss": 1.3647,
+      "step": 363900
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3094090156176056e-05,
+      "loss": 1.3344,
+      "step": 364000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.3086769403691436e-05,
+      "loss": 1.3416,
+      "step": 364100
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3079374704212022e-05,
+      "loss": 1.3712,
+      "step": 364200
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.307198000473261e-05,
+      "loss": 1.3582,
+      "step": 364300
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3064585305253193e-05,
+      "loss": 1.3546,
+      "step": 364400
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3057190605773782e-05,
+      "loss": 1.3509,
+      "step": 364500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3049795906294368e-05,
+      "loss": 1.3188,
+      "step": 364600
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3042401206814957e-05,
+      "loss": 1.3573,
+      "step": 364700
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3035006507335542e-05,
+      "loss": 1.3377,
+      "step": 364800
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.302761180785613e-05,
+      "loss": 1.3533,
+      "step": 364900
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3020217108376714e-05,
+      "loss": 1.3818,
+      "step": 365000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.3012822408897303e-05,
+      "loss": 1.3548,
+      "step": 365100
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.300542770941789e-05,
+      "loss": 1.3449,
+      "step": 365200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.2998033009938477e-05,
+      "loss": 1.3479,
+      "step": 365300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.2990638310459063e-05,
+      "loss": 1.3131,
+      "step": 365400
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2983243610979652e-05,
+      "loss": 1.3591,
+      "step": 365500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2975848911500238e-05,
+      "loss": 1.3443,
+      "step": 365600
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2968454212020823e-05,
+      "loss": 1.3336,
+      "step": 365700
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2961059512541412e-05,
+      "loss": 1.3556,
+      "step": 365800
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2953664813061998e-05,
+      "loss": 1.3977,
+      "step": 365900
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2946270113582584e-05,
+      "loss": 1.3506,
+      "step": 366000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2938875414103173e-05,
+      "loss": 1.3391,
+      "step": 366100
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.293148071462376e-05,
+      "loss": 1.3433,
+      "step": 366200
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2924086015144344e-05,
+      "loss": 1.3471,
+      "step": 366300
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2916765262659724e-05,
+      "loss": 1.3525,
+      "step": 366400
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2909370563180313e-05,
+      "loss": 1.3621,
+      "step": 366500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.29019758637009e-05,
+      "loss": 1.3505,
+      "step": 366600
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2894581164221488e-05,
+      "loss": 1.3749,
+      "step": 366700
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2887186464742074e-05,
+      "loss": 1.3439,
+      "step": 366800
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2879791765262663e-05,
+      "loss": 1.3735,
+      "step": 366900
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2872397065783245e-05,
+      "loss": 1.3718,
+      "step": 367000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2865002366303834e-05,
+      "loss": 1.3532,
+      "step": 367100
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.285760766682442e-05,
+      "loss": 1.3453,
+      "step": 367200
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.285021296734501e-05,
+      "loss": 1.3509,
+      "step": 367300
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2842818267865594e-05,
+      "loss": 1.3741,
+      "step": 367400
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2835423568386183e-05,
+      "loss": 1.3597,
+      "step": 367500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.282802886890677e-05,
+      "loss": 1.3519,
+      "step": 367600
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2820634169427355e-05,
+      "loss": 1.3491,
+      "step": 367700
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.281323946994794e-05,
+      "loss": 1.3902,
+      "step": 367800
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.280584477046853e-05,
+      "loss": 1.3428,
+      "step": 367900
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2798450070989115e-05,
+      "loss": 1.3533,
+      "step": 368000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.2791055371509704e-05,
+      "loss": 1.3521,
+      "step": 368100
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.278366067203029e-05,
+      "loss": 1.3501,
+      "step": 368200
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2776265972550875e-05,
+      "loss": 1.3528,
+      "step": 368300
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2768871273071464e-05,
+      "loss": 1.3558,
+      "step": 368400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2761550520586845e-05,
+      "loss": 1.3692,
+      "step": 368500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.275415582110743e-05,
+      "loss": 1.3566,
+      "step": 368600
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.274676112162802e-05,
+      "loss": 1.3515,
+      "step": 368700
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.2739366422148605e-05,
+      "loss": 1.361,
+      "step": 368800
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2731971722669194e-05,
+      "loss": 1.3528,
+      "step": 368900
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2724577023189776e-05,
+      "loss": 1.3766,
+      "step": 369000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2717182323710365e-05,
+      "loss": 1.356,
+      "step": 369100
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.270978762423095e-05,
+      "loss": 1.3618,
+      "step": 369200
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.270239292475154e-05,
+      "loss": 1.3446,
+      "step": 369300
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2694998225272126e-05,
+      "loss": 1.3522,
+      "step": 369400
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2687603525792715e-05,
+      "loss": 1.3401,
+      "step": 369500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2680208826313297e-05,
+      "loss": 1.3564,
+      "step": 369600
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2672814126833886e-05,
+      "loss": 1.3251,
+      "step": 369700
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.266541942735447e-05,
+      "loss": 1.3456,
+      "step": 369800
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.265802472787506e-05,
+      "loss": 1.3496,
+      "step": 369900
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2650630028395646e-05,
+      "loss": 1.346,
+      "step": 370000
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.2643235328916235e-05,
+      "loss": 1.3617,
+      "step": 370100
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.263584062943682e-05,
+      "loss": 1.3442,
+      "step": 370200
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2628445929957407e-05,
+      "loss": 1.3626,
+      "step": 370300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2621051230477992e-05,
+      "loss": 1.3643,
+      "step": 370400
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.261365653099858e-05,
+      "loss": 1.3718,
+      "step": 370500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.260633577851396e-05,
+      "loss": 1.3166,
+      "step": 370600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.259894107903455e-05,
+      "loss": 1.3684,
+      "step": 370700
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2591546379555136e-05,
+      "loss": 1.313,
+      "step": 370800
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2584151680075725e-05,
+      "loss": 1.3632,
+      "step": 370900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2576756980596308e-05,
+      "loss": 1.343,
+      "step": 371000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2569362281116897e-05,
+      "loss": 1.3527,
+      "step": 371100
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2561967581637482e-05,
+      "loss": 1.3526,
+      "step": 371200
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.255457288215807e-05,
+      "loss": 1.3481,
+      "step": 371300
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2547178182678657e-05,
+      "loss": 1.3524,
+      "step": 371400
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2539783483199246e-05,
+      "loss": 1.3399,
+      "step": 371500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2532388783719828e-05,
+      "loss": 1.3331,
+      "step": 371600
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2524994084240417e-05,
+      "loss": 1.3776,
+      "step": 371700
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2517599384761003e-05,
+      "loss": 1.3537,
+      "step": 371800
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2510204685281592e-05,
+      "loss": 1.3599,
+      "step": 371900
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2502809985802178e-05,
+      "loss": 1.3614,
+      "step": 372000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2495415286322767e-05,
+      "loss": 1.3527,
+      "step": 372100
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2488020586843352e-05,
+      "loss": 1.3766,
+      "step": 372200
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2480625887363938e-05,
+      "loss": 1.3389,
+      "step": 372300
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2473231187884523e-05,
+      "loss": 1.3551,
+      "step": 372400
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2465836488405112e-05,
+      "loss": 1.3791,
+      "step": 372500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2458515735920493e-05,
+      "loss": 1.3556,
+      "step": 372600
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2451121036441082e-05,
+      "loss": 1.3639,
+      "step": 372700
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2443726336961668e-05,
+      "loss": 1.3547,
+      "step": 372800
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.2436331637482253e-05,
+      "loss": 1.3513,
+      "step": 372900
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.242893693800284e-05,
+      "loss": 1.3674,
+      "step": 373000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2421542238523428e-05,
+      "loss": 1.356,
+      "step": 373100
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2414147539044014e-05,
+      "loss": 1.3506,
+      "step": 373200
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2406752839564603e-05,
+      "loss": 1.3393,
+      "step": 373300
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2399358140085188e-05,
+      "loss": 1.3737,
+      "step": 373400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2391963440605777e-05,
+      "loss": 1.3702,
+      "step": 373500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.238456874112636e-05,
+      "loss": 1.3421,
+      "step": 373600
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.237717404164695e-05,
+      "loss": 1.3553,
+      "step": 373700
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2369779342167534e-05,
+      "loss": 1.3766,
+      "step": 373800
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2362384642688123e-05,
+      "loss": 1.363,
+      "step": 373900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.235498994320871e-05,
+      "loss": 1.3612,
+      "step": 374000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2347595243729298e-05,
+      "loss": 1.3513,
+      "step": 374100
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2340200544249883e-05,
+      "loss": 1.3717,
+      "step": 374200
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.233280584477047e-05,
+      "loss": 1.3578,
+      "step": 374300
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2325411145291055e-05,
+      "loss": 1.3693,
+      "step": 374400
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2318016445811644e-05,
+      "loss": 1.3801,
+      "step": 374500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.231062174633223e-05,
+      "loss": 1.3535,
+      "step": 374600
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.230322704685282e-05,
+      "loss": 1.3847,
+      "step": 374700
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2295832347373404e-05,
+      "loss": 1.3473,
+      "step": 374800
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.2288511594888785e-05,
+      "loss": 1.3502,
+      "step": 374900
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.228111689540937e-05,
+      "loss": 1.377,
+      "step": 375000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.227372219592996e-05,
+      "loss": 1.3696,
+      "step": 375100
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2266327496450545e-05,
+      "loss": 1.3537,
+      "step": 375200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2258932796971134e-05,
+      "loss": 1.3625,
+      "step": 375300
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.225153809749172e-05,
+      "loss": 1.3548,
+      "step": 375400
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.224414339801231e-05,
+      "loss": 1.3559,
+      "step": 375500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.223674869853289e-05,
+      "loss": 1.3543,
+      "step": 375600
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.222935399905348e-05,
+      "loss": 1.3594,
+      "step": 375700
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2221959299574065e-05,
+      "loss": 1.3565,
+      "step": 375800
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2214564600094654e-05,
+      "loss": 1.3714,
+      "step": 375900
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.220716990061524e-05,
+      "loss": 1.3748,
+      "step": 376000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.219977520113583e-05,
+      "loss": 1.3814,
+      "step": 376100
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.219238050165641e-05,
+      "loss": 1.375,
+      "step": 376200
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2184985802177e-05,
+      "loss": 1.3437,
+      "step": 376300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2177591102697586e-05,
+      "loss": 1.3762,
+      "step": 376400
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2170196403218175e-05,
+      "loss": 1.342,
+      "step": 376500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.216280170373876e-05,
+      "loss": 1.345,
+      "step": 376600
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.215540700425935e-05,
+      "loss": 1.3638,
+      "step": 376700
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2148012304779935e-05,
+      "loss": 1.3701,
+      "step": 376800
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.214061760530052e-05,
+      "loss": 1.3759,
+      "step": 376900
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.21332968528159e-05,
+      "loss": 1.3683,
+      "step": 377000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.212590215333649e-05,
+      "loss": 1.3851,
+      "step": 377100
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2118507453857076e-05,
+      "loss": 1.3368,
+      "step": 377200
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2111112754377665e-05,
+      "loss": 1.3602,
+      "step": 377300
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.210371805489825e-05,
+      "loss": 1.3416,
+      "step": 377400
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.209632335541884e-05,
+      "loss": 1.3721,
+      "step": 377500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2088928655939422e-05,
+      "loss": 1.3514,
+      "step": 377600
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.208153395646001e-05,
+      "loss": 1.3524,
+      "step": 377700
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2074139256980597e-05,
+      "loss": 1.3703,
+      "step": 377800
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2066744557501186e-05,
+      "loss": 1.3542,
+      "step": 377900
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.205934985802177e-05,
+      "loss": 1.3647,
+      "step": 378000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.205195515854236e-05,
+      "loss": 1.3634,
+      "step": 378100
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.2044560459062943e-05,
+      "loss": 1.3594,
+      "step": 378200
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.203716575958353e-05,
+      "loss": 1.3639,
+      "step": 378300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2029771060104117e-05,
+      "loss": 1.3542,
+      "step": 378400
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2022376360624706e-05,
+      "loss": 1.3474,
+      "step": 378500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2014981661145292e-05,
+      "loss": 1.3861,
+      "step": 378600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.200758696166588e-05,
+      "loss": 1.3639,
+      "step": 378700
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2000192262186467e-05,
+      "loss": 1.3835,
+      "step": 378800
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.1992797562707052e-05,
+      "loss": 1.3601,
+      "step": 378900
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1985402863227638e-05,
+      "loss": 1.3475,
+      "step": 379000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1978008163748227e-05,
+      "loss": 1.3506,
+      "step": 379100
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1970613464268813e-05,
+      "loss": 1.3434,
+      "step": 379200
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.19632187647894e-05,
+      "loss": 1.3621,
+      "step": 379300
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1955898012304782e-05,
+      "loss": 1.3617,
+      "step": 379400
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1948503312825368e-05,
+      "loss": 1.3635,
+      "step": 379500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1941108613345953e-05,
+      "loss": 1.3626,
+      "step": 379600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1933713913866542e-05,
+      "loss": 1.3663,
+      "step": 379700
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1926319214387128e-05,
+      "loss": 1.36,
+      "step": 379800
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1918924514907717e-05,
+      "loss": 1.3745,
+      "step": 379900
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1911529815428303e-05,
+      "loss": 1.3676,
+      "step": 380000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.190413511594889e-05,
+      "loss": 1.3575,
+      "step": 380100
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1896740416469474e-05,
+      "loss": 1.3795,
+      "step": 380200
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1889345716990063e-05,
+      "loss": 1.3607,
+      "step": 380300
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.188195101751065e-05,
+      "loss": 1.3566,
+      "step": 380400
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1874556318031238e-05,
+      "loss": 1.3641,
+      "step": 380500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1867161618551823e-05,
+      "loss": 1.3397,
+      "step": 380600
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1859766919072412e-05,
+      "loss": 1.3481,
+      "step": 380700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1852372219592998e-05,
+      "loss": 1.3568,
+      "step": 380800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.1844977520113584e-05,
+      "loss": 1.3878,
+      "step": 380900
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.183758282063417e-05,
+      "loss": 1.3658,
+      "step": 381000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1830188121154758e-05,
+      "loss": 1.3635,
+      "step": 381100
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1822793421675344e-05,
+      "loss": 1.3748,
+      "step": 381200
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1815398722195933e-05,
+      "loss": 1.3379,
+      "step": 381300
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1808077969711313e-05,
+      "loss": 1.3838,
+      "step": 381400
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.18006832702319e-05,
+      "loss": 1.363,
+      "step": 381500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1793288570752485e-05,
+      "loss": 1.3644,
+      "step": 381600
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1785893871273074e-05,
+      "loss": 1.4008,
+      "step": 381700
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.177849917179366e-05,
+      "loss": 1.3566,
+      "step": 381800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1771104472314245e-05,
+      "loss": 1.3817,
+      "step": 381900
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1763709772834834e-05,
+      "loss": 1.3296,
+      "step": 382000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.175631507335542e-05,
+      "loss": 1.3526,
+      "step": 382100
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1748920373876005e-05,
+      "loss": 1.3616,
+      "step": 382200
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1741525674396594e-05,
+      "loss": 1.3482,
+      "step": 382300
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.173413097491718e-05,
+      "loss": 1.3729,
+      "step": 382400
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.172673627543777e-05,
+      "loss": 1.3498,
+      "step": 382500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1719341575958355e-05,
+      "loss": 1.3323,
+      "step": 382600
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.171194687647894e-05,
+      "loss": 1.3727,
+      "step": 382700
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.170455217699953e-05,
+      "loss": 1.3347,
+      "step": 382800
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.1697157477520115e-05,
+      "loss": 1.3469,
+      "step": 382900
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.16897627780407e-05,
+      "loss": 1.3796,
+      "step": 383000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.168236807856129e-05,
+      "loss": 1.3719,
+      "step": 383100
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1674973379081875e-05,
+      "loss": 1.3529,
+      "step": 383200
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1667578679602464e-05,
+      "loss": 1.3666,
+      "step": 383300
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.166018398012305e-05,
+      "loss": 1.3678,
+      "step": 383400
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1652789280643635e-05,
+      "loss": 1.361,
+      "step": 383500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.1645468528159016e-05,
+      "loss": 1.354,
+      "step": 383600
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.16380738286796e-05,
+      "loss": 1.3837,
+      "step": 383700
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.163067912920019e-05,
+      "loss": 1.3643,
+      "step": 383800
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1623284429720776e-05,
+      "loss": 1.3387,
+      "step": 383900
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1615889730241365e-05,
+      "loss": 1.3837,
+      "step": 384000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.160849503076195e-05,
+      "loss": 1.3352,
+      "step": 384100
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1601100331282537e-05,
+      "loss": 1.3388,
+      "step": 384200
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1593705631803126e-05,
+      "loss": 1.3733,
+      "step": 384300
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.158631093232371e-05,
+      "loss": 1.3625,
+      "step": 384400
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1578916232844297e-05,
+      "loss": 1.3501,
+      "step": 384500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1571521533364886e-05,
+      "loss": 1.3437,
+      "step": 384600
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.156412683388547e-05,
+      "loss": 1.3637,
+      "step": 384700
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1556732134406057e-05,
+      "loss": 1.3936,
+      "step": 384800
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1549337434926646e-05,
+      "loss": 1.3549,
+      "step": 384900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.1541942735447232e-05,
+      "loss": 1.3613,
+      "step": 385000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.153454803596782e-05,
+      "loss": 1.362,
+      "step": 385100
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1527153336488406e-05,
+      "loss": 1.3761,
+      "step": 385200
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1519758637008992e-05,
+      "loss": 1.3901,
+      "step": 385300
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.151236393752958e-05,
+      "loss": 1.3521,
+      "step": 385400
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1504969238050167e-05,
+      "loss": 1.3484,
+      "step": 385500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.1497574538570752e-05,
+      "loss": 1.3832,
+      "step": 385600
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.149017983909134e-05,
+      "loss": 1.3364,
+      "step": 385700
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1482859086606722e-05,
+      "loss": 1.3713,
+      "step": 385800
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1475464387127307e-05,
+      "loss": 1.3254,
+      "step": 385900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1468069687647897e-05,
+      "loss": 1.3668,
+      "step": 386000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1460674988168482e-05,
+      "loss": 1.3247,
+      "step": 386100
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1453280288689068e-05,
+      "loss": 1.3417,
+      "step": 386200
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1445885589209653e-05,
+      "loss": 1.3744,
+      "step": 386300
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1438490889730242e-05,
+      "loss": 1.3603,
+      "step": 386400
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1431096190250828e-05,
+      "loss": 1.3609,
+      "step": 386500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1423701490771417e-05,
+      "loss": 1.3729,
+      "step": 386600
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1416306791292003e-05,
+      "loss": 1.3592,
+      "step": 386700
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.140891209181259e-05,
+      "loss": 1.3698,
+      "step": 386800
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1401517392333177e-05,
+      "loss": 1.3506,
+      "step": 386900
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1394122692853763e-05,
+      "loss": 1.3826,
+      "step": 387000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.138672799337435e-05,
+      "loss": 1.3469,
+      "step": 387100
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1379333293894938e-05,
+      "loss": 1.3606,
+      "step": 387200
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1371938594415523e-05,
+      "loss": 1.3496,
+      "step": 387300
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1364543894936112e-05,
+      "loss": 1.3739,
+      "step": 387400
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1357149195456698e-05,
+      "loss": 1.3686,
+      "step": 387500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1349754495977284e-05,
+      "loss": 1.3555,
+      "step": 387600
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1342359796497873e-05,
+      "loss": 1.3511,
+      "step": 387700
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.133496509701846e-05,
+      "loss": 1.3657,
+      "step": 387800
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1327570397539044e-05,
+      "loss": 1.3599,
+      "step": 387900
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1320175698059633e-05,
+      "loss": 1.3341,
+      "step": 388000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.131278099858022e-05,
+      "loss": 1.3282,
+      "step": 388100
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1305386299100804e-05,
+      "loss": 1.3674,
+      "step": 388200
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1297991599621393e-05,
+      "loss": 1.3606,
+      "step": 388300
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.129059690014198e-05,
+      "loss": 1.3486,
+      "step": 388400
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1283202200662568e-05,
+      "loss": 1.3482,
+      "step": 388500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1275807501183154e-05,
+      "loss": 1.3463,
+      "step": 388600
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1268412801703743e-05,
+      "loss": 1.3668,
+      "step": 388700
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1261018102224325e-05,
+      "loss": 1.3637,
+      "step": 388800
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1253697349739705e-05,
+      "loss": 1.3715,
+      "step": 388900
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1246302650260294e-05,
+      "loss": 1.3766,
+      "step": 389000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.123890795078088e-05,
+      "loss": 1.3801,
+      "step": 389100
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.123151325130147e-05,
+      "loss": 1.372,
+      "step": 389200
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1224118551822055e-05,
+      "loss": 1.3755,
+      "step": 389300
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1216723852342644e-05,
+      "loss": 1.3414,
+      "step": 389400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.120932915286323e-05,
+      "loss": 1.3542,
+      "step": 389500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1201934453383815e-05,
+      "loss": 1.3812,
+      "step": 389600
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.11945397539044e-05,
+      "loss": 1.3531,
+      "step": 389700
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.118714505442499e-05,
+      "loss": 1.3911,
+      "step": 389800
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1179750354945575e-05,
+      "loss": 1.3526,
+      "step": 389900
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1172355655466164e-05,
+      "loss": 1.3434,
+      "step": 390000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.116496095598675e-05,
+      "loss": 1.36,
+      "step": 390100
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1157566256507336e-05,
+      "loss": 1.372,
+      "step": 390200
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1150171557027925e-05,
+      "loss": 1.3773,
+      "step": 390300
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.114277685754851e-05,
+      "loss": 1.3466,
+      "step": 390400
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1135382158069096e-05,
+      "loss": 1.3554,
+      "step": 390500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1127987458589685e-05,
+      "loss": 1.3829,
+      "step": 390600
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.112059275911027e-05,
+      "loss": 1.3756,
+      "step": 390700
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1113198059630856e-05,
+      "loss": 1.3478,
+      "step": 390800
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1105803360151445e-05,
+      "loss": 1.3361,
+      "step": 390900
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1098482607666826e-05,
+      "loss": 1.3957,
+      "step": 391000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.109108790818741e-05,
+      "loss": 1.3728,
+      "step": 391100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1083693208708e-05,
+      "loss": 1.375,
+      "step": 391200
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1076298509228586e-05,
+      "loss": 1.3352,
+      "step": 391300
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.106890380974917e-05,
+      "loss": 1.3668,
+      "step": 391400
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1061509110269757e-05,
+      "loss": 1.3691,
+      "step": 391500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1054114410790346e-05,
+      "loss": 1.3457,
+      "step": 391600
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1046719711310932e-05,
+      "loss": 1.3506,
+      "step": 391700
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.103932501183152e-05,
+      "loss": 1.357,
+      "step": 391800
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1031930312352107e-05,
+      "loss": 1.3642,
+      "step": 391900
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1024535612872696e-05,
+      "loss": 1.341,
+      "step": 392000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.101714091339328e-05,
+      "loss": 1.3864,
+      "step": 392100
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1009746213913867e-05,
+      "loss": 1.3601,
+      "step": 392200
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.1002351514434452e-05,
+      "loss": 1.3812,
+      "step": 392300
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.099495681495504e-05,
+      "loss": 1.3769,
+      "step": 392400
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.0987562115475627e-05,
+      "loss": 1.3615,
+      "step": 392500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0980167415996216e-05,
+      "loss": 1.3577,
+      "step": 392600
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0972772716516802e-05,
+      "loss": 1.3689,
+      "step": 392700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0965378017037387e-05,
+      "loss": 1.3629,
+      "step": 392800
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0957983317557976e-05,
+      "loss": 1.3564,
+      "step": 392900
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0950588618078562e-05,
+      "loss": 1.351,
+      "step": 393000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.0943193918599148e-05,
+      "loss": 1.3678,
+      "step": 393100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.093587316611453e-05,
+      "loss": 1.3662,
+      "step": 393200
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0928478466635117e-05,
+      "loss": 1.3438,
+      "step": 393300
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0921083767155703e-05,
+      "loss": 1.3553,
+      "step": 393400
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.091368906767629e-05,
+      "loss": 1.3524,
+      "step": 393500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0906294368196878e-05,
+      "loss": 1.3671,
+      "step": 393600
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0898899668717463e-05,
+      "loss": 1.3646,
+      "step": 393700
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0891504969238052e-05,
+      "loss": 1.3618,
+      "step": 393800
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0884110269758638e-05,
+      "loss": 1.3627,
+      "step": 393900
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0876715570279227e-05,
+      "loss": 1.3611,
+      "step": 394000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.086932087079981e-05,
+      "loss": 1.3712,
+      "step": 394100
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0861926171320398e-05,
+      "loss": 1.3859,
+      "step": 394200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0854531471840984e-05,
+      "loss": 1.3838,
+      "step": 394300
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0847136772361573e-05,
+      "loss": 1.3674,
+      "step": 394400
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.083974207288216e-05,
+      "loss": 1.3708,
+      "step": 394500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0832347373402747e-05,
+      "loss": 1.3507,
+      "step": 394600
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.082495267392333e-05,
+      "loss": 1.3661,
+      "step": 394700
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.081755797444392e-05,
+      "loss": 1.3538,
+      "step": 394800
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0810163274964504e-05,
+      "loss": 1.3354,
+      "step": 394900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0802768575485093e-05,
+      "loss": 1.3672,
+      "step": 395000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.079537387600568e-05,
+      "loss": 1.379,
+      "step": 395100
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.0787979176526268e-05,
+      "loss": 1.3556,
+      "step": 395200
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.078065842404165e-05,
+      "loss": 1.3433,
+      "step": 395300
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0773263724562234e-05,
+      "loss": 1.3462,
+      "step": 395400
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.076586902508282e-05,
+      "loss": 1.3458,
+      "step": 395500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.075847432560341e-05,
+      "loss": 1.3521,
+      "step": 395600
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0751079626123994e-05,
+      "loss": 1.3496,
+      "step": 395700
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.0743684926644583e-05,
+      "loss": 1.3636,
+      "step": 395800
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.073629022716517e-05,
+      "loss": 1.3717,
+      "step": 395900
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0728895527685758e-05,
+      "loss": 1.3657,
+      "step": 396000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.072150082820634e-05,
+      "loss": 1.3682,
+      "step": 396100
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.071410612872693e-05,
+      "loss": 1.3621,
+      "step": 396200
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0706711429247515e-05,
+      "loss": 1.3663,
+      "step": 396300
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.0699316729768104e-05,
+      "loss": 1.3633,
+      "step": 396400
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.069192203028869e-05,
+      "loss": 1.351,
+      "step": 396500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.068452733080928e-05,
+      "loss": 1.3534,
+      "step": 396600
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.067713263132986e-05,
+      "loss": 1.3563,
+      "step": 396700
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.066973793185045e-05,
+      "loss": 1.3717,
+      "step": 396800
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0662343232371036e-05,
+      "loss": 1.3717,
+      "step": 396900
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0654948532891625e-05,
+      "loss": 1.3533,
+      "step": 397000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.064755383341221e-05,
+      "loss": 1.3486,
+      "step": 397100
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.06401591339328e-05,
+      "loss": 1.3755,
+      "step": 397200
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.063283838144818e-05,
+      "loss": 1.3437,
+      "step": 397300
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0625443681968765e-05,
+      "loss": 1.3579,
+      "step": 397400
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.061804898248935e-05,
+      "loss": 1.3634,
+      "step": 397500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.061065428300994e-05,
+      "loss": 1.3371,
+      "step": 397600
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0603259583530526e-05,
+      "loss": 1.348,
+      "step": 397700
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0595864884051115e-05,
+      "loss": 1.3728,
+      "step": 397800
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.05884701845717e-05,
+      "loss": 1.3601,
+      "step": 397900
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.058107548509229e-05,
+      "loss": 1.3654,
+      "step": 398000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.057368078561287e-05,
+      "loss": 1.3873,
+      "step": 398100
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.056628608613346e-05,
+      "loss": 1.3464,
+      "step": 398200
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0558891386654046e-05,
+      "loss": 1.3842,
+      "step": 398300
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0551496687174635e-05,
+      "loss": 1.3672,
+      "step": 398400
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.054410198769522e-05,
+      "loss": 1.3623,
+      "step": 398500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.053670728821581e-05,
+      "loss": 1.3657,
+      "step": 398600
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0529312588736392e-05,
+      "loss": 1.3536,
+      "step": 398700
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.052191788925698e-05,
+      "loss": 1.342,
+      "step": 398800
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0514523189777567e-05,
+      "loss": 1.3739,
+      "step": 398900
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0507128490298156e-05,
+      "loss": 1.354,
+      "step": 399000
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.049973379081874e-05,
+      "loss": 1.3457,
+      "step": 399100
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.049233909133933e-05,
+      "loss": 1.3563,
+      "step": 399200
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0484944391859916e-05,
+      "loss": 1.3486,
+      "step": 399300
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0477549692380502e-05,
+      "loss": 1.3496,
+      "step": 399400
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0470228939895882e-05,
+      "loss": 1.3916,
+      "step": 399500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.046283424041647e-05,
+      "loss": 1.3605,
+      "step": 399600
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0455439540937057e-05,
+      "loss": 1.3406,
+      "step": 399700
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0448044841457646e-05,
+      "loss": 1.3797,
+      "step": 399800
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.044065014197823e-05,
+      "loss": 1.3541,
+      "step": 399900
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0433255442498817e-05,
+      "loss": 1.3753,
+      "step": 400000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0425860743019403e-05,
+      "loss": 1.3537,
+      "step": 400100
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0418466043539992e-05,
+      "loss": 1.3532,
+      "step": 400200
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0411071344060578e-05,
+      "loss": 1.3417,
+      "step": 400300
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0403676644581167e-05,
+      "loss": 1.3837,
+      "step": 400400
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0396281945101752e-05,
+      "loss": 1.3718,
+      "step": 400500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.038888724562234e-05,
+      "loss": 1.3726,
+      "step": 400600
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0381492546142924e-05,
+      "loss": 1.3725,
+      "step": 400700
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0374097846663513e-05,
+      "loss": 1.3595,
+      "step": 400800
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0366703147184098e-05,
+      "loss": 1.3646,
+      "step": 400900
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0359308447704687e-05,
+      "loss": 1.3744,
+      "step": 401000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0351913748225273e-05,
+      "loss": 1.355,
+      "step": 401100
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0344519048745862e-05,
+      "loss": 1.3529,
+      "step": 401200
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0337124349266448e-05,
+      "loss": 1.346,
+      "step": 401300
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0329729649787033e-05,
+      "loss": 1.3438,
+      "step": 401400
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.032233495030762e-05,
+      "loss": 1.3379,
+      "step": 401500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0315014197823003e-05,
+      "loss": 1.3548,
+      "step": 401600
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0307619498343588e-05,
+      "loss": 1.3422,
+      "step": 401700
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0300224798864177e-05,
+      "loss": 1.38,
+      "step": 401800
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0292830099384763e-05,
+      "loss": 1.343,
+      "step": 401900
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.028543539990535e-05,
+      "loss": 1.3457,
+      "step": 402000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0278040700425934e-05,
+      "loss": 1.3529,
+      "step": 402100
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0270646000946523e-05,
+      "loss": 1.3702,
+      "step": 402200
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.026325130146711e-05,
+      "loss": 1.3999,
+      "step": 402300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0255856601987698e-05,
+      "loss": 1.3891,
+      "step": 402400
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0248461902508284e-05,
+      "loss": 1.3632,
+      "step": 402500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0241067203028873e-05,
+      "loss": 1.357,
+      "step": 402600
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0233672503549455e-05,
+      "loss": 1.3587,
+      "step": 402700
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0226277804070044e-05,
+      "loss": 1.3786,
+      "step": 402800
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.021888310459063e-05,
+      "loss": 1.3533,
+      "step": 402900
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.021148840511122e-05,
+      "loss": 1.3539,
+      "step": 403000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0204093705631804e-05,
+      "loss": 1.3745,
+      "step": 403100
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0196699006152393e-05,
+      "loss": 1.366,
+      "step": 403200
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0189304306672975e-05,
+      "loss": 1.355,
+      "step": 403300
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0181909607193564e-05,
+      "loss": 1.3504,
+      "step": 403400
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.017451490771415e-05,
+      "loss": 1.3413,
+      "step": 403500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.016712020823474e-05,
+      "loss": 1.3297,
+      "step": 403600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.015979945575012e-05,
+      "loss": 1.3773,
+      "step": 403700
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0152404756270705e-05,
+      "loss": 1.3612,
+      "step": 403800
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.0145010056791294e-05,
+      "loss": 1.3555,
+      "step": 403900
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.013761535731188e-05,
+      "loss": 1.3575,
+      "step": 404000
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0130220657832466e-05,
+      "loss": 1.3592,
+      "step": 404100
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0122825958353055e-05,
+      "loss": 1.3643,
+      "step": 404200
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.011543125887364e-05,
+      "loss": 1.3626,
+      "step": 404300
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.010803655939423e-05,
+      "loss": 1.379,
+      "step": 404400
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0100641859914815e-05,
+      "loss": 1.36,
+      "step": 404500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0093247160435404e-05,
+      "loss": 1.3551,
+      "step": 404600
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0085852460955986e-05,
+      "loss": 1.3859,
+      "step": 404700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0078457761476575e-05,
+      "loss": 1.3758,
+      "step": 404800
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.007106306199716e-05,
+      "loss": 1.3323,
+      "step": 404900
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.006366836251775e-05,
+      "loss": 1.3705,
+      "step": 405000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0056273663038335e-05,
+      "loss": 1.3614,
+      "step": 405100
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0048878963558924e-05,
+      "loss": 1.3642,
+      "step": 405200
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0041484264079507e-05,
+      "loss": 1.3397,
+      "step": 405300
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0034089564600096e-05,
+      "loss": 1.3774,
+      "step": 405400
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.002669486512068e-05,
+      "loss": 1.3685,
+      "step": 405500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.001930016564127e-05,
+      "loss": 1.3352,
+      "step": 405600
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0011905466161856e-05,
+      "loss": 1.3751,
+      "step": 405700
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.0004584713677237e-05,
+      "loss": 1.3057,
+      "step": 405800
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.9997190014197826e-05,
+      "loss": 1.3074,
+      "step": 405900
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.998979531471841e-05,
+      "loss": 1.3093,
+      "step": 406000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9982400615238997e-05,
+      "loss": 1.297,
+      "step": 406100
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9975005915759586e-05,
+      "loss": 1.3071,
+      "step": 406200
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.996761121628017e-05,
+      "loss": 1.3385,
+      "step": 406300
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9960216516800757e-05,
+      "loss": 1.3119,
+      "step": 406400
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9952821817321346e-05,
+      "loss": 1.2962,
+      "step": 406500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9945427117841932e-05,
+      "loss": 1.29,
+      "step": 406600
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9938032418362517e-05,
+      "loss": 1.2855,
+      "step": 406700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9930637718883106e-05,
+      "loss": 1.303,
+      "step": 406800
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9923243019403692e-05,
+      "loss": 1.2951,
+      "step": 406900
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.991584831992428e-05,
+      "loss": 1.2959,
+      "step": 407000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9908453620444867e-05,
+      "loss": 1.3224,
+      "step": 407100
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9901058920965456e-05,
+      "loss": 1.3235,
+      "step": 407200
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9893664221486038e-05,
+      "loss": 1.2923,
+      "step": 407300
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9886269522006627e-05,
+      "loss": 1.3111,
+      "step": 407400
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9878874822527213e-05,
+      "loss": 1.2695,
+      "step": 407500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9871480123047802e-05,
+      "loss": 1.3087,
+      "step": 407600
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9864085423568387e-05,
+      "loss": 1.3093,
+      "step": 407700
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9856690724088976e-05,
+      "loss": 1.3171,
+      "step": 407800
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9849369971604357e-05,
+      "loss": 1.3192,
+      "step": 407900
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9841975272124942e-05,
+      "loss": 1.3245,
+      "step": 408000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9834580572645528e-05,
+      "loss": 1.3174,
+      "step": 408100
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9827185873166114e-05,
+      "loss": 1.31,
+      "step": 408200
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9819791173686703e-05,
+      "loss": 1.3185,
+      "step": 408300
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.981239647420729e-05,
+      "loss": 1.3052,
+      "step": 408400
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9805001774727877e-05,
+      "loss": 1.3141,
+      "step": 408500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.9797607075248463e-05,
+      "loss": 1.2849,
+      "step": 408600
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.979021237576905e-05,
+      "loss": 1.3123,
+      "step": 408700
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9782817676289638e-05,
+      "loss": 1.3189,
+      "step": 408800
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9775422976810223e-05,
+      "loss": 1.2995,
+      "step": 408900
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.976802827733081e-05,
+      "loss": 1.3152,
+      "step": 409000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9760633577851398e-05,
+      "loss": 1.307,
+      "step": 409100
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9753238878371984e-05,
+      "loss": 1.3232,
+      "step": 409200
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.974584417889257e-05,
+      "loss": 1.3295,
+      "step": 409300
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.973844947941316e-05,
+      "loss": 1.2697,
+      "step": 409400
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9731054779933744e-05,
+      "loss": 1.2815,
+      "step": 409500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9723660080454333e-05,
+      "loss": 1.2953,
+      "step": 409600
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.971626538097492e-05,
+      "loss": 1.31,
+      "step": 409700
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.9708870681495508e-05,
+      "loss": 1.3144,
+      "step": 409800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.970147598201609e-05,
+      "loss": 1.2992,
+      "step": 409900
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.969408128253668e-05,
+      "loss": 1.3126,
+      "step": 410000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9686686583057265e-05,
+      "loss": 1.3389,
+      "step": 410100
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9679291883577854e-05,
+      "loss": 1.3095,
+      "step": 410200
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9671971131093234e-05,
+      "loss": 1.3192,
+      "step": 410300
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.966457643161382e-05,
+      "loss": 1.3074,
+      "step": 410400
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.965718173213441e-05,
+      "loss": 1.295,
+      "step": 410500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9649787032654994e-05,
+      "loss": 1.3407,
+      "step": 410600
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.964239233317558e-05,
+      "loss": 1.3018,
+      "step": 410700
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9634997633696166e-05,
+      "loss": 1.2903,
+      "step": 410800
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9627602934216755e-05,
+      "loss": 1.2941,
+      "step": 410900
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.962020823473734e-05,
+      "loss": 1.3169,
+      "step": 411000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.961281353525793e-05,
+      "loss": 1.3098,
+      "step": 411100
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.9605418835778515e-05,
+      "loss": 1.3229,
+      "step": 411200
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.95980241362991e-05,
+      "loss": 1.3166,
+      "step": 411300
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.959062943681969e-05,
+      "loss": 1.3148,
+      "step": 411400
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9583234737340275e-05,
+      "loss": 1.3592,
+      "step": 411500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.957584003786086e-05,
+      "loss": 1.2856,
+      "step": 411600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.956844533838145e-05,
+      "loss": 1.2963,
+      "step": 411700
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9561050638902036e-05,
+      "loss": 1.2959,
+      "step": 411800
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.955365593942262e-05,
+      "loss": 1.3177,
+      "step": 411900
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.954626123994321e-05,
+      "loss": 1.2948,
+      "step": 412000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9538866540463796e-05,
+      "loss": 1.2978,
+      "step": 412100
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9531471840984385e-05,
+      "loss": 1.29,
+      "step": 412200
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9524151088499765e-05,
+      "loss": 1.3074,
+      "step": 412300
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.951675638902035e-05,
+      "loss": 1.3061,
+      "step": 412400
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.950936168954094e-05,
+      "loss": 1.3222,
+      "step": 412500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9501966990061522e-05,
+      "loss": 1.3071,
+      "step": 412600
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.949457229058211e-05,
+      "loss": 1.3076,
+      "step": 412700
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9487177591102697e-05,
+      "loss": 1.3064,
+      "step": 412800
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9479782891623286e-05,
+      "loss": 1.2984,
+      "step": 412900
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.947238819214387e-05,
+      "loss": 1.301,
+      "step": 413000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.946499349266446e-05,
+      "loss": 1.3197,
+      "step": 413100
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9457598793185043e-05,
+      "loss": 1.2934,
+      "step": 413200
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9450204093705632e-05,
+      "loss": 1.3354,
+      "step": 413300
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9442809394226218e-05,
+      "loss": 1.3234,
+      "step": 413400
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9435414694746807e-05,
+      "loss": 1.3227,
+      "step": 413500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9428019995267392e-05,
+      "loss": 1.2968,
+      "step": 413600
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.942062529578798e-05,
+      "loss": 1.3243,
+      "step": 413700
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9413230596308567e-05,
+      "loss": 1.2985,
+      "step": 413800
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9405835896829152e-05,
+      "loss": 1.3337,
+      "step": 413900
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.939844119734974e-05,
+      "loss": 1.2948,
+      "step": 414000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.9391046497870327e-05,
+      "loss": 1.3173,
+      "step": 414100
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9383651798390913e-05,
+      "loss": 1.3159,
+      "step": 414200
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9376331045906297e-05,
+      "loss": 1.2999,
+      "step": 414300
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9368936346426882e-05,
+      "loss": 1.305,
+      "step": 414400
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.936154164694747e-05,
+      "loss": 1.3407,
+      "step": 414500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9354146947468054e-05,
+      "loss": 1.3466,
+      "step": 414600
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9346752247988643e-05,
+      "loss": 1.3178,
+      "step": 414700
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.9339357548509228e-05,
+      "loss": 1.3016,
+      "step": 414800
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9331962849029817e-05,
+      "loss": 1.3056,
+      "step": 414900
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9324568149550403e-05,
+      "loss": 1.3012,
+      "step": 415000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9317173450070992e-05,
+      "loss": 1.3319,
+      "step": 415100
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9309778750591574e-05,
+      "loss": 1.3177,
+      "step": 415200
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9302384051112163e-05,
+      "loss": 1.3214,
+      "step": 415300
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.929498935163275e-05,
+      "loss": 1.3224,
+      "step": 415400
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9287594652153338e-05,
+      "loss": 1.3119,
+      "step": 415500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9280199952673923e-05,
+      "loss": 1.3042,
+      "step": 415600
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9272805253194512e-05,
+      "loss": 1.3209,
+      "step": 415700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9265410553715098e-05,
+      "loss": 1.2965,
+      "step": 415800
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.9258015854235684e-05,
+      "loss": 1.3169,
+      "step": 415900
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.925062115475627e-05,
+      "loss": 1.3387,
+      "step": 416000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.924322645527686e-05,
+      "loss": 1.2908,
+      "step": 416100
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9235831755797444e-05,
+      "loss": 1.287,
+      "step": 416200
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9228437056318033e-05,
+      "loss": 1.2972,
+      "step": 416300
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.922104235683862e-05,
+      "loss": 1.31,
+      "step": 416400
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9213647657359204e-05,
+      "loss": 1.3056,
+      "step": 416500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9206326904874585e-05,
+      "loss": 1.2996,
+      "step": 416600
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.9198932205395174e-05,
+      "loss": 1.3038,
+      "step": 416700
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.919153750591576e-05,
+      "loss": 1.3108,
+      "step": 416800
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.918414280643635e-05,
+      "loss": 1.331,
+      "step": 416900
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9176748106956934e-05,
+      "loss": 1.3094,
+      "step": 417000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9169353407477523e-05,
+      "loss": 1.3322,
+      "step": 417100
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9161958707998105e-05,
+      "loss": 1.2981,
+      "step": 417200
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9154564008518694e-05,
+      "loss": 1.3286,
+      "step": 417300
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.914716930903928e-05,
+      "loss": 1.3048,
+      "step": 417400
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.913977460955987e-05,
+      "loss": 1.3199,
+      "step": 417500
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9132379910080455e-05,
+      "loss": 1.3247,
+      "step": 417600
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9124985210601044e-05,
+      "loss": 1.3317,
+      "step": 417700
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.911759051112163e-05,
+      "loss": 1.3038,
+      "step": 417800
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9110195811642215e-05,
+      "loss": 1.3216,
+      "step": 417900
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.91028011121628e-05,
+      "loss": 1.3252,
+      "step": 418000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.909540641268339e-05,
+      "loss": 1.3289,
+      "step": 418100
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9088011713203975e-05,
+      "loss": 1.316,
+      "step": 418200
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9080617013724564e-05,
+      "loss": 1.3306,
+      "step": 418300
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.907322231424515e-05,
+      "loss": 1.3096,
+      "step": 418400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9065827614765736e-05,
+      "loss": 1.3107,
+      "step": 418500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.905843291528632e-05,
+      "loss": 1.3004,
+      "step": 418600
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.905103821580691e-05,
+      "loss": 1.3107,
+      "step": 418700
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9043643516327496e-05,
+      "loss": 1.3253,
+      "step": 418800
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.903632276384288e-05,
+      "loss": 1.3182,
+      "step": 418900
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9028928064363465e-05,
+      "loss": 1.3144,
+      "step": 419000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9021533364884054e-05,
+      "loss": 1.324,
+      "step": 419100
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9014138665404637e-05,
+      "loss": 1.3132,
+      "step": 419200
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.9006743965925226e-05,
+      "loss": 1.3285,
+      "step": 419300
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.899934926644581e-05,
+      "loss": 1.304,
+      "step": 419400
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.89919545669664e-05,
+      "loss": 1.3236,
+      "step": 419500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8984559867486986e-05,
+      "loss": 1.319,
+      "step": 419600
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8977165168007575e-05,
+      "loss": 1.3406,
+      "step": 419700
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.896977046852816e-05,
+      "loss": 1.31,
+      "step": 419800
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8962375769048746e-05,
+      "loss": 1.3161,
+      "step": 419900
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8954981069569332e-05,
+      "loss": 1.3136,
+      "step": 420000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.894758637008992e-05,
+      "loss": 1.3108,
+      "step": 420100
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8940191670610507e-05,
+      "loss": 1.3159,
+      "step": 420200
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8932796971131096e-05,
+      "loss": 1.295,
+      "step": 420300
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.892540227165168e-05,
+      "loss": 1.289,
+      "step": 420400
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8918007572172267e-05,
+      "loss": 1.3052,
+      "step": 420500
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8910612872692853e-05,
+      "loss": 1.3103,
+      "step": 420600
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.890321817321344e-05,
+      "loss": 1.3039,
+      "step": 420700
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.8895823473734027e-05,
+      "loss": 1.3267,
+      "step": 420800
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.888850272124941e-05,
+      "loss": 1.3086,
+      "step": 420900
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8881108021769997e-05,
+      "loss": 1.3426,
+      "step": 421000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8873713322290586e-05,
+      "loss": 1.2891,
+      "step": 421100
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8866318622811168e-05,
+      "loss": 1.3103,
+      "step": 421200
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8858923923331757e-05,
+      "loss": 1.2971,
+      "step": 421300
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.8851529223852343e-05,
+      "loss": 1.307,
+      "step": 421400
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.884413452437293e-05,
+      "loss": 1.3149,
+      "step": 421500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8836739824893517e-05,
+      "loss": 1.3105,
+      "step": 421600
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8829345125414106e-05,
+      "loss": 1.3224,
+      "step": 421700
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.882195042593469e-05,
+      "loss": 1.3207,
+      "step": 421800
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8814555726455278e-05,
+      "loss": 1.3008,
+      "step": 421900
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8807161026975863e-05,
+      "loss": 1.3036,
+      "step": 422000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8799766327496452e-05,
+      "loss": 1.3577,
+      "step": 422100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8792371628017038e-05,
+      "loss": 1.3085,
+      "step": 422200
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8784976928537627e-05,
+      "loss": 1.3151,
+      "step": 422300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8777582229058213e-05,
+      "loss": 1.3044,
+      "step": 422400
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8770187529578798e-05,
+      "loss": 1.3218,
+      "step": 422500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8762792830099384e-05,
+      "loss": 1.3193,
+      "step": 422600
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8755398130619973e-05,
+      "loss": 1.3035,
+      "step": 422700
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.874800343114056e-05,
+      "loss": 1.3294,
+      "step": 422800
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8740608731661148e-05,
+      "loss": 1.3061,
+      "step": 422900
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8733214032181733e-05,
+      "loss": 1.3061,
+      "step": 423000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8725893279697117e-05,
+      "loss": 1.3326,
+      "step": 423100
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.87184985802177e-05,
+      "loss": 1.3122,
+      "step": 423200
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8711103880738288e-05,
+      "loss": 1.2883,
+      "step": 423300
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8703709181258874e-05,
+      "loss": 1.3216,
+      "step": 423400
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8696314481779463e-05,
+      "loss": 1.3202,
+      "step": 423500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.868891978230005e-05,
+      "loss": 1.3252,
+      "step": 423600
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8681525082820638e-05,
+      "loss": 1.3244,
+      "step": 423700
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.867413038334122e-05,
+      "loss": 1.3235,
+      "step": 423800
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.866673568386181e-05,
+      "loss": 1.3133,
+      "step": 423900
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8659340984382395e-05,
+      "loss": 1.3134,
+      "step": 424000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.8651946284902984e-05,
+      "loss": 1.3105,
+      "step": 424100
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.864455158542357e-05,
+      "loss": 1.3372,
+      "step": 424200
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8637156885944158e-05,
+      "loss": 1.2953,
+      "step": 424300
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8629762186464744e-05,
+      "loss": 1.3244,
+      "step": 424400
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.862236748698533e-05,
+      "loss": 1.3118,
+      "step": 424500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8614972787505915e-05,
+      "loss": 1.3281,
+      "step": 424600
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8607578088026504e-05,
+      "loss": 1.3204,
+      "step": 424700
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.860018338854709e-05,
+      "loss": 1.3054,
+      "step": 424800
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.859278868906768e-05,
+      "loss": 1.3257,
+      "step": 424900
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8585393989588264e-05,
+      "loss": 1.3074,
+      "step": 425000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.857799929010885e-05,
+      "loss": 1.2786,
+      "step": 425100
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8570604590629436e-05,
+      "loss": 1.3185,
+      "step": 425200
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.8563209891150025e-05,
+      "loss": 1.3151,
+      "step": 425300
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.855581519167061e-05,
+      "loss": 1.3146,
+      "step": 425400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.85484204921912e-05,
+      "loss": 1.3346,
+      "step": 425500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.854109973970658e-05,
+      "loss": 1.3238,
+      "step": 425600
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.853370504022717e-05,
+      "loss": 1.334,
+      "step": 425700
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.852631034074775e-05,
+      "loss": 1.3283,
+      "step": 425800
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.851891564126834e-05,
+      "loss": 1.3043,
+      "step": 425900
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8511520941788926e-05,
+      "loss": 1.33,
+      "step": 426000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8504126242309515e-05,
+      "loss": 1.3048,
+      "step": 426100
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.84967315428301e-05,
+      "loss": 1.3415,
+      "step": 426200
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.848941079034548e-05,
+      "loss": 1.3081,
+      "step": 426300
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.848201609086607e-05,
+      "loss": 1.3149,
+      "step": 426400
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8474621391386656e-05,
+      "loss": 1.2996,
+      "step": 426500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.846722669190724e-05,
+      "loss": 1.3145,
+      "step": 426600
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8459831992427827e-05,
+      "loss": 1.3267,
+      "step": 426700
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8452437292948416e-05,
+      "loss": 1.3134,
+      "step": 426800
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.8445042593469e-05,
+      "loss": 1.3304,
+      "step": 426900
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.843764789398959e-05,
+      "loss": 1.313,
+      "step": 427000
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8430253194510176e-05,
+      "loss": 1.3097,
+      "step": 427100
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8422858495030762e-05,
+      "loss": 1.2905,
+      "step": 427200
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.841546379555135e-05,
+      "loss": 1.335,
+      "step": 427300
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8408069096071937e-05,
+      "loss": 1.3508,
+      "step": 427400
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8400674396592522e-05,
+      "loss": 1.2862,
+      "step": 427500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.839327969711311e-05,
+      "loss": 1.3333,
+      "step": 427600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8385884997633697e-05,
+      "loss": 1.3146,
+      "step": 427700
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8378490298154282e-05,
+      "loss": 1.3535,
+      "step": 427800
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.837109559867487e-05,
+      "loss": 1.3073,
+      "step": 427900
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8363700899195457e-05,
+      "loss": 1.307,
+      "step": 428000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8356306199716046e-05,
+      "loss": 1.3288,
+      "step": 428100
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8348911500236632e-05,
+      "loss": 1.3138,
+      "step": 428200
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.834151680075722e-05,
+      "loss": 1.3284,
+      "step": 428300
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8334122101277803e-05,
+      "loss": 1.3484,
+      "step": 428400
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8326727401798392e-05,
+      "loss": 1.3264,
+      "step": 428500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8319332702318978e-05,
+      "loss": 1.3349,
+      "step": 428600
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8311938002839567e-05,
+      "loss": 1.3181,
+      "step": 428700
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8304543303360152e-05,
+      "loss": 1.3138,
+      "step": 428800
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.829714860388074e-05,
+      "loss": 1.3066,
+      "step": 428900
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8289753904401327e-05,
+      "loss": 1.3207,
+      "step": 429000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8282359204921913e-05,
+      "loss": 1.3024,
+      "step": 429100
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.82749645054425e-05,
+      "loss": 1.3032,
+      "step": 429200
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8267569805963087e-05,
+      "loss": 1.3102,
+      "step": 429300
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8260175106483673e-05,
+      "loss": 1.3163,
+      "step": 429400
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8252780407004262e-05,
+      "loss": 1.3271,
+      "step": 429500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8245385707524848e-05,
+      "loss": 1.3158,
+      "step": 429600
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.8237991008045437e-05,
+      "loss": 1.2956,
+      "step": 429700
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.823059630856602e-05,
+      "loss": 1.2976,
+      "step": 429800
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8223201609086608e-05,
+      "loss": 1.3431,
+      "step": 429900
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8215806909607194e-05,
+      "loss": 1.3298,
+      "step": 430000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8208412210127783e-05,
+      "loss": 1.3158,
+      "step": 430100
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8201017510648368e-05,
+      "loss": 1.3344,
+      "step": 430200
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.819369675816375e-05,
+      "loss": 1.3292,
+      "step": 430300
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8186302058684334e-05,
+      "loss": 1.3028,
+      "step": 430400
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8178907359204923e-05,
+      "loss": 1.3061,
+      "step": 430500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.817151265972551e-05,
+      "loss": 1.3017,
+      "step": 430600
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8164117960246098e-05,
+      "loss": 1.3068,
+      "step": 430700
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8156723260766684e-05,
+      "loss": 1.3143,
+      "step": 430800
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8149328561287273e-05,
+      "loss": 1.3069,
+      "step": 430900
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.814193386180786e-05,
+      "loss": 1.3217,
+      "step": 431000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8134539162328444e-05,
+      "loss": 1.2817,
+      "step": 431100
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.812714446284903e-05,
+      "loss": 1.3094,
+      "step": 431200
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.811974976336962e-05,
+      "loss": 1.3225,
+      "step": 431300
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8112355063890204e-05,
+      "loss": 1.3027,
+      "step": 431400
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8104960364410793e-05,
+      "loss": 1.3292,
+      "step": 431500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.809756566493138e-05,
+      "loss": 1.3392,
+      "step": 431600
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.8090170965451965e-05,
+      "loss": 1.3255,
+      "step": 431700
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.808277626597255e-05,
+      "loss": 1.3341,
+      "step": 431800
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.807538156649314e-05,
+      "loss": 1.3231,
+      "step": 431900
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8067986867013725e-05,
+      "loss": 1.3124,
+      "step": 432000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8060592167534314e-05,
+      "loss": 1.3196,
+      "step": 432100
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.80531974680549e-05,
+      "loss": 1.3053,
+      "step": 432200
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.804587671557028e-05,
+      "loss": 1.3196,
+      "step": 432300
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8038482016090866e-05,
+      "loss": 1.3137,
+      "step": 432400
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8031087316611455e-05,
+      "loss": 1.3438,
+      "step": 432500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.802369261713204e-05,
+      "loss": 1.3149,
+      "step": 432600
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8016297917652626e-05,
+      "loss": 1.3215,
+      "step": 432700
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8008903218173215e-05,
+      "loss": 1.3219,
+      "step": 432800
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.80015085186938e-05,
+      "loss": 1.3019,
+      "step": 432900
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.799411381921439e-05,
+      "loss": 1.3268,
+      "step": 433000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7986719119734975e-05,
+      "loss": 1.3214,
+      "step": 433100
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.797932442025556e-05,
+      "loss": 1.3056,
+      "step": 433200
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.797192972077615e-05,
+      "loss": 1.3126,
+      "step": 433300
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7964535021296736e-05,
+      "loss": 1.3256,
+      "step": 433400
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.795714032181732e-05,
+      "loss": 1.3382,
+      "step": 433500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.794974562233791e-05,
+      "loss": 1.3231,
+      "step": 433600
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7942350922858496e-05,
+      "loss": 1.3147,
+      "step": 433700
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.793495622337908e-05,
+      "loss": 1.3406,
+      "step": 433800
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.792756152389967e-05,
+      "loss": 1.326,
+      "step": 433900
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7920166824420256e-05,
+      "loss": 1.3253,
+      "step": 434000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7912772124940845e-05,
+      "loss": 1.3584,
+      "step": 434100
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.790537742546143e-05,
+      "loss": 1.3207,
+      "step": 434200
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.789805667297681e-05,
+      "loss": 1.3198,
+      "step": 434300
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7890661973497397e-05,
+      "loss": 1.3257,
+      "step": 434400
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7883267274017983e-05,
+      "loss": 1.2902,
+      "step": 434500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.787587257453857e-05,
+      "loss": 1.3508,
+      "step": 434600
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7868477875059157e-05,
+      "loss": 1.3252,
+      "step": 434700
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7861083175579746e-05,
+      "loss": 1.3117,
+      "step": 434800
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7853688476100332e-05,
+      "loss": 1.305,
+      "step": 434900
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7846293776620918e-05,
+      "loss": 1.3227,
+      "step": 435000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7838899077141507e-05,
+      "loss": 1.3127,
+      "step": 435100
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7831504377662092e-05,
+      "loss": 1.3145,
+      "step": 435200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7824109678182678e-05,
+      "loss": 1.3278,
+      "step": 435300
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7816714978703267e-05,
+      "loss": 1.3359,
+      "step": 435400
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7809320279223852e-05,
+      "loss": 1.3127,
+      "step": 435500
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.780192557974444e-05,
+      "loss": 1.3307,
+      "step": 435600
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7794530880265027e-05,
+      "loss": 1.331,
+      "step": 435700
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7787136180785613e-05,
+      "loss": 1.3317,
+      "step": 435800
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7779741481306202e-05,
+      "loss": 1.3251,
+      "step": 435900
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7772346781826787e-05,
+      "loss": 1.3208,
+      "step": 436000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7764952082347373e-05,
+      "loss": 1.33,
+      "step": 436100
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7757557382867962e-05,
+      "loss": 1.3235,
+      "step": 436200
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7750162683388548e-05,
+      "loss": 1.3241,
+      "step": 436300
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7742767983909133e-05,
+      "loss": 1.3134,
+      "step": 436400
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7735373284429722e-05,
+      "loss": 1.3205,
+      "step": 436500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7727978584950308e-05,
+      "loss": 1.3342,
+      "step": 436600
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.772065783246569e-05,
+      "loss": 1.3081,
+      "step": 436700
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7713263132986278e-05,
+      "loss": 1.3024,
+      "step": 436800
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7705868433506863e-05,
+      "loss": 1.312,
+      "step": 436900
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.769847373402745e-05,
+      "loss": 1.3154,
+      "step": 437000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7691079034548034e-05,
+      "loss": 1.3309,
+      "step": 437100
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7683684335068623e-05,
+      "loss": 1.3259,
+      "step": 437200
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.767628963558921e-05,
+      "loss": 1.3199,
+      "step": 437300
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7668894936109798e-05,
+      "loss": 1.3255,
+      "step": 437400
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7661500236630384e-05,
+      "loss": 1.3273,
+      "step": 437500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7654105537150973e-05,
+      "loss": 1.3096,
+      "step": 437600
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.764671083767156e-05,
+      "loss": 1.3248,
+      "step": 437700
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7639316138192144e-05,
+      "loss": 1.3265,
+      "step": 437800
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.763192143871273e-05,
+      "loss": 1.3172,
+      "step": 437900
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.762452673923332e-05,
+      "loss": 1.2988,
+      "step": 438000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7617132039753904e-05,
+      "loss": 1.3203,
+      "step": 438100
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7609737340274493e-05,
+      "loss": 1.299,
+      "step": 438200
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.760234264079508e-05,
+      "loss": 1.3233,
+      "step": 438300
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7594947941315665e-05,
+      "loss": 1.3057,
+      "step": 438400
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7587553241836254e-05,
+      "loss": 1.3306,
+      "step": 438500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.758015854235684e-05,
+      "loss": 1.3206,
+      "step": 438600
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7572763842877425e-05,
+      "loss": 1.3287,
+      "step": 438700
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7565369143398014e-05,
+      "loss": 1.3222,
+      "step": 438800
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7558048390913394e-05,
+      "loss": 1.3195,
+      "step": 438900
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.755065369143398e-05,
+      "loss": 1.3377,
+      "step": 439000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.7543258991954566e-05,
+      "loss": 1.3091,
+      "step": 439100
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7535864292475155e-05,
+      "loss": 1.322,
+      "step": 439200
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.752846959299574e-05,
+      "loss": 1.3038,
+      "step": 439300
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.752107489351633e-05,
+      "loss": 1.3374,
+      "step": 439400
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7513680194036915e-05,
+      "loss": 1.3355,
+      "step": 439500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7506285494557504e-05,
+      "loss": 1.3014,
+      "step": 439600
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7498890795078086e-05,
+      "loss": 1.3244,
+      "step": 439700
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.7491496095598675e-05,
+      "loss": 1.3296,
+      "step": 439800
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.748410139611926e-05,
+      "loss": 1.3198,
+      "step": 439900
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.747670669663985e-05,
+      "loss": 1.3444,
+      "step": 440000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7469311997160436e-05,
+      "loss": 1.3117,
+      "step": 440100
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7461917297681025e-05,
+      "loss": 1.3113,
+      "step": 440200
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.745452259820161e-05,
+      "loss": 1.3096,
+      "step": 440300
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7447127898722196e-05,
+      "loss": 1.3078,
+      "step": 440400
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.743973319924278e-05,
+      "loss": 1.3229,
+      "step": 440500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.743233849976337e-05,
+      "loss": 1.3469,
+      "step": 440600
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7424943800283956e-05,
+      "loss": 1.2954,
+      "step": 440700
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7417549100804545e-05,
+      "loss": 1.3156,
+      "step": 440800
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.741015440132513e-05,
+      "loss": 1.3009,
+      "step": 440900
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7402759701845717e-05,
+      "loss": 1.3199,
+      "step": 441000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.7395365002366306e-05,
+      "loss": 1.3091,
+      "step": 441100
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.738797030288689e-05,
+      "loss": 1.3387,
+      "step": 441200
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7380575603407477e-05,
+      "loss": 1.35,
+      "step": 441300
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.737325485092286e-05,
+      "loss": 1.3129,
+      "step": 441400
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7365860151443446e-05,
+      "loss": 1.3392,
+      "step": 441500
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7358465451964035e-05,
+      "loss": 1.3075,
+      "step": 441600
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7351070752484618e-05,
+      "loss": 1.2911,
+      "step": 441700
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.7343676053005207e-05,
+      "loss": 1.3145,
+      "step": 441800
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7336281353525792e-05,
+      "loss": 1.3285,
+      "step": 441900
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.732888665404638e-05,
+      "loss": 1.3345,
+      "step": 442000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7321491954566967e-05,
+      "loss": 1.3333,
+      "step": 442100
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7314097255087556e-05,
+      "loss": 1.3261,
+      "step": 442200
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7306702555608138e-05,
+      "loss": 1.3124,
+      "step": 442300
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7299307856128727e-05,
+      "loss": 1.3013,
+      "step": 442400
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.7291913156649313e-05,
+      "loss": 1.3096,
+      "step": 442500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7284518457169902e-05,
+      "loss": 1.3168,
+      "step": 442600
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7277123757690488e-05,
+      "loss": 1.3436,
+      "step": 442700
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7269729058211077e-05,
+      "loss": 1.3138,
+      "step": 442800
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7262334358731662e-05,
+      "loss": 1.3065,
+      "step": 442900
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7254939659252248e-05,
+      "loss": 1.3174,
+      "step": 443000
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7247544959772833e-05,
+      "loss": 1.3175,
+      "step": 443100
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7240150260293422e-05,
+      "loss": 1.3192,
+      "step": 443200
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7232755560814008e-05,
+      "loss": 1.3283,
+      "step": 443300
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7225434808329392e-05,
+      "loss": 1.3126,
+      "step": 443400
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7218040108849978e-05,
+      "loss": 1.3312,
+      "step": 443500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7210645409370563e-05,
+      "loss": 1.311,
+      "step": 443600
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.720325070989115e-05,
+      "loss": 1.3149,
+      "step": 443700
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7195856010411738e-05,
+      "loss": 1.3187,
+      "step": 443800
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7188461310932324e-05,
+      "loss": 1.3222,
+      "step": 443900
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7181066611452913e-05,
+      "loss": 1.3159,
+      "step": 444000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7173671911973498e-05,
+      "loss": 1.3199,
+      "step": 444100
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7166277212494087e-05,
+      "loss": 1.3582,
+      "step": 444200
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.715888251301467e-05,
+      "loss": 1.3554,
+      "step": 444300
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.715148781353526e-05,
+      "loss": 1.2932,
+      "step": 444400
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7144093114055844e-05,
+      "loss": 1.3013,
+      "step": 444500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7136698414576433e-05,
+      "loss": 1.3086,
+      "step": 444600
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.712930371509702e-05,
+      "loss": 1.317,
+      "step": 444700
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7121909015617608e-05,
+      "loss": 1.3181,
+      "step": 444800
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7114514316138193e-05,
+      "loss": 1.3348,
+      "step": 444900
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.710711961665878e-05,
+      "loss": 1.3014,
+      "step": 445000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7099724917179365e-05,
+      "loss": 1.3325,
+      "step": 445100
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7092330217699954e-05,
+      "loss": 1.3484,
+      "step": 445200
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.708493551822054e-05,
+      "loss": 1.3093,
+      "step": 445300
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7077614765735923e-05,
+      "loss": 1.3348,
+      "step": 445400
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.707022006625651e-05,
+      "loss": 1.3234,
+      "step": 445500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7062825366777095e-05,
+      "loss": 1.32,
+      "step": 445600
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.705543066729768e-05,
+      "loss": 1.3341,
+      "step": 445700
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.704803596781827e-05,
+      "loss": 1.3174,
+      "step": 445800
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7040641268338855e-05,
+      "loss": 1.3154,
+      "step": 445900
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7033246568859444e-05,
+      "loss": 1.3126,
+      "step": 446000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.702585186938003e-05,
+      "loss": 1.311,
+      "step": 446100
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.701845716990062e-05,
+      "loss": 1.2956,
+      "step": 446200
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.70110624704212e-05,
+      "loss": 1.3377,
+      "step": 446300
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.700366777094179e-05,
+      "loss": 1.3328,
+      "step": 446400
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.6996273071462375e-05,
+      "loss": 1.3112,
+      "step": 446500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.6988878371982964e-05,
+      "loss": 1.3229,
+      "step": 446600
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.698148367250355e-05,
+      "loss": 1.3158,
+      "step": 446700
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.697408897302414e-05,
+      "loss": 1.32,
+      "step": 446800
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.696669427354472e-05,
+      "loss": 1.3164,
+      "step": 446900
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.695929957406531e-05,
+      "loss": 1.3064,
+      "step": 447000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6951904874585896e-05,
+      "loss": 1.3424,
+      "step": 447100
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6944510175106485e-05,
+      "loss": 1.3339,
+      "step": 447200
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.693711547562707e-05,
+      "loss": 1.3187,
+      "step": 447300
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.692972077614766e-05,
+      "loss": 1.312,
+      "step": 447400
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6922326076668245e-05,
+      "loss": 1.3285,
+      "step": 447500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6915005324183626e-05,
+      "loss": 1.3107,
+      "step": 447600
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.690761062470421e-05,
+      "loss": 1.3171,
+      "step": 447700
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.69002159252248e-05,
+      "loss": 1.3338,
+      "step": 447800
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6892821225745386e-05,
+      "loss": 1.3284,
+      "step": 447900
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6885426526265975e-05,
+      "loss": 1.3153,
+      "step": 448000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.687803182678656e-05,
+      "loss": 1.2977,
+      "step": 448100
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.687063712730715e-05,
+      "loss": 1.326,
+      "step": 448200
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6863242427827732e-05,
+      "loss": 1.3189,
+      "step": 448300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.685584772834832e-05,
+      "loss": 1.3132,
+      "step": 448400
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6848453028868907e-05,
+      "loss": 1.3349,
+      "step": 448500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6841058329389496e-05,
+      "loss": 1.3497,
+      "step": 448600
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.683366362991008e-05,
+      "loss": 1.3184,
+      "step": 448700
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.682626893043067e-05,
+      "loss": 1.3172,
+      "step": 448800
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6818874230951253e-05,
+      "loss": 1.321,
+      "step": 448900
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.681147953147184e-05,
+      "loss": 1.3209,
+      "step": 449000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6804084831992427e-05,
+      "loss": 1.3281,
+      "step": 449100
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6796690132513016e-05,
+      "loss": 1.3115,
+      "step": 449200
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6789295433033602e-05,
+      "loss": 1.3381,
+      "step": 449300
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.678190073355419e-05,
+      "loss": 1.3086,
+      "step": 449400
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6774506034074777e-05,
+      "loss": 1.3236,
+      "step": 449500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6767111334595362e-05,
+      "loss": 1.3444,
+      "step": 449600
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6759790582110743e-05,
+      "loss": 1.321,
+      "step": 449700
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6752395882631332e-05,
+      "loss": 1.3097,
+      "step": 449800
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6745001183151917e-05,
+      "loss": 1.3299,
+      "step": 449900
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6737606483672506e-05,
+      "loss": 1.3173,
+      "step": 450000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6730211784193092e-05,
+      "loss": 1.3186,
+      "step": 450100
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6722817084713678e-05,
+      "loss": 1.3287,
+      "step": 450200
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6715422385234263e-05,
+      "loss": 1.3325,
+      "step": 450300
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6708027685754852e-05,
+      "loss": 1.3165,
+      "step": 450400
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6700632986275438e-05,
+      "loss": 1.296,
+      "step": 450500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6693238286796027e-05,
+      "loss": 1.293,
+      "step": 450600
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6685843587316613e-05,
+      "loss": 1.305,
+      "step": 450700
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6678448887837202e-05,
+      "loss": 1.3075,
+      "step": 450800
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6671054188357784e-05,
+      "loss": 1.3307,
+      "step": 450900
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6663659488878373e-05,
+      "loss": 1.3202,
+      "step": 451000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.665626478939896e-05,
+      "loss": 1.3247,
+      "step": 451100
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6648870089919548e-05,
+      "loss": 1.3434,
+      "step": 451200
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6641475390440133e-05,
+      "loss": 1.3104,
+      "step": 451300
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6634080690960722e-05,
+      "loss": 1.3011,
+      "step": 451400
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6626685991481308e-05,
+      "loss": 1.2938,
+      "step": 451500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6619291292001894e-05,
+      "loss": 1.338,
+      "step": 451600
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.661189659252248e-05,
+      "loss": 1.3106,
+      "step": 451700
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6604575840037863e-05,
+      "loss": 1.3305,
+      "step": 451800
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.659718114055845e-05,
+      "loss": 1.3023,
+      "step": 451900
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6589786441079034e-05,
+      "loss": 1.3374,
+      "step": 452000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6582391741599623e-05,
+      "loss": 1.3123,
+      "step": 452100
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.657499704212021e-05,
+      "loss": 1.3249,
+      "step": 452200
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6567602342640795e-05,
+      "loss": 1.3257,
+      "step": 452300
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6560207643161384e-05,
+      "loss": 1.3005,
+      "step": 452400
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.655281294368197e-05,
+      "loss": 1.3334,
+      "step": 452500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.654541824420256e-05,
+      "loss": 1.3416,
+      "step": 452600
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6538023544723144e-05,
+      "loss": 1.3299,
+      "step": 452700
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6530628845243733e-05,
+      "loss": 1.3295,
+      "step": 452800
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6523234145764315e-05,
+      "loss": 1.317,
+      "step": 452900
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6515839446284904e-05,
+      "loss": 1.3268,
+      "step": 453000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.650844474680549e-05,
+      "loss": 1.2952,
+      "step": 453100
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.650105004732608e-05,
+      "loss": 1.3316,
+      "step": 453200
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6493655347846665e-05,
+      "loss": 1.3075,
+      "step": 453300
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6486260648367254e-05,
+      "loss": 1.3111,
+      "step": 453400
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6478865948887836e-05,
+      "loss": 1.3325,
+      "step": 453500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6471471249408425e-05,
+      "loss": 1.3321,
+      "step": 453600
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.646407654992901e-05,
+      "loss": 1.3242,
+      "step": 453700
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.645675579744439e-05,
+      "loss": 1.3271,
+      "step": 453800
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.644936109796498e-05,
+      "loss": 1.3341,
+      "step": 453900
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6441966398485566e-05,
+      "loss": 1.3171,
+      "step": 454000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6434571699006155e-05,
+      "loss": 1.2992,
+      "step": 454100
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.642717699952674e-05,
+      "loss": 1.3427,
+      "step": 454200
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6419782300047326e-05,
+      "loss": 1.311,
+      "step": 454300
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6412387600567915e-05,
+      "loss": 1.3405,
+      "step": 454400
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.64049929010885e-05,
+      "loss": 1.3322,
+      "step": 454500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6397598201609086e-05,
+      "loss": 1.2961,
+      "step": 454600
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6390203502129675e-05,
+      "loss": 1.3218,
+      "step": 454700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.638280880265026e-05,
+      "loss": 1.319,
+      "step": 454800
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6375414103170847e-05,
+      "loss": 1.2974,
+      "step": 454900
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6368019403691436e-05,
+      "loss": 1.317,
+      "step": 455000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.636062470421202e-05,
+      "loss": 1.3258,
+      "step": 455100
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.635323000473261e-05,
+      "loss": 1.3199,
+      "step": 455200
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6345835305253196e-05,
+      "loss": 1.3278,
+      "step": 455300
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6338440605773785e-05,
+      "loss": 1.3355,
+      "step": 455400
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6331045906294367e-05,
+      "loss": 1.3397,
+      "step": 455500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6323651206814956e-05,
+      "loss": 1.3179,
+      "step": 455600
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6316256507335542e-05,
+      "loss": 1.3412,
+      "step": 455700
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.630886180785613e-05,
+      "loss": 1.3348,
+      "step": 455800
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6301467108376716e-05,
+      "loss": 1.3337,
+      "step": 455900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6294146355892097e-05,
+      "loss": 1.3003,
+      "step": 456000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6286751656412686e-05,
+      "loss": 1.3339,
+      "step": 456100
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.627935695693327e-05,
+      "loss": 1.3042,
+      "step": 456200
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6271962257453857e-05,
+      "loss": 1.3123,
+      "step": 456300
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6264567557974443e-05,
+      "loss": 1.3199,
+      "step": 456400
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6257172858495032e-05,
+      "loss": 1.3189,
+      "step": 456500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6249778159015618e-05,
+      "loss": 1.3221,
+      "step": 456600
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.6242383459536207e-05,
+      "loss": 1.3291,
+      "step": 456700
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6234988760056792e-05,
+      "loss": 1.3121,
+      "step": 456800
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6227594060577378e-05,
+      "loss": 1.3266,
+      "step": 456900
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6220199361097967e-05,
+      "loss": 1.3144,
+      "step": 457000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6212804661618552e-05,
+      "loss": 1.3239,
+      "step": 457100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6205409962139138e-05,
+      "loss": 1.3232,
+      "step": 457200
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6198015262659727e-05,
+      "loss": 1.2982,
+      "step": 457300
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6190620563180313e-05,
+      "loss": 1.3373,
+      "step": 457400
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.61832258637009e-05,
+      "loss": 1.3407,
+      "step": 457500
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6175831164221487e-05,
+      "loss": 1.3172,
+      "step": 457600
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6168436464742073e-05,
+      "loss": 1.3153,
+      "step": 457700
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6161041765262662e-05,
+      "loss": 1.3215,
+      "step": 457800
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6153647065783248e-05,
+      "loss": 1.3311,
+      "step": 457900
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6146252366303837e-05,
+      "loss": 1.2825,
+      "step": 458000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6138857666824422e-05,
+      "loss": 1.3204,
+      "step": 458100
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.61315369143398e-05,
+      "loss": 1.32,
+      "step": 458200
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.612414221486039e-05,
+      "loss": 1.3211,
+      "step": 458300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6116747515380974e-05,
+      "loss": 1.3042,
+      "step": 458400
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6109426762896358e-05,
+      "loss": 1.3353,
+      "step": 458500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6102032063416944e-05,
+      "loss": 1.3321,
+      "step": 458600
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6094637363937533e-05,
+      "loss": 1.3069,
+      "step": 458700
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6087242664458115e-05,
+      "loss": 1.3155,
+      "step": 458800
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6079847964978704e-05,
+      "loss": 1.3116,
+      "step": 458900
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.607245326549929e-05,
+      "loss": 1.3339,
+      "step": 459000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.606505856601988e-05,
+      "loss": 1.3148,
+      "step": 459100
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6057663866540464e-05,
+      "loss": 1.3203,
+      "step": 459200
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6050269167061053e-05,
+      "loss": 1.2841,
+      "step": 459300
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.604287446758164e-05,
+      "loss": 1.3188,
+      "step": 459400
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6035479768102225e-05,
+      "loss": 1.3186,
+      "step": 459500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.602808506862281e-05,
+      "loss": 1.3187,
+      "step": 459600
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.60206903691434e-05,
+      "loss": 1.3222,
+      "step": 459700
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6013295669663985e-05,
+      "loss": 1.3505,
+      "step": 459800
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.6005900970184574e-05,
+      "loss": 1.3107,
+      "step": 459900
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.599850627070516e-05,
+      "loss": 1.3121,
+      "step": 460000
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5991111571225745e-05,
+      "loss": 1.3158,
+      "step": 460100
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.598371687174633e-05,
+      "loss": 1.3174,
+      "step": 460200
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.597632217226692e-05,
+      "loss": 1.331,
+      "step": 460300
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5968927472787505e-05,
+      "loss": 1.3001,
+      "step": 460400
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.5961532773308094e-05,
+      "loss": 1.3252,
+      "step": 460500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.595413807382868e-05,
+      "loss": 1.3169,
+      "step": 460600
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.594674337434927e-05,
+      "loss": 1.3255,
+      "step": 460700
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.593934867486985e-05,
+      "loss": 1.3048,
+      "step": 460800
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.593195397539044e-05,
+      "loss": 1.3218,
+      "step": 460900
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5924559275911026e-05,
+      "loss": 1.3091,
+      "step": 461000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5917164576431615e-05,
+      "loss": 1.3193,
+      "step": 461100
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.59097698769522e-05,
+      "loss": 1.3618,
+      "step": 461200
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.590237517747279e-05,
+      "loss": 1.3294,
+      "step": 461300
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5894980477993375e-05,
+      "loss": 1.3209,
+      "step": 461400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.588758577851396e-05,
+      "loss": 1.3081,
+      "step": 461500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5880191079034547e-05,
+      "loss": 1.3139,
+      "step": 461600
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5872796379555136e-05,
+      "loss": 1.3445,
+      "step": 461700
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.586540168007572e-05,
+      "loss": 1.3122,
+      "step": 461800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.585800698059631e-05,
+      "loss": 1.3144,
+      "step": 461900
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5850612281116896e-05,
+      "loss": 1.3199,
+      "step": 462000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.584321758163748e-05,
+      "loss": 1.3159,
+      "step": 462100
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.583582288215807e-05,
+      "loss": 1.3181,
+      "step": 462200
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5828428182678656e-05,
+      "loss": 1.3075,
+      "step": 462300
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5821033483199242e-05,
+      "loss": 1.3083,
+      "step": 462400
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.581363878371983e-05,
+      "loss": 1.3271,
+      "step": 462500
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5806244084240417e-05,
+      "loss": 1.3035,
+      "step": 462600
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5798849384761006e-05,
+      "loss": 1.337,
+      "step": 462700
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.5791528632276383e-05,
+      "loss": 1.3396,
+      "step": 462800
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.578413393279697e-05,
+      "loss": 1.3239,
+      "step": 462900
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5776739233317557e-05,
+      "loss": 1.3315,
+      "step": 463000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5769344533838146e-05,
+      "loss": 1.3083,
+      "step": 463100
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5761949834358732e-05,
+      "loss": 1.3186,
+      "step": 463200
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.575455513487932e-05,
+      "loss": 1.3096,
+      "step": 463300
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5747160435399907e-05,
+      "loss": 1.316,
+      "step": 463400
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5739765735920492e-05,
+      "loss": 1.3168,
+      "step": 463500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5732371036441078e-05,
+      "loss": 1.3204,
+      "step": 463600
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5724976336961667e-05,
+      "loss": 1.3035,
+      "step": 463700
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5717581637482253e-05,
+      "loss": 1.3266,
+      "step": 463800
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.571018693800284e-05,
+      "loss": 1.3291,
+      "step": 463900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5702792238523427e-05,
+      "loss": 1.3129,
+      "step": 464000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5695397539044013e-05,
+      "loss": 1.3114,
+      "step": 464100
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.56880028395646e-05,
+      "loss": 1.3404,
+      "step": 464200
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5680608140085188e-05,
+      "loss": 1.3237,
+      "step": 464300
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5673213440605773e-05,
+      "loss": 1.3351,
+      "step": 464400
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5665818741126362e-05,
+      "loss": 1.3233,
+      "step": 464500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5658424041646948e-05,
+      "loss": 1.3406,
+      "step": 464600
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5651029342167537e-05,
+      "loss": 1.3241,
+      "step": 464700
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5643708589682914e-05,
+      "loss": 1.2915,
+      "step": 464800
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5636313890203503e-05,
+      "loss": 1.3257,
+      "step": 464900
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.562891919072409e-05,
+      "loss": 1.3468,
+      "step": 465000
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5621524491244678e-05,
+      "loss": 1.3514,
+      "step": 465100
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5614129791765263e-05,
+      "loss": 1.3454,
+      "step": 465200
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5606735092285852e-05,
+      "loss": 1.3152,
+      "step": 465300
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5599340392806435e-05,
+      "loss": 1.3122,
+      "step": 465400
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.5591945693327024e-05,
+      "loss": 1.3161,
+      "step": 465500
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.558455099384761e-05,
+      "loss": 1.3005,
+      "step": 465600
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5577156294368198e-05,
+      "loss": 1.3242,
+      "step": 465700
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5569761594888784e-05,
+      "loss": 1.3209,
+      "step": 465800
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5562366895409373e-05,
+      "loss": 1.3109,
+      "step": 465900
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.555497219592996e-05,
+      "loss": 1.3375,
+      "step": 466000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5547577496450544e-05,
+      "loss": 1.3073,
+      "step": 466100
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.554018279697113e-05,
+      "loss": 1.2985,
+      "step": 466200
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.553278809749172e-05,
+      "loss": 1.3262,
+      "step": 466300
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5525393398012304e-05,
+      "loss": 1.3234,
+      "step": 466400
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5517998698532893e-05,
+      "loss": 1.3505,
+      "step": 466500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.551060399905348e-05,
+      "loss": 1.3172,
+      "step": 466600
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5503209299574068e-05,
+      "loss": 1.3248,
+      "step": 466700
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.549581460009465e-05,
+      "loss": 1.3305,
+      "step": 466800
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.548841990061524e-05,
+      "loss": 1.3285,
+      "step": 466900
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.548109914813062e-05,
+      "loss": 1.293,
+      "step": 467000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.547370444865121e-05,
+      "loss": 1.324,
+      "step": 467100
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5466309749171795e-05,
+      "loss": 1.3245,
+      "step": 467200
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5458915049692384e-05,
+      "loss": 1.3246,
+      "step": 467300
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5451520350212966e-05,
+      "loss": 1.3213,
+      "step": 467400
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.5444125650733555e-05,
+      "loss": 1.3233,
+      "step": 467500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.543673095125414e-05,
+      "loss": 1.3197,
+      "step": 467600
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.542933625177473e-05,
+      "loss": 1.3403,
+      "step": 467700
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5421941552295315e-05,
+      "loss": 1.3305,
+      "step": 467800
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5414546852815904e-05,
+      "loss": 1.3211,
+      "step": 467900
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.540715215333649e-05,
+      "loss": 1.3138,
+      "step": 468000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5399757453857075e-05,
+      "loss": 1.3351,
+      "step": 468100
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.539236275437766e-05,
+      "loss": 1.3292,
+      "step": 468200
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.538496805489825e-05,
+      "loss": 1.3093,
+      "step": 468300
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5377573355418836e-05,
+      "loss": 1.3408,
+      "step": 468400
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5370178655939425e-05,
+      "loss": 1.3489,
+      "step": 468500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.536278395646001e-05,
+      "loss": 1.3413,
+      "step": 468600
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5355389256980596e-05,
+      "loss": 1.3191,
+      "step": 468700
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.534799455750118e-05,
+      "loss": 1.3306,
+      "step": 468800
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.534059985802177e-05,
+      "loss": 1.3207,
+      "step": 468900
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5333205158542356e-05,
+      "loss": 1.3008,
+      "step": 469000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5325810459062945e-05,
+      "loss": 1.3296,
+      "step": 469100
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5318489706578326e-05,
+      "loss": 1.321,
+      "step": 469200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5311095007098915e-05,
+      "loss": 1.3535,
+      "step": 469300
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5303700307619497e-05,
+      "loss": 1.3262,
+      "step": 469400
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5296305608140086e-05,
+      "loss": 1.3295,
+      "step": 469500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5288910908660672e-05,
+      "loss": 1.3303,
+      "step": 469600
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.528151620918126e-05,
+      "loss": 1.3194,
+      "step": 469700
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5274121509701846e-05,
+      "loss": 1.3047,
+      "step": 469800
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5266726810222435e-05,
+      "loss": 1.3347,
+      "step": 469900
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.525933211074302e-05,
+      "loss": 1.3244,
+      "step": 470000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5251937411263607e-05,
+      "loss": 1.3141,
+      "step": 470100
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5244542711784194e-05,
+      "loss": 1.3284,
+      "step": 470200
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.523714801230478e-05,
+      "loss": 1.3122,
+      "step": 470300
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5229753312825367e-05,
+      "loss": 1.3263,
+      "step": 470400
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5222358613345954e-05,
+      "loss": 1.3488,
+      "step": 470500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5214963913866542e-05,
+      "loss": 1.3126,
+      "step": 470600
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5207569214387127e-05,
+      "loss": 1.3342,
+      "step": 470700
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5200174514907715e-05,
+      "loss": 1.3197,
+      "step": 470800
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5192779815428302e-05,
+      "loss": 1.3212,
+      "step": 470900
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.518538511594889e-05,
+      "loss": 1.3356,
+      "step": 471000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5177990416469477e-05,
+      "loss": 1.3145,
+      "step": 471100
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5170669663984857e-05,
+      "loss": 1.3199,
+      "step": 471200
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5163274964505444e-05,
+      "loss": 1.3131,
+      "step": 471300
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5155880265026028e-05,
+      "loss": 1.3222,
+      "step": 471400
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5148485565546616e-05,
+      "loss": 1.3396,
+      "step": 471500
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5141090866067203e-05,
+      "loss": 1.3201,
+      "step": 471600
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.513369616658779e-05,
+      "loss": 1.3197,
+      "step": 471700
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5126301467108378e-05,
+      "loss": 1.3451,
+      "step": 471800
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5118906767628965e-05,
+      "loss": 1.3368,
+      "step": 471900
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5111512068149549e-05,
+      "loss": 1.3204,
+      "step": 472000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5104117368670136e-05,
+      "loss": 1.3387,
+      "step": 472100
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5096722669190724e-05,
+      "loss": 1.3113,
+      "step": 472200
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.5089327969711311e-05,
+      "loss": 1.3052,
+      "step": 472300
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5081933270231898e-05,
+      "loss": 1.3321,
+      "step": 472400
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5074538570752486e-05,
+      "loss": 1.314,
+      "step": 472500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5067143871273073e-05,
+      "loss": 1.3318,
+      "step": 472600
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5059749171793659e-05,
+      "loss": 1.3022,
+      "step": 472700
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5052354472314246e-05,
+      "loss": 1.3446,
+      "step": 472800
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5044959772834832e-05,
+      "loss": 1.3182,
+      "step": 472900
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5037565073355419e-05,
+      "loss": 1.3318,
+      "step": 473000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5030170373876006e-05,
+      "loss": 1.3427,
+      "step": 473100
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5022849621391388e-05,
+      "loss": 1.2966,
+      "step": 473200
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5015454921911976e-05,
+      "loss": 1.3331,
+      "step": 473300
+    }
+  ],
+  "max_steps": 676160,
+  "num_train_epochs": 10,
+  "total_flos": 2.891701043509248e+18,
+  "trial_name": null,
+  "trial_params": null
+}