diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,6 +1,6 @@
 {
-  "best_metric": 2.5151257514953613,
-  "best_model_checkpoint": "./model_tweets_2020_Q1_50/checkpoint-128000",
+  "best_metric": 2.4417428970336914,
+  "best_model_checkpoint": "./model_tweets_2020_Q1_50/checkpoint-1888000",
   "epoch": 9.834292176820574,
   "eval_steps": 8000,
   "global_step": 2400000,
@@ -10,3312 +10,3312 @@
   "log_history": [
     {
       "epoch": 0.03,
-      "eval_loss": 2.6128780841827393,
-      "eval_runtime": 220.0915,
-      "eval_samples_per_second": 933.748,
-      "eval_steps_per_second": 58.362,
+      "eval_loss": 2.89373779296875,
+      "eval_runtime": 220.7068,
+      "eval_samples_per_second": 931.145,
+      "eval_steps_per_second": 58.199,
       "step": 8000
     },
     {
       "epoch": 0.07,
-      "learning_rate": 9.939131159843243e-06,
-      "loss": 2.7516,
+      "learning_rate": 4.0726666666666665e-07,
+      "loss": 3.073,
       "step": 16000
     },
     {
       "epoch": 0.07,
-      "eval_loss": 2.560152769088745,
-      "eval_runtime": 222.3718,
-      "eval_samples_per_second": 924.173,
-      "eval_steps_per_second": 57.764,
+      "eval_loss": 2.76598858833313,
+      "eval_runtime": 221.0774,
+      "eval_samples_per_second": 929.584,
+      "eval_steps_per_second": 58.102,
       "step": 16000
     },
     {
       "epoch": 0.1,
-      "eval_loss": 2.561044216156006,
-      "eval_runtime": 219.7796,
-      "eval_samples_per_second": 935.073,
-      "eval_steps_per_second": 58.445,
+      "eval_loss": 2.7232513427734375,
+      "eval_runtime": 221.6288,
+      "eval_samples_per_second": 927.271,
+      "eval_steps_per_second": 57.957,
       "step": 24000
     },
     {
       "epoch": 0.13,
-      "learning_rate": 9.872425581589261e-06,
-      "loss": 2.6478,
+      "learning_rate": 4.0453333333333336e-07,
+      "loss": 2.8244,
       "step": 32000
     },
     {
       "epoch": 0.13,
-      "eval_loss": 2.558018684387207,
-      "eval_runtime": 219.7961,
-      "eval_samples_per_second": 935.003,
-      "eval_steps_per_second": 58.441,
+      "eval_loss": 2.687758207321167,
+      "eval_runtime": 220.1262,
+      "eval_samples_per_second": 933.601,
+      "eval_steps_per_second": 58.353,
       "step": 32000
     },
     {
       "epoch": 0.16,
-      "eval_loss": 2.5498745441436768,
-      "eval_runtime": 221.2066,
-      "eval_samples_per_second": 929.041,
-      "eval_steps_per_second": 58.068,
+      "eval_loss": 2.6519503593444824,
+      "eval_runtime": 220.495,
+      "eval_samples_per_second": 932.039,
+      "eval_steps_per_second": 58.255,
       "step": 40000
     },
     {
       "epoch": 0.2,
-      "learning_rate": 9.80572000333528e-06,
-      "loss": 2.6344,
+      "learning_rate": 4.018e-07,
+      "loss": 2.7542,
       "step": 48000
     },
     {
       "epoch": 0.2,
-      "eval_loss": 2.542255163192749,
-      "eval_runtime": 219.3858,
-      "eval_samples_per_second": 936.752,
-      "eval_steps_per_second": 58.55,
+      "eval_loss": 2.63004469871521,
+      "eval_runtime": 220.2803,
+      "eval_samples_per_second": 932.948,
+      "eval_steps_per_second": 58.312,
       "step": 48000
     },
     {
       "epoch": 0.23,
-      "eval_loss": 2.5431501865386963,
-      "eval_runtime": 220.781,
-      "eval_samples_per_second": 930.832,
-      "eval_steps_per_second": 58.18,
+      "eval_loss": 2.613522529602051,
+      "eval_runtime": 221.2317,
+      "eval_samples_per_second": 928.936,
+      "eval_steps_per_second": 58.061,
       "step": 56000
     },
     {
       "epoch": 0.26,
-      "learning_rate": 9.739014425081299e-06,
-      "loss": 2.6174,
+      "learning_rate": 3.9906666666666667e-07,
+      "loss": 2.7083,
       "step": 64000
     },
     {
       "epoch": 0.26,
-      "eval_loss": 2.550448417663574,
-      "eval_runtime": 220.5,
-      "eval_samples_per_second": 932.018,
-      "eval_steps_per_second": 58.254,
+      "eval_loss": 2.6067709922790527,
+      "eval_runtime": 220.3177,
+      "eval_samples_per_second": 932.789,
+      "eval_steps_per_second": 58.302,
       "step": 64000
     },
     {
       "epoch": 0.3,
-      "eval_loss": 2.5348613262176514,
-      "eval_runtime": 220.0149,
-      "eval_samples_per_second": 934.073,
-      "eval_steps_per_second": 58.382,
+      "eval_loss": 2.5854294300079346,
+      "eval_runtime": 220.8061,
+      "eval_samples_per_second": 930.726,
+      "eval_steps_per_second": 58.173,
       "step": 72000
     },
     {
       "epoch": 0.33,
-      "learning_rate": 9.672308846827316e-06,
-      "loss": 2.5993,
+      "learning_rate": 3.963333333333333e-07,
+      "loss": 2.6752,
       "step": 80000
     },
     {
       "epoch": 0.33,
-      "eval_loss": 2.533210039138794,
-      "eval_runtime": 224.2726,
-      "eval_samples_per_second": 916.34,
-      "eval_steps_per_second": 57.274,
+      "eval_loss": 2.575528860092163,
+      "eval_runtime": 221.8521,
+      "eval_samples_per_second": 926.338,
+      "eval_steps_per_second": 57.899,
       "step": 80000
     },
     {
       "epoch": 0.36,
-      "eval_loss": 2.5342018604278564,
-      "eval_runtime": 221.6101,
-      "eval_samples_per_second": 927.349,
-      "eval_steps_per_second": 57.962,
+      "eval_loss": 2.5720720291137695,
+      "eval_runtime": 221.4472,
+      "eval_samples_per_second": 928.032,
+      "eval_steps_per_second": 58.005,
       "step": 88000
     },
     {
       "epoch": 0.39,
-      "learning_rate": 9.605603268573334e-06,
-      "loss": 2.5948,
+      "learning_rate": 3.936e-07,
+      "loss": 2.6657,
       "step": 96000
     },
     {
       "epoch": 0.39,
-      "eval_loss": 2.5296876430511475,
-      "eval_runtime": 221.2575,
-      "eval_samples_per_second": 928.827,
-      "eval_steps_per_second": 58.055,
+      "eval_loss": 2.5709290504455566,
+      "eval_runtime": 220.9006,
+      "eval_samples_per_second": 930.328,
+      "eval_steps_per_second": 58.148,
       "step": 96000
     },
     {
       "epoch": 0.43,
-      "eval_loss": 2.5301403999328613,
-      "eval_runtime": 221.9648,
-      "eval_samples_per_second": 925.868,
-      "eval_steps_per_second": 57.87,
+      "eval_loss": 2.5656096935272217,
+      "eval_runtime": 220.5433,
+      "eval_samples_per_second": 931.835,
+      "eval_steps_per_second": 58.243,
       "step": 104000
     },
     {
       "epoch": 0.46,
-      "learning_rate": 9.538897690319354e-06,
-      "loss": 2.5846,
+      "learning_rate": 3.908666666666667e-07,
+      "loss": 2.6534,
       "step": 112000
     },
     {
       "epoch": 0.46,
-      "eval_loss": 2.5199761390686035,
-      "eval_runtime": 221.7756,
-      "eval_samples_per_second": 926.658,
-      "eval_steps_per_second": 57.919,
+      "eval_loss": 2.5558407306671143,
+      "eval_runtime": 221.6371,
+      "eval_samples_per_second": 927.236,
+      "eval_steps_per_second": 57.955,
       "step": 112000
     },
     {
       "epoch": 0.49,
-      "eval_loss": 2.5174739360809326,
-      "eval_runtime": 222.132,
-      "eval_samples_per_second": 925.171,
-      "eval_steps_per_second": 57.826,
+      "eval_loss": 2.5495829582214355,
+      "eval_runtime": 220.7733,
+      "eval_samples_per_second": 930.864,
+      "eval_steps_per_second": 58.182,
       "step": 120000
     },
     {
       "epoch": 0.52,
-      "learning_rate": 9.472192112065373e-06,
-      "loss": 2.5774,
+      "learning_rate": 3.8813333333333334e-07,
+      "loss": 2.646,
       "step": 128000
     },
     {
       "epoch": 0.52,
-      "eval_loss": 2.5151257514953613,
-      "eval_runtime": 222.5971,
-      "eval_samples_per_second": 923.238,
-      "eval_steps_per_second": 57.705,
+      "eval_loss": 2.547106981277466,
+      "eval_runtime": 221.6448,
+      "eval_samples_per_second": 927.204,
+      "eval_steps_per_second": 57.953,
       "step": 128000
     },
     {
       "epoch": 0.56,
-      "eval_loss": 2.512612819671631,
-      "eval_runtime": 225.9428,
-      "eval_samples_per_second": 909.567,
-      "eval_steps_per_second": 56.851,
+      "eval_loss": 2.5408244132995605,
+      "eval_runtime": 221.6302,
+      "eval_samples_per_second": 927.265,
+      "eval_steps_per_second": 57.957,
       "step": 136000
     },
     {
       "epoch": 0.59,
-      "learning_rate": 9.405486533811392e-06,
-      "loss": 2.5584,
+      "learning_rate": 3.854e-07,
+      "loss": 2.625,
       "step": 144000
     },
     {
       "epoch": 0.59,
-      "eval_loss": 2.5006816387176514,
-      "eval_runtime": 225.1567,
-      "eval_samples_per_second": 912.742,
-      "eval_steps_per_second": 57.049,
+      "eval_loss": 2.531517744064331,
+      "eval_runtime": 223.6683,
+      "eval_samples_per_second": 918.816,
+      "eval_steps_per_second": 57.429,
       "step": 144000
     },
     {
       "epoch": 0.62,
-      "eval_loss": 2.5149295330047607,
-      "eval_runtime": 222.6305,
-      "eval_samples_per_second": 923.099,
-      "eval_steps_per_second": 57.697,
+      "eval_loss": 2.5364675521850586,
+      "eval_runtime": 224.1465,
+      "eval_samples_per_second": 916.856,
+      "eval_steps_per_second": 57.306,
       "step": 152000
     },
     {
       "epoch": 0.66,
-      "learning_rate": 9.338780955557409e-06,
-      "loss": 2.5578,
+      "learning_rate": 3.8266666666666665e-07,
+      "loss": 2.6222,
       "step": 160000
     },
     {
       "epoch": 0.66,
-      "eval_loss": 2.519784688949585,
-      "eval_runtime": 224.6765,
-      "eval_samples_per_second": 914.693,
-      "eval_steps_per_second": 57.171,
+      "eval_loss": 2.5372273921966553,
+      "eval_runtime": 221.7325,
+      "eval_samples_per_second": 926.837,
+      "eval_steps_per_second": 57.93,
       "step": 160000
     },
     {
       "epoch": 0.69,
-      "eval_loss": 2.5136947631835938,
-      "eval_runtime": 223.084,
-      "eval_samples_per_second": 921.223,
-      "eval_steps_per_second": 57.579,
+      "eval_loss": 2.534186363220215,
+      "eval_runtime": 222.3525,
+      "eval_samples_per_second": 924.253,
+      "eval_steps_per_second": 57.769,
       "step": 168000
     },
     {
       "epoch": 0.72,
-      "learning_rate": 9.272075377303427e-06,
-      "loss": 2.5699,
+      "learning_rate": 3.799333333333333e-07,
+      "loss": 2.6256,
       "step": 176000
     },
     {
       "epoch": 0.72,
-      "eval_loss": 2.5189852714538574,
-      "eval_runtime": 221.4167,
-      "eval_samples_per_second": 928.159,
-      "eval_steps_per_second": 58.013,
+      "eval_loss": 2.5308265686035156,
+      "eval_runtime": 221.538,
+      "eval_samples_per_second": 927.651,
+      "eval_steps_per_second": 57.981,
       "step": 176000
     },
     {
       "epoch": 0.75,
-      "eval_loss": 2.5280351638793945,
-      "eval_runtime": 220.8999,
-      "eval_samples_per_second": 930.331,
-      "eval_steps_per_second": 58.149,
+      "eval_loss": 2.5311617851257324,
+      "eval_runtime": 224.2919,
+      "eval_samples_per_second": 916.261,
+      "eval_steps_per_second": 57.269,
       "step": 184000
     },
     {
       "epoch": 0.79,
-      "learning_rate": 9.205369799049446e-06,
-      "loss": 2.5648,
+      "learning_rate": 3.772e-07,
+      "loss": 2.6074,
       "step": 192000
     },
     {
       "epoch": 0.79,
-      "eval_loss": 2.5205190181732178,
-      "eval_runtime": 220.8432,
-      "eval_samples_per_second": 930.57,
-      "eval_steps_per_second": 58.163,
+      "eval_loss": 2.522848129272461,
+      "eval_runtime": 224.4507,
+      "eval_samples_per_second": 915.613,
+      "eval_steps_per_second": 57.229,
       "step": 192000
     },
     {
       "epoch": 0.82,
-      "eval_loss": 2.5266730785369873,
-      "eval_runtime": 223.4826,
-      "eval_samples_per_second": 919.579,
-      "eval_steps_per_second": 57.477,
+      "eval_loss": 2.529161214828491,
+      "eval_runtime": 222.7477,
+      "eval_samples_per_second": 922.613,
+      "eval_steps_per_second": 57.666,
       "step": 200000
     },
     {
       "epoch": 0.85,
-      "learning_rate": 9.138664220795464e-06,
-      "loss": 2.5701,
+      "learning_rate": 3.7446666666666667e-07,
+      "loss": 2.6071,
       "step": 208000
     },
     {
       "epoch": 0.85,
-      "eval_loss": 2.533360719680786,
-      "eval_runtime": 222.9517,
-      "eval_samples_per_second": 921.769,
-      "eval_steps_per_second": 57.613,
+      "eval_loss": 2.5295047760009766,
+      "eval_runtime": 223.0891,
+      "eval_samples_per_second": 921.201,
+      "eval_steps_per_second": 57.578,
       "step": 208000
     },
     {
       "epoch": 0.89,
-      "eval_loss": 2.5359556674957275,
-      "eval_runtime": 223.0368,
-      "eval_samples_per_second": 921.418,
-      "eval_steps_per_second": 57.591,
+      "eval_loss": 2.523491621017456,
+      "eval_runtime": 221.4007,
+      "eval_samples_per_second": 928.227,
+      "eval_steps_per_second": 58.017,
       "step": 216000
     },
     {
       "epoch": 0.92,
-      "learning_rate": 9.071958642541483e-06,
-      "loss": 2.5688,
+      "learning_rate": 3.7173333333333333e-07,
+      "loss": 2.5955,
       "step": 224000
     },
     {
       "epoch": 0.92,
-      "eval_loss": 2.5403811931610107,
-      "eval_runtime": 219.8534,
-      "eval_samples_per_second": 934.759,
-      "eval_steps_per_second": 58.425,
+      "eval_loss": 2.5219199657440186,
+      "eval_runtime": 221.3605,
+      "eval_samples_per_second": 928.395,
+      "eval_steps_per_second": 58.028,
       "step": 224000
     },
     {
       "epoch": 0.95,
-      "eval_loss": 2.536496162414551,
-      "eval_runtime": 222.4622,
-      "eval_samples_per_second": 923.797,
-      "eval_steps_per_second": 57.74,
+      "eval_loss": 2.5190882682800293,
+      "eval_runtime": 221.1449,
+      "eval_samples_per_second": 929.3,
+      "eval_steps_per_second": 58.084,
       "step": 232000
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.005253064287502e-06,
-      "loss": 2.5788,
+      "learning_rate": 3.69e-07,
+      "loss": 2.6036,
       "step": 240000
     },
     {
       "epoch": 0.98,
-      "eval_loss": 2.5484516620635986,
-      "eval_runtime": 221.8944,
-      "eval_samples_per_second": 926.161,
-      "eval_steps_per_second": 57.888,
+      "eval_loss": 2.517120361328125,
+      "eval_runtime": 220.9198,
+      "eval_samples_per_second": 930.247,
+      "eval_steps_per_second": 58.143,
       "step": 240000
     },
     {
       "epoch": 1.02,
-      "eval_loss": 2.5406737327575684,
-      "eval_runtime": 223.364,
-      "eval_samples_per_second": 920.068,
-      "eval_steps_per_second": 57.507,
+      "eval_loss": 2.5102434158325195,
+      "eval_runtime": 221.7647,
+      "eval_samples_per_second": 926.703,
+      "eval_steps_per_second": 57.922,
       "step": 248000
     },
     {
       "epoch": 1.05,
-      "learning_rate": 8.93854748603352e-06,
-      "loss": 2.5857,
+      "learning_rate": 3.6626666666666664e-07,
+      "loss": 2.6046,
       "step": 256000
     },
     {
       "epoch": 1.05,
-      "eval_loss": 2.5444178581237793,
-      "eval_runtime": 222.6684,
-      "eval_samples_per_second": 922.942,
-      "eval_steps_per_second": 57.687,
+      "eval_loss": 2.5070137977600098,
+      "eval_runtime": 221.6584,
+      "eval_samples_per_second": 927.147,
+      "eval_steps_per_second": 57.95,
       "step": 256000
     },
     {
       "epoch": 1.08,
-      "eval_loss": 2.55073881149292,
-      "eval_runtime": 221.5347,
-      "eval_samples_per_second": 927.665,
-      "eval_steps_per_second": 57.982,
+      "eval_loss": 2.5109376907348633,
+      "eval_runtime": 221.3382,
+      "eval_samples_per_second": 928.489,
+      "eval_steps_per_second": 58.033,
       "step": 264000
     },
     {
       "epoch": 1.11,
-      "learning_rate": 8.871841907779539e-06,
-      "loss": 2.576,
+      "learning_rate": 3.6353333333333335e-07,
+      "loss": 2.5892,
       "step": 272000
     },
     {
       "epoch": 1.11,
-      "eval_loss": 2.5567193031311035,
-      "eval_runtime": 221.584,
-      "eval_samples_per_second": 927.459,
-      "eval_steps_per_second": 57.969,
+      "eval_loss": 2.5104565620422363,
+      "eval_runtime": 222.1683,
+      "eval_samples_per_second": 925.019,
+      "eval_steps_per_second": 57.817,
       "step": 272000
     },
     {
       "epoch": 1.15,
-      "eval_loss": 2.5561089515686035,
-      "eval_runtime": 221.035,
-      "eval_samples_per_second": 929.762,
-      "eval_steps_per_second": 58.113,
+      "eval_loss": 2.508704423904419,
+      "eval_runtime": 222.9629,
+      "eval_samples_per_second": 921.723,
+      "eval_steps_per_second": 57.61,
       "step": 280000
     },
     {
       "epoch": 1.18,
-      "learning_rate": 8.805136329525557e-06,
-      "loss": 2.5919,
+      "learning_rate": 3.608e-07,
+      "loss": 2.5929,
       "step": 288000
     },
     {
       "epoch": 1.18,
-      "eval_loss": 2.564077138900757,
-      "eval_runtime": 221.6286,
-      "eval_samples_per_second": 927.272,
-      "eval_steps_per_second": 57.957,
+      "eval_loss": 2.509392738342285,
+      "eval_runtime": 223.3494,
+      "eval_samples_per_second": 920.128,
+      "eval_steps_per_second": 57.511,
       "step": 288000
     },
     {
       "epoch": 1.21,
-      "eval_loss": 2.5574235916137695,
-      "eval_runtime": 221.5082,
-      "eval_samples_per_second": 927.776,
-      "eval_steps_per_second": 57.989,
+      "eval_loss": 2.508585214614868,
+      "eval_runtime": 222.7314,
+      "eval_samples_per_second": 922.681,
+      "eval_steps_per_second": 57.67,
       "step": 296000
     },
     {
       "epoch": 1.25,
-      "learning_rate": 8.738430751271576e-06,
-      "loss": 2.5893,
+      "learning_rate": 3.5806666666666666e-07,
+      "loss": 2.5857,
       "step": 304000
     },
     {
       "epoch": 1.25,
-      "eval_loss": 2.5562856197357178,
-      "eval_runtime": 221.2963,
-      "eval_samples_per_second": 928.664,
-      "eval_steps_per_second": 58.044,
+      "eval_loss": 2.4991345405578613,
+      "eval_runtime": 223.3332,
+      "eval_samples_per_second": 920.195,
+      "eval_steps_per_second": 57.515,
       "step": 304000
     },
     {
       "epoch": 1.28,
-      "eval_loss": 2.5707902908325195,
-      "eval_runtime": 220.7861,
-      "eval_samples_per_second": 930.811,
-      "eval_steps_per_second": 58.178,
+      "eval_loss": 2.508927822113037,
+      "eval_runtime": 224.1404,
+      "eval_samples_per_second": 916.881,
+      "eval_steps_per_second": 57.308,
       "step": 312000
     },
     {
       "epoch": 1.31,
-      "learning_rate": 8.671725173017595e-06,
-      "loss": 2.5896,
+      "learning_rate": 3.553333333333333e-07,
+      "loss": 2.5828,
       "step": 320000
     },
     {
       "epoch": 1.31,
-      "eval_loss": 2.571282386779785,
-      "eval_runtime": 222.5864,
-      "eval_samples_per_second": 923.282,
-      "eval_steps_per_second": 57.708,
+      "eval_loss": 2.501734972000122,
+      "eval_runtime": 223.1146,
+      "eval_samples_per_second": 921.096,
+      "eval_steps_per_second": 57.571,
       "step": 320000
     },
     {
       "epoch": 1.34,
-      "eval_loss": 2.5756185054779053,
-      "eval_runtime": 220.8371,
-      "eval_samples_per_second": 930.596,
-      "eval_steps_per_second": 58.165,
+      "eval_loss": 2.503918409347534,
+      "eval_runtime": 223.3327,
+      "eval_samples_per_second": 920.196,
+      "eval_steps_per_second": 57.515,
       "step": 328000
     },
     {
       "epoch": 1.38,
-      "learning_rate": 8.605019594763613e-06,
-      "loss": 2.6066,
+      "learning_rate": 3.5259999999999997e-07,
+      "loss": 2.5812,
       "step": 336000
     },
     {
       "epoch": 1.38,
-      "eval_loss": 2.5867831707000732,
-      "eval_runtime": 221.6633,
-      "eval_samples_per_second": 927.127,
-      "eval_steps_per_second": 57.948,
+      "eval_loss": 2.5064587593078613,
+      "eval_runtime": 224.1587,
+      "eval_samples_per_second": 916.806,
+      "eval_steps_per_second": 57.303,
       "step": 336000
     },
     {
       "epoch": 1.41,
-      "eval_loss": 2.5921335220336914,
-      "eval_runtime": 220.8265,
-      "eval_samples_per_second": 930.64,
-      "eval_steps_per_second": 58.168,
+      "eval_loss": 2.508263111114502,
+      "eval_runtime": 222.503,
+      "eval_samples_per_second": 923.628,
+      "eval_steps_per_second": 57.73,
       "step": 344000
     },
     {
       "epoch": 1.44,
-      "learning_rate": 8.538314016509632e-06,
-      "loss": 2.6121,
+      "learning_rate": 3.498666666666667e-07,
+      "loss": 2.5775,
       "step": 352000
     },
     {
       "epoch": 1.44,
-      "eval_loss": 2.605332136154175,
-      "eval_runtime": 224.0349,
-      "eval_samples_per_second": 917.312,
-      "eval_steps_per_second": 57.335,
+      "eval_loss": 2.509936571121216,
+      "eval_runtime": 223.039,
+      "eval_samples_per_second": 921.408,
+      "eval_steps_per_second": 57.591,
       "step": 352000
     },
     {
       "epoch": 1.48,
-      "eval_loss": 2.60457706451416,
-      "eval_runtime": 220.8455,
-      "eval_samples_per_second": 930.56,
-      "eval_steps_per_second": 58.163,
+      "eval_loss": 2.5078811645507812,
+      "eval_runtime": 221.7646,
+      "eval_samples_per_second": 926.703,
+      "eval_steps_per_second": 57.922,
       "step": 360000
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.471608438255649e-06,
-      "loss": 2.6161,
+      "learning_rate": 3.4713333333333333e-07,
+      "loss": 2.5711,
       "step": 368000
     },
     {
       "epoch": 1.51,
-      "eval_loss": 2.5993847846984863,
-      "eval_runtime": 222.5248,
-      "eval_samples_per_second": 923.537,
-      "eval_steps_per_second": 57.724,
+      "eval_loss": 2.4922046661376953,
+      "eval_runtime": 223.0544,
+      "eval_samples_per_second": 921.345,
+      "eval_steps_per_second": 57.587,
       "step": 368000
     },
     {
       "epoch": 1.54,
-      "eval_loss": 2.6035287380218506,
-      "eval_runtime": 221.0365,
-      "eval_samples_per_second": 929.756,
-      "eval_steps_per_second": 58.113,
+      "eval_loss": 2.5012030601501465,
+      "eval_runtime": 222.0392,
+      "eval_samples_per_second": 925.557,
+      "eval_steps_per_second": 57.85,
       "step": 376000
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.404902860001667e-06,
-      "loss": 2.6313,
+      "learning_rate": 3.444e-07,
+      "loss": 2.5797,
       "step": 384000
     },
     {
       "epoch": 1.57,
-      "eval_loss": 2.6119377613067627,
-      "eval_runtime": 221.8478,
-      "eval_samples_per_second": 926.356,
-      "eval_steps_per_second": 57.9,
+      "eval_loss": 2.49989914894104,
+      "eval_runtime": 223.8829,
+      "eval_samples_per_second": 917.935,
+      "eval_steps_per_second": 57.374,
       "step": 384000
     },
     {
       "epoch": 1.61,
-      "eval_loss": 2.601569175720215,
-      "eval_runtime": 221.1561,
-      "eval_samples_per_second": 929.253,
-      "eval_steps_per_second": 58.081,
+      "eval_loss": 2.4881107807159424,
+      "eval_runtime": 222.4413,
+      "eval_samples_per_second": 923.884,
+      "eval_steps_per_second": 57.746,
       "step": 392000
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.338197281747686e-06,
-      "loss": 2.6342,
+      "learning_rate": 3.416666666666667e-07,
+      "loss": 2.5718,
       "step": 400000
     },
     {
       "epoch": 1.64,
-      "eval_loss": 2.62074875831604,
-      "eval_runtime": 222.2544,
-      "eval_samples_per_second": 924.661,
-      "eval_steps_per_second": 57.794,
+      "eval_loss": 2.4960451126098633,
+      "eval_runtime": 222.8741,
+      "eval_samples_per_second": 922.09,
+      "eval_steps_per_second": 57.633,
       "step": 400000
     },
     {
       "epoch": 1.67,
-      "eval_loss": 2.623107433319092,
-      "eval_runtime": 221.8176,
-      "eval_samples_per_second": 926.482,
-      "eval_steps_per_second": 57.908,
+      "eval_loss": 2.490837574005127,
+      "eval_runtime": 222.3679,
+      "eval_samples_per_second": 924.189,
+      "eval_steps_per_second": 57.765,
       "step": 408000
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.271491703493705e-06,
-      "loss": 2.6358,
+      "learning_rate": 3.3893333333333335e-07,
+      "loss": 2.5627,
       "step": 416000
     },
     {
       "epoch": 1.7,
-      "eval_loss": 2.6287713050842285,
-      "eval_runtime": 224.4588,
-      "eval_samples_per_second": 915.58,
-      "eval_steps_per_second": 57.227,
+      "eval_loss": 2.4970648288726807,
+      "eval_runtime": 223.472,
+      "eval_samples_per_second": 919.623,
+      "eval_steps_per_second": 57.479,
       "step": 416000
     },
     {
       "epoch": 1.74,
-      "eval_loss": 2.633507490158081,
-      "eval_runtime": 222.6705,
-      "eval_samples_per_second": 922.933,
-      "eval_steps_per_second": 57.686,
+      "eval_loss": 2.4916465282440186,
+      "eval_runtime": 222.5109,
+      "eval_samples_per_second": 923.595,
+      "eval_steps_per_second": 57.728,
       "step": 424000
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.204786125239725e-06,
-      "loss": 2.6463,
+      "learning_rate": 3.3619999999999995e-07,
+      "loss": 2.5641,
       "step": 432000
     },
     {
       "epoch": 1.77,
-      "eval_loss": 2.63563871383667,
-      "eval_runtime": 221.9208,
-      "eval_samples_per_second": 926.051,
-      "eval_steps_per_second": 57.881,
+      "eval_loss": 2.4971389770507812,
+      "eval_runtime": 222.1533,
+      "eval_samples_per_second": 925.082,
+      "eval_steps_per_second": 57.82,
       "step": 432000
     },
     {
       "epoch": 1.8,
-      "eval_loss": 2.6429896354675293,
-      "eval_runtime": 222.0647,
-      "eval_samples_per_second": 925.451,
-      "eval_steps_per_second": 57.844,
+      "eval_loss": 2.495426654815674,
+      "eval_runtime": 223.2728,
+      "eval_samples_per_second": 920.444,
+      "eval_steps_per_second": 57.531,
       "step": 440000
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.138080546985743e-06,
-      "loss": 2.6561,
+      "learning_rate": 3.3346666666666666e-07,
+      "loss": 2.5633,
       "step": 448000
     },
     {
       "epoch": 1.84,
-      "eval_loss": 2.643951416015625,
-      "eval_runtime": 222.3078,
-      "eval_samples_per_second": 924.439,
-      "eval_steps_per_second": 57.78,
+      "eval_loss": 2.485994815826416,
+      "eval_runtime": 222.7264,
+      "eval_samples_per_second": 922.702,
+      "eval_steps_per_second": 57.672,
       "step": 448000
     },
     {
       "epoch": 1.87,
-      "eval_loss": 2.6444947719573975,
-      "eval_runtime": 222.0032,
-      "eval_samples_per_second": 925.707,
-      "eval_steps_per_second": 57.86,
+      "eval_loss": 2.4893651008605957,
+      "eval_runtime": 223.4251,
+      "eval_samples_per_second": 919.816,
+      "eval_steps_per_second": 57.491,
       "step": 456000
     },
     {
       "epoch": 1.9,
-      "learning_rate": 8.07137496873176e-06,
-      "loss": 2.6748,
+      "learning_rate": 3.307333333333333e-07,
+      "loss": 2.5676,
       "step": 464000
     },
     {
       "epoch": 1.9,
-      "eval_loss": 2.6506636142730713,
-      "eval_runtime": 223.5942,
-      "eval_samples_per_second": 919.121,
-      "eval_steps_per_second": 57.448,
+      "eval_loss": 2.489337205886841,
+      "eval_runtime": 222.9423,
+      "eval_samples_per_second": 921.808,
+      "eval_steps_per_second": 57.616,
       "step": 464000
     },
     {
       "epoch": 1.93,
-      "eval_loss": 2.6547434329986572,
-      "eval_runtime": 221.7433,
-      "eval_samples_per_second": 926.792,
-      "eval_steps_per_second": 57.927,
+      "eval_loss": 2.4883553981781006,
+      "eval_runtime": 223.2404,
+      "eval_samples_per_second": 920.577,
+      "eval_steps_per_second": 57.539,
       "step": 472000
     },
     {
       "epoch": 1.97,
-      "learning_rate": 8.004669390477779e-06,
-      "loss": 2.6847,
+      "learning_rate": 3.28e-07,
+      "loss": 2.5687,
       "step": 480000
     },
     {
       "epoch": 1.97,
-      "eval_loss": 2.664285898208618,
-      "eval_runtime": 221.7152,
-      "eval_samples_per_second": 926.91,
-      "eval_steps_per_second": 57.935,
+      "eval_loss": 2.4921038150787354,
+      "eval_runtime": 223.8809,
+      "eval_samples_per_second": 917.943,
+      "eval_steps_per_second": 57.374,
       "step": 480000
     },
     {
       "epoch": 2.0,
-      "eval_loss": 2.670330762863159,
-      "eval_runtime": 222.5951,
-      "eval_samples_per_second": 923.246,
-      "eval_steps_per_second": 57.706,
+      "eval_loss": 2.4873294830322266,
+      "eval_runtime": 222.8771,
+      "eval_samples_per_second": 922.078,
+      "eval_steps_per_second": 57.633,
       "step": 488000
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.937963812223798e-06,
-      "loss": 2.6772,
+      "learning_rate": 3.252666666666667e-07,
+      "loss": 2.5633,
       "step": 496000
     },
     {
       "epoch": 2.03,
-      "eval_loss": 2.674689292907715,
-      "eval_runtime": 222.9188,
-      "eval_samples_per_second": 921.905,
-      "eval_steps_per_second": 57.622,
+      "eval_loss": 2.4919497966766357,
+      "eval_runtime": 222.6439,
+      "eval_samples_per_second": 923.043,
+      "eval_steps_per_second": 57.693,
       "step": 496000
     },
     {
       "epoch": 2.07,
-      "eval_loss": 2.6655595302581787,
-      "eval_runtime": 221.9863,
-      "eval_samples_per_second": 925.778,
-      "eval_steps_per_second": 57.864,
+      "eval_loss": 2.482137441635132,
+      "eval_runtime": 222.747,
+      "eval_samples_per_second": 922.616,
+      "eval_steps_per_second": 57.666,
       "step": 504000
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.871258233969816e-06,
-      "loss": 2.6668,
+      "learning_rate": 3.2253333333333334e-07,
+      "loss": 2.5547,
       "step": 512000
     },
     {
       "epoch": 2.1,
-      "eval_loss": 2.676820993423462,
-      "eval_runtime": 222.0249,
-      "eval_samples_per_second": 925.617,
-      "eval_steps_per_second": 57.854,
+      "eval_loss": 2.490872621536255,
+      "eval_runtime": 222.6765,
+      "eval_samples_per_second": 922.908,
+      "eval_steps_per_second": 57.685,
       "step": 512000
     },
     {
       "epoch": 2.13,
-      "eval_loss": 2.669207811355591,
-      "eval_runtime": 221.7932,
-      "eval_samples_per_second": 926.584,
-      "eval_steps_per_second": 57.914,
+      "eval_loss": 2.4818356037139893,
+      "eval_runtime": 223.7166,
+      "eval_samples_per_second": 918.617,
+      "eval_steps_per_second": 57.416,
       "step": 520000
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.804552655715835e-06,
-      "loss": 2.6802,
+      "learning_rate": 3.198e-07,
+      "loss": 2.5617,
       "step": 528000
     },
     {
       "epoch": 2.16,
-      "eval_loss": 2.6729602813720703,
-      "eval_runtime": 221.933,
-      "eval_samples_per_second": 926.0,
-      "eval_steps_per_second": 57.878,
+      "eval_loss": 2.4854869842529297,
+      "eval_runtime": 223.7715,
+      "eval_samples_per_second": 918.392,
+      "eval_steps_per_second": 57.402,
       "step": 528000
     },
     {
       "epoch": 2.2,
-      "eval_loss": 2.67464017868042,
-      "eval_runtime": 222.2198,
-      "eval_samples_per_second": 924.805,
-      "eval_steps_per_second": 57.803,
+      "eval_loss": 2.48504638671875,
+      "eval_runtime": 223.6654,
+      "eval_samples_per_second": 918.828,
+      "eval_steps_per_second": 57.43,
       "step": 536000
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.737847077461853e-06,
-      "loss": 2.6856,
+      "learning_rate": 3.1706666666666665e-07,
+      "loss": 2.5569,
       "step": 544000
     },
     {
       "epoch": 2.23,
-      "eval_loss": 2.678727865219116,
-      "eval_runtime": 222.413,
-      "eval_samples_per_second": 924.002,
-      "eval_steps_per_second": 57.753,
+      "eval_loss": 2.480282783508301,
+      "eval_runtime": 222.7744,
+      "eval_samples_per_second": 922.503,
+      "eval_steps_per_second": 57.659,
       "step": 544000
     },
     {
       "epoch": 2.26,
-      "eval_loss": 2.6777920722961426,
-      "eval_runtime": 222.6488,
-      "eval_samples_per_second": 923.023,
-      "eval_steps_per_second": 57.692,
+      "eval_loss": 2.4775896072387695,
+      "eval_runtime": 223.0018,
+      "eval_samples_per_second": 921.562,
+      "eval_steps_per_second": 57.6,
       "step": 552000
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.671141499207872e-06,
-      "loss": 2.6874,
+      "learning_rate": 3.1433333333333336e-07,
+      "loss": 2.5535,
       "step": 560000
     },
     {
       "epoch": 2.29,
-      "eval_loss": 2.6909406185150146,
-      "eval_runtime": 221.9152,
-      "eval_samples_per_second": 926.074,
-      "eval_steps_per_second": 57.882,
+      "eval_loss": 2.4824471473693848,
+      "eval_runtime": 223.1733,
+      "eval_samples_per_second": 920.854,
+      "eval_steps_per_second": 57.556,
       "step": 560000
     },
     {
       "epoch": 2.33,
-      "eval_loss": 2.691913604736328,
-      "eval_runtime": 222.3607,
-      "eval_samples_per_second": 924.219,
-      "eval_steps_per_second": 57.766,
+      "eval_loss": 2.4821510314941406,
+      "eval_runtime": 224.1586,
+      "eval_samples_per_second": 916.806,
+      "eval_steps_per_second": 57.303,
       "step": 568000
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.604435920953891e-06,
-      "loss": 2.6956,
+      "learning_rate": 3.116e-07,
+      "loss": 2.5534,
       "step": 576000
     },
     {
       "epoch": 2.36,
-      "eval_loss": 2.69474720954895,
-      "eval_runtime": 222.1948,
-      "eval_samples_per_second": 924.909,
-      "eval_steps_per_second": 57.81,
+      "eval_loss": 2.476337432861328,
+      "eval_runtime": 223.4733,
+      "eval_samples_per_second": 919.618,
+      "eval_steps_per_second": 57.479,
       "step": 576000
     },
     {
       "epoch": 2.39,
-      "eval_loss": 2.7031822204589844,
-      "eval_runtime": 223.2718,
-      "eval_samples_per_second": 920.447,
-      "eval_steps_per_second": 57.531,
+      "eval_loss": 2.47969388961792,
+      "eval_runtime": 224.2217,
+      "eval_samples_per_second": 916.548,
+      "eval_steps_per_second": 57.287,
       "step": 584000
     },
     {
       "epoch": 2.43,
-      "learning_rate": 7.537730342699909e-06,
-      "loss": 2.7081,
+      "learning_rate": 3.0886666666666667e-07,
+      "loss": 2.5583,
       "step": 592000
     },
     {
       "epoch": 2.43,
-      "eval_loss": 2.707876205444336,
-      "eval_runtime": 222.7056,
-      "eval_samples_per_second": 922.788,
-      "eval_steps_per_second": 57.677,
+      "eval_loss": 2.4872305393218994,
+      "eval_runtime": 224.237,
+      "eval_samples_per_second": 916.486,
+      "eval_steps_per_second": 57.283,
       "step": 592000
     },
     {
       "epoch": 2.46,
-      "eval_loss": 2.710313320159912,
-      "eval_runtime": 223.6841,
-      "eval_samples_per_second": 918.751,
-      "eval_steps_per_second": 57.425,
+      "eval_loss": 2.4812192916870117,
+      "eval_runtime": 222.6272,
+      "eval_samples_per_second": 923.113,
+      "eval_steps_per_second": 57.697,
       "step": 600000
     },
     {
       "epoch": 2.49,
-      "learning_rate": 7.471024764445928e-06,
-      "loss": 2.7124,
+      "learning_rate": 3.061333333333333e-07,
+      "loss": 2.5545,
       "step": 608000
     },
     {
       "epoch": 2.49,
-      "eval_loss": 2.7138588428497314,
-      "eval_runtime": 222.1167,
-      "eval_samples_per_second": 925.234,
-      "eval_steps_per_second": 57.83,
+      "eval_loss": 2.474827527999878,
+      "eval_runtime": 223.4042,
+      "eval_samples_per_second": 919.902,
+      "eval_steps_per_second": 57.497,
       "step": 608000
     },
     {
       "epoch": 2.52,
-      "eval_loss": 2.710862636566162,
-      "eval_runtime": 222.0676,
-      "eval_samples_per_second": 925.439,
-      "eval_steps_per_second": 57.843,
+      "eval_loss": 2.4735865592956543,
+      "eval_runtime": 224.1504,
+      "eval_samples_per_second": 916.84,
+      "eval_steps_per_second": 57.305,
       "step": 616000
     },
     {
       "epoch": 2.56,
-      "learning_rate": 7.4043191861919465e-06,
-      "loss": 2.7221,
+      "learning_rate": 3.034e-07,
+      "loss": 2.5561,
       "step": 624000
     },
     {
       "epoch": 2.56,
-      "eval_loss": 2.7152557373046875,
-      "eval_runtime": 223.9434,
-      "eval_samples_per_second": 917.687,
-      "eval_steps_per_second": 57.358,
+      "eval_loss": 2.4714128971099854,
+      "eval_runtime": 223.2085,
+      "eval_samples_per_second": 920.709,
+      "eval_steps_per_second": 57.547,
       "step": 624000
     },
     {
       "epoch": 2.59,
-      "eval_loss": 2.735896348953247,
-      "eval_runtime": 222.2379,
-      "eval_samples_per_second": 924.73,
-      "eval_steps_per_second": 57.798,
+      "eval_loss": 2.485759973526001,
+      "eval_runtime": 222.8361,
+      "eval_samples_per_second": 922.247,
+      "eval_steps_per_second": 57.643,
       "step": 632000
     },
     {
       "epoch": 2.62,
-      "learning_rate": 7.337613607937964e-06,
-      "loss": 2.7131,
+      "learning_rate": 3.0066666666666663e-07,
+      "loss": 2.5384,
       "step": 640000
     },
     {
       "epoch": 2.62,
-      "eval_loss": 2.7278735637664795,
-      "eval_runtime": 222.8278,
-      "eval_samples_per_second": 922.282,
-      "eval_steps_per_second": 57.645,
+      "eval_loss": 2.482938289642334,
+      "eval_runtime": 223.4494,
+      "eval_samples_per_second": 919.716,
+      "eval_steps_per_second": 57.485,
       "step": 640000
     },
     {
       "epoch": 2.66,
-      "eval_loss": 2.737804651260376,
-      "eval_runtime": 223.5516,
-      "eval_samples_per_second": 919.296,
-      "eval_steps_per_second": 57.459,
+      "eval_loss": 2.47662091255188,
+      "eval_runtime": 222.9171,
+      "eval_samples_per_second": 921.912,
+      "eval_steps_per_second": 57.622,
       "step": 648000
     },
     {
       "epoch": 2.69,
-      "learning_rate": 7.270908029683983e-06,
-      "loss": 2.7268,
+      "learning_rate": 2.9793333333333334e-07,
+      "loss": 2.541,
       "step": 656000
     },
     {
       "epoch": 2.69,
-      "eval_loss": 2.73801851272583,
-      "eval_runtime": 222.415,
-      "eval_samples_per_second": 923.993,
-      "eval_steps_per_second": 57.752,
+      "eval_loss": 2.4835963249206543,
+      "eval_runtime": 223.4062,
+      "eval_samples_per_second": 919.894,
+      "eval_steps_per_second": 57.496,
       "step": 656000
     },
     {
       "epoch": 2.72,
-      "eval_loss": 2.727534294128418,
-      "eval_runtime": 222.4068,
-      "eval_samples_per_second": 924.027,
-      "eval_steps_per_second": 57.755,
+      "eval_loss": 2.465118408203125,
+      "eval_runtime": 226.1239,
+      "eval_samples_per_second": 908.838,
+      "eval_steps_per_second": 56.805,
       "step": 664000
     },
     {
       "epoch": 2.75,
-      "learning_rate": 7.2042024514300015e-06,
-      "loss": 2.7373,
+      "learning_rate": 2.952e-07,
+      "loss": 2.5439,
       "step": 672000
     },
     {
       "epoch": 2.75,
-      "eval_loss": 2.7439777851104736,
-      "eval_runtime": 224.2928,
-      "eval_samples_per_second": 916.258,
-      "eval_steps_per_second": 57.269,
+      "eval_loss": 2.4797005653381348,
+      "eval_runtime": 224.1173,
+      "eval_samples_per_second": 916.975,
+      "eval_steps_per_second": 57.314,
       "step": 672000
     },
     {
       "epoch": 2.79,
-      "eval_loss": 2.7381510734558105,
-      "eval_runtime": 223.3597,
-      "eval_samples_per_second": 920.086,
-      "eval_steps_per_second": 57.508,
+      "eval_loss": 2.4702000617980957,
+      "eval_runtime": 223.8532,
+      "eval_samples_per_second": 918.057,
+      "eval_steps_per_second": 57.381,
       "step": 680000
     },
     {
       "epoch": 2.82,
-      "learning_rate": 7.13749687317602e-06,
-      "loss": 2.7576,
+      "learning_rate": 2.9246666666666665e-07,
+      "loss": 2.5597,
       "step": 688000
     },
     {
       "epoch": 2.82,
-      "eval_loss": 2.743011236190796,
-      "eval_runtime": 223.7189,
-      "eval_samples_per_second": 918.608,
-      "eval_steps_per_second": 57.416,
+      "eval_loss": 2.475144386291504,
+      "eval_runtime": 223.4589,
+      "eval_samples_per_second": 919.677,
+      "eval_steps_per_second": 57.483,
       "step": 688000
     },
     {
       "epoch": 2.85,
-      "eval_loss": 2.742142915725708,
-      "eval_runtime": 223.6634,
-      "eval_samples_per_second": 918.836,
-      "eval_steps_per_second": 57.43,
+      "eval_loss": 2.474367618560791,
+      "eval_runtime": 222.9092,
+      "eval_samples_per_second": 921.945,
+      "eval_steps_per_second": 57.624,
       "step": 696000
     },
     {
       "epoch": 2.88,
-      "learning_rate": 7.070791294922038e-06,
-      "loss": 2.7495,
+      "learning_rate": 2.897333333333333e-07,
+      "loss": 2.5491,
       "step": 704000
     },
     {
       "epoch": 2.88,
-      "eval_loss": 2.751878261566162,
-      "eval_runtime": 223.6777,
-      "eval_samples_per_second": 918.777,
-      "eval_steps_per_second": 57.426,
+      "eval_loss": 2.4756221771240234,
+      "eval_runtime": 223.5443,
+      "eval_samples_per_second": 919.325,
+      "eval_steps_per_second": 57.461,
       "step": 704000
     },
     {
       "epoch": 2.92,
-      "eval_loss": 2.749375104904175,
-      "eval_runtime": 223.9511,
-      "eval_samples_per_second": 917.656,
-      "eval_steps_per_second": 57.356,
+      "eval_loss": 2.4731247425079346,
+      "eval_runtime": 223.5397,
+      "eval_samples_per_second": 919.345,
+      "eval_steps_per_second": 57.462,
       "step": 712000
     },
     {
       "epoch": 2.95,
-      "learning_rate": 7.0040857166680564e-06,
-      "loss": 2.7626,
+      "learning_rate": 2.8699999999999996e-07,
+      "loss": 2.5505,
       "step": 720000
     },
     {
       "epoch": 2.95,
-      "eval_loss": 2.754082202911377,
-      "eval_runtime": 222.8179,
-      "eval_samples_per_second": 922.323,
-      "eval_steps_per_second": 57.648,
+      "eval_loss": 2.475615978240967,
+      "eval_runtime": 223.941,
+      "eval_samples_per_second": 917.697,
+      "eval_steps_per_second": 57.359,
       "step": 720000
     },
     {
       "epoch": 2.98,
-      "eval_loss": 2.7496798038482666,
-      "eval_runtime": 223.5475,
-      "eval_samples_per_second": 919.312,
-      "eval_steps_per_second": 57.46,
+      "eval_loss": 2.4703986644744873,
+      "eval_runtime": 224.1288,
+      "eval_samples_per_second": 916.928,
+      "eval_steps_per_second": 57.311,
       "step": 728000
     },
     {
       "epoch": 3.02,
-      "learning_rate": 6.937380138414076e-06,
-      "loss": 2.7551,
+      "learning_rate": 2.8426666666666667e-07,
+      "loss": 2.5432,
       "step": 736000
     },
     {
       "epoch": 3.02,
-      "eval_loss": 2.762470245361328,
-      "eval_runtime": 223.2771,
-      "eval_samples_per_second": 920.426,
-      "eval_steps_per_second": 57.529,
+      "eval_loss": 2.4762611389160156,
+      "eval_runtime": 223.7009,
+      "eval_samples_per_second": 918.682,
+      "eval_steps_per_second": 57.42,
       "step": 736000
     },
     {
       "epoch": 3.05,
-      "eval_loss": 2.764946937561035,
-      "eval_runtime": 222.8366,
-      "eval_samples_per_second": 922.245,
-      "eval_steps_per_second": 57.643,
+      "eval_loss": 2.4743261337280273,
+      "eval_runtime": 224.407,
+      "eval_samples_per_second": 915.791,
+      "eval_steps_per_second": 57.24,
       "step": 744000
     },
     {
       "epoch": 3.08,
-      "learning_rate": 6.8706745601600945e-06,
-      "loss": 2.7606,
+      "learning_rate": 2.815333333333333e-07,
+      "loss": 2.5485,
       "step": 752000
     },
     {
       "epoch": 3.08,
-      "eval_loss": 2.759011745452881,
-      "eval_runtime": 223.0437,
-      "eval_samples_per_second": 921.389,
-      "eval_steps_per_second": 57.59,
+      "eval_loss": 2.4626660346984863,
+      "eval_runtime": 224.0612,
+      "eval_samples_per_second": 917.205,
+      "eval_steps_per_second": 57.328,
       "step": 752000
     },
     {
       "epoch": 3.11,
-      "eval_loss": 2.7598087787628174,
-      "eval_runtime": 223.0646,
-      "eval_samples_per_second": 921.303,
-      "eval_steps_per_second": 57.584,
+      "eval_loss": 2.471444606781006,
+      "eval_runtime": 223.4318,
+      "eval_samples_per_second": 919.788,
+      "eval_steps_per_second": 57.49,
       "step": 760000
     },
     {
       "epoch": 3.15,
-      "learning_rate": 6.803968981906113e-06,
-      "loss": 2.7709,
+      "learning_rate": 2.7880000000000003e-07,
+      "loss": 2.5482,
       "step": 768000
     },
     {
       "epoch": 3.15,
-      "eval_loss": 2.7685601711273193,
-      "eval_runtime": 223.0947,
-      "eval_samples_per_second": 921.178,
-      "eval_steps_per_second": 57.576,
+      "eval_loss": 2.4684672355651855,
+      "eval_runtime": 224.1026,
+      "eval_samples_per_second": 917.035,
+      "eval_steps_per_second": 57.317,
       "step": 768000
     },
     {
       "epoch": 3.18,
-      "eval_loss": 2.769714117050171,
-      "eval_runtime": 224.3593,
-      "eval_samples_per_second": 915.986,
-      "eval_steps_per_second": 57.252,
+      "eval_loss": 2.4672694206237793,
+      "eval_runtime": 224.9545,
+      "eval_samples_per_second": 913.562,
+      "eval_steps_per_second": 57.1,
       "step": 776000
     },
     {
       "epoch": 3.21,
-      "learning_rate": 6.737263403652131e-06,
-      "loss": 2.7687,
+      "learning_rate": 2.7606666666666664e-07,
+      "loss": 2.5411,
       "step": 784000
     },
     {
       "epoch": 3.21,
-      "eval_loss": 2.7807157039642334,
-      "eval_runtime": 224.1624,
-      "eval_samples_per_second": 916.791,
-      "eval_steps_per_second": 57.302,
+      "eval_loss": 2.4726006984710693,
+      "eval_runtime": 224.2901,
+      "eval_samples_per_second": 916.269,
+      "eval_steps_per_second": 57.27,
       "step": 784000
     },
     {
       "epoch": 3.25,
-      "eval_loss": 2.783041000366211,
-      "eval_runtime": 224.6759,
-      "eval_samples_per_second": 914.695,
-      "eval_steps_per_second": 57.171,
+      "eval_loss": 2.476133108139038,
+      "eval_runtime": 224.2831,
+      "eval_samples_per_second": 916.297,
+      "eval_steps_per_second": 57.271,
       "step": 792000
     },
     {
       "epoch": 3.28,
-      "learning_rate": 6.6705578253981495e-06,
-      "loss": 2.7745,
+      "learning_rate": 2.733333333333333e-07,
+      "loss": 2.5407,
       "step": 800000
     },
     {
       "epoch": 3.28,
-      "eval_loss": 2.7677245140075684,
-      "eval_runtime": 223.5451,
-      "eval_samples_per_second": 919.322,
-      "eval_steps_per_second": 57.46,
+      "eval_loss": 2.4611737728118896,
+      "eval_runtime": 223.8119,
+      "eval_samples_per_second": 918.226,
+      "eval_steps_per_second": 57.392,
       "step": 800000
     },
     {
       "epoch": 3.31,
-      "eval_loss": 2.789714813232422,
-      "eval_runtime": 224.1358,
-      "eval_samples_per_second": 916.9,
-      "eval_steps_per_second": 57.309,
+      "eval_loss": 2.4742894172668457,
+      "eval_runtime": 224.8606,
+      "eval_samples_per_second": 913.944,
+      "eval_steps_per_second": 57.124,
       "step": 808000
     },
     {
       "epoch": 3.34,
-      "learning_rate": 6.603852247144168e-06,
-      "loss": 2.7596,
+      "learning_rate": 2.706e-07,
+      "loss": 2.5307,
       "step": 816000
     },
     {
       "epoch": 3.34,
-      "eval_loss": 2.782496690750122,
-      "eval_runtime": 224.2552,
-      "eval_samples_per_second": 916.411,
-      "eval_steps_per_second": 57.278,
+      "eval_loss": 2.469853401184082,
+      "eval_runtime": 224.177,
+      "eval_samples_per_second": 916.731,
+      "eval_steps_per_second": 57.298,
       "step": 816000
     },
     {
       "epoch": 3.38,
-      "eval_loss": 2.782914400100708,
-      "eval_runtime": 225.4775,
-      "eval_samples_per_second": 911.444,
-      "eval_steps_per_second": 56.968,
+      "eval_loss": 2.4721498489379883,
+      "eval_runtime": 223.5611,
+      "eval_samples_per_second": 919.256,
+      "eval_steps_per_second": 57.456,
       "step": 824000
     },
     {
       "epoch": 3.41,
-      "learning_rate": 6.537146668890187e-06,
-      "loss": 2.7749,
+      "learning_rate": 2.6786666666666666e-07,
+      "loss": 2.5391,
       "step": 832000
     },
     {
       "epoch": 3.41,
-      "eval_loss": 2.7868270874023438,
-      "eval_runtime": 224.6633,
-      "eval_samples_per_second": 914.747,
-      "eval_steps_per_second": 57.174,
+      "eval_loss": 2.461381435394287,
+      "eval_runtime": 224.3794,
+      "eval_samples_per_second": 915.904,
+      "eval_steps_per_second": 57.247,
       "step": 832000
     },
     {
       "epoch": 3.44,
-      "eval_loss": 2.7875137329101562,
-      "eval_runtime": 223.5713,
-      "eval_samples_per_second": 919.214,
-      "eval_steps_per_second": 57.454,
+      "eval_loss": 2.4641225337982178,
+      "eval_runtime": 224.9887,
+      "eval_samples_per_second": 913.423,
+      "eval_steps_per_second": 57.092,
       "step": 840000
     },
     {
       "epoch": 3.47,
-      "learning_rate": 6.4704410906362044e-06,
-      "loss": 2.7788,
+      "learning_rate": 2.651333333333333e-07,
+      "loss": 2.5378,
       "step": 848000
     },
     {
       "epoch": 3.47,
-      "eval_loss": 2.7906622886657715,
-      "eval_runtime": 223.506,
-      "eval_samples_per_second": 919.483,
-      "eval_steps_per_second": 57.471,
+      "eval_loss": 2.4652435779571533,
+      "eval_runtime": 225.299,
+      "eval_samples_per_second": 912.166,
+      "eval_steps_per_second": 57.013,
       "step": 848000
     },
     {
       "epoch": 3.51,
-      "eval_loss": 2.7915995121002197,
-      "eval_runtime": 225.1506,
-      "eval_samples_per_second": 912.767,
-      "eval_steps_per_second": 57.051,
+      "eval_loss": 2.4640512466430664,
+      "eval_runtime": 224.7841,
+      "eval_samples_per_second": 914.255,
+      "eval_steps_per_second": 57.144,
       "step": 856000
     },
     {
       "epoch": 3.54,
-      "learning_rate": 6.403735512382223e-06,
-      "loss": 2.7792,
+      "learning_rate": 2.624e-07,
+      "loss": 2.5399,
       "step": 864000
     },
     {
       "epoch": 3.54,
-      "eval_loss": 2.793468713760376,
-      "eval_runtime": 223.5626,
-      "eval_samples_per_second": 919.25,
-      "eval_steps_per_second": 57.456,
+      "eval_loss": 2.469067096710205,
+      "eval_runtime": 225.0567,
+      "eval_samples_per_second": 913.148,
+      "eval_steps_per_second": 57.075,
       "step": 864000
     },
     {
       "epoch": 3.57,
-      "eval_loss": 2.789273977279663,
-      "eval_runtime": 224.3088,
-      "eval_samples_per_second": 916.192,
-      "eval_steps_per_second": 57.265,
+      "eval_loss": 2.4611856937408447,
+      "eval_runtime": 224.6227,
+      "eval_samples_per_second": 914.912,
+      "eval_steps_per_second": 57.185,
       "step": 872000
     },
     {
       "epoch": 3.61,
-      "learning_rate": 6.337029934128242e-06,
-      "loss": 2.7871,
+      "learning_rate": 2.596666666666667e-07,
+      "loss": 2.5412,
       "step": 880000
     },
     {
       "epoch": 3.61,
-      "eval_loss": 2.8002703189849854,
-      "eval_runtime": 223.9319,
-      "eval_samples_per_second": 917.734,
-      "eval_steps_per_second": 57.361,
+      "eval_loss": 2.469621419906616,
+      "eval_runtime": 225.3239,
+      "eval_samples_per_second": 912.065,
+      "eval_steps_per_second": 57.007,
       "step": 880000
     },
     {
       "epoch": 3.64,
-      "eval_loss": 2.79728364944458,
-      "eval_runtime": 224.811,
-      "eval_samples_per_second": 914.145,
-      "eval_steps_per_second": 57.137,
+      "eval_loss": 2.4638073444366455,
+      "eval_runtime": 224.7878,
+      "eval_samples_per_second": 914.24,
+      "eval_steps_per_second": 57.143,
       "step": 888000
     },
     {
       "epoch": 3.67,
-      "learning_rate": 6.270324355874261e-06,
-      "loss": 2.7967,
+      "learning_rate": 2.5693333333333333e-07,
+      "loss": 2.5389,
       "step": 896000
     },
     {
       "epoch": 3.67,
-      "eval_loss": 2.810227394104004,
-      "eval_runtime": 223.8856,
-      "eval_samples_per_second": 917.924,
-      "eval_steps_per_second": 57.373,
+      "eval_loss": 2.4658303260803223,
+      "eval_runtime": 224.4987,
+      "eval_samples_per_second": 915.417,
+      "eval_steps_per_second": 57.216,
       "step": 896000
     },
     {
       "epoch": 3.7,
-      "eval_loss": 2.811846971511841,
-      "eval_runtime": 226.2552,
-      "eval_samples_per_second": 908.31,
-      "eval_steps_per_second": 56.772,
+      "eval_loss": 2.4725189208984375,
+      "eval_runtime": 226.711,
+      "eval_samples_per_second": 906.484,
+      "eval_steps_per_second": 56.658,
       "step": 904000
     },
     {
       "epoch": 3.74,
-      "learning_rate": 6.20361877762028e-06,
-      "loss": 2.7896,
+      "learning_rate": 2.542e-07,
+      "loss": 2.5325,
       "step": 912000
     },
     {
       "epoch": 3.74,
-      "eval_loss": 2.805860757827759,
-      "eval_runtime": 224.7634,
-      "eval_samples_per_second": 914.339,
-      "eval_steps_per_second": 57.149,
+      "eval_loss": 2.46415114402771,
+      "eval_runtime": 225.3655,
+      "eval_samples_per_second": 911.896,
+      "eval_steps_per_second": 56.996,
       "step": 912000
     },
     {
       "epoch": 3.77,
-      "eval_loss": 2.8134682178497314,
-      "eval_runtime": 223.9747,
-      "eval_samples_per_second": 917.559,
-      "eval_steps_per_second": 57.35,
+      "eval_loss": 2.4599404335021973,
+      "eval_runtime": 224.7087,
+      "eval_samples_per_second": 914.562,
+      "eval_steps_per_second": 57.163,
       "step": 920000
     },
     {
       "epoch": 3.8,
-      "learning_rate": 6.1369131993662975e-06,
-      "loss": 2.8021,
+      "learning_rate": 2.5146666666666664e-07,
+      "loss": 2.5351,
       "step": 928000
     },
     {
       "epoch": 3.8,
-      "eval_loss": 2.810849905014038,
-      "eval_runtime": 226.3005,
-      "eval_samples_per_second": 908.129,
-      "eval_steps_per_second": 56.761,
+      "eval_loss": 2.4616599082946777,
+      "eval_runtime": 226.6966,
+      "eval_samples_per_second": 906.542,
+      "eval_steps_per_second": 56.662,
       "step": 928000
     },
     {
       "epoch": 3.84,
-      "eval_loss": 2.8164422512054443,
-      "eval_runtime": 227.0284,
-      "eval_samples_per_second": 905.217,
-      "eval_steps_per_second": 56.579,
+      "eval_loss": 2.464627265930176,
+      "eval_runtime": 224.9933,
+      "eval_samples_per_second": 913.405,
+      "eval_steps_per_second": 57.091,
       "step": 936000
     },
     {
       "epoch": 3.87,
-      "learning_rate": 6.070207621112316e-06,
-      "loss": 2.7931,
+      "learning_rate": 2.4873333333333335e-07,
+      "loss": 2.522,
       "step": 944000
     },
     {
       "epoch": 3.87,
-      "eval_loss": 2.8173489570617676,
-      "eval_runtime": 225.5541,
-      "eval_samples_per_second": 911.134,
-      "eval_steps_per_second": 56.949,
+      "eval_loss": 2.4665021896362305,
+      "eval_runtime": 225.0978,
+      "eval_samples_per_second": 912.981,
+      "eval_steps_per_second": 57.064,
       "step": 944000
     },
     {
       "epoch": 3.9,
-      "eval_loss": 2.829514741897583,
-      "eval_runtime": 225.3778,
-      "eval_samples_per_second": 911.847,
-      "eval_steps_per_second": 56.993,
+      "eval_loss": 2.4761972427368164,
+      "eval_runtime": 224.2641,
+      "eval_samples_per_second": 916.375,
+      "eval_steps_per_second": 57.276,
       "step": 952000
     },
     {
       "epoch": 3.93,
-      "learning_rate": 6.003502042858335e-06,
-      "loss": 2.8105,
+      "learning_rate": 2.46e-07,
+      "loss": 2.5331,
       "step": 960000
     },
     {
       "epoch": 3.93,
-      "eval_loss": 2.8248379230499268,
-      "eval_runtime": 224.9909,
-      "eval_samples_per_second": 913.415,
-      "eval_steps_per_second": 57.091,
+      "eval_loss": 2.4668779373168945,
+      "eval_runtime": 225.069,
+      "eval_samples_per_second": 913.098,
+      "eval_steps_per_second": 57.071,
       "step": 960000
     },
     {
       "epoch": 3.97,
-      "eval_loss": 2.812260627746582,
-      "eval_runtime": 225.6651,
-      "eval_samples_per_second": 910.686,
-      "eval_steps_per_second": 56.921,
+      "eval_loss": 2.4549825191497803,
+      "eval_runtime": 224.677,
+      "eval_samples_per_second": 914.691,
+      "eval_steps_per_second": 57.171,
       "step": 968000
     },
     {
       "epoch": 4.0,
-      "learning_rate": 5.936796464604353e-06,
-      "loss": 2.805,
+      "learning_rate": 2.4326666666666666e-07,
+      "loss": 2.5276,
       "step": 976000
     },
     {
       "epoch": 4.0,
-      "eval_loss": 2.829453468322754,
-      "eval_runtime": 224.9048,
-      "eval_samples_per_second": 913.764,
-      "eval_steps_per_second": 57.113,
+      "eval_loss": 2.466226577758789,
+      "eval_runtime": 224.6009,
+      "eval_samples_per_second": 915.001,
+      "eval_steps_per_second": 57.19,
       "step": 976000
     },
     {
       "epoch": 4.03,
-      "eval_loss": 2.8286798000335693,
-      "eval_runtime": 224.572,
-      "eval_samples_per_second": 915.119,
-      "eval_steps_per_second": 57.198,
+      "eval_loss": 2.464536190032959,
+      "eval_runtime": 224.76,
+      "eval_samples_per_second": 914.353,
+      "eval_steps_per_second": 57.15,
       "step": 984000
     },
     {
       "epoch": 4.06,
-      "learning_rate": 5.870090886350371e-06,
-      "loss": 2.7959,
+      "learning_rate": 2.405333333333333e-07,
+      "loss": 2.5206,
       "step": 992000
     },
     {
       "epoch": 4.06,
-      "eval_loss": 2.826554298400879,
-      "eval_runtime": 224.6548,
-      "eval_samples_per_second": 914.781,
-      "eval_steps_per_second": 57.177,
+      "eval_loss": 2.4586591720581055,
+      "eval_runtime": 225.6548,
+      "eval_samples_per_second": 910.727,
+      "eval_steps_per_second": 56.923,
       "step": 992000
     },
     {
       "epoch": 4.1,
-      "eval_loss": 2.844482898712158,
-      "eval_runtime": 225.9544,
-      "eval_samples_per_second": 909.52,
-      "eval_steps_per_second": 56.848,
+      "eval_loss": 2.47253680229187,
+      "eval_runtime": 225.9081,
+      "eval_samples_per_second": 909.706,
+      "eval_steps_per_second": 56.859,
       "step": 1000000
     },
     {
       "epoch": 4.13,
-      "learning_rate": 5.80338530809639e-06,
-      "loss": 2.8126,
+      "learning_rate": 2.3779999999999997e-07,
+      "loss": 2.5294,
       "step": 1008000
     },
     {
       "epoch": 4.13,
-      "eval_loss": 2.8259875774383545,
-      "eval_runtime": 226.7424,
-      "eval_samples_per_second": 906.359,
-      "eval_steps_per_second": 56.65,
+      "eval_loss": 2.458824634552002,
+      "eval_runtime": 224.8489,
+      "eval_samples_per_second": 913.991,
+      "eval_steps_per_second": 57.127,
       "step": 1008000
     },
     {
       "epoch": 4.16,
-      "eval_loss": 2.829375743865967,
-      "eval_runtime": 225.2876,
-      "eval_samples_per_second": 912.212,
-      "eval_steps_per_second": 57.016,
+      "eval_loss": 2.4591076374053955,
+      "eval_runtime": 225.8271,
+      "eval_samples_per_second": 910.032,
+      "eval_steps_per_second": 56.88,
       "step": 1016000
     },
     {
       "epoch": 4.2,
-      "learning_rate": 5.736679729842408e-06,
-      "loss": 2.8085,
+      "learning_rate": 2.3506666666666668e-07,
+      "loss": 2.5312,
       "step": 1024000
     },
     {
       "epoch": 4.2,
-      "eval_loss": 2.8393983840942383,
-      "eval_runtime": 224.8219,
-      "eval_samples_per_second": 914.101,
-      "eval_steps_per_second": 57.134,
+      "eval_loss": 2.4680891036987305,
+      "eval_runtime": 226.319,
+      "eval_samples_per_second": 908.055,
+      "eval_steps_per_second": 56.756,
       "step": 1024000
     },
     {
       "epoch": 4.23,
-      "eval_loss": 2.835977077484131,
-      "eval_runtime": 226.9491,
-      "eval_samples_per_second": 905.533,
-      "eval_steps_per_second": 56.599,
+      "eval_loss": 2.4624712467193604,
+      "eval_runtime": 226.6472,
+      "eval_samples_per_second": 906.74,
+      "eval_steps_per_second": 56.674,
       "step": 1032000
     },
     {
       "epoch": 4.26,
-      "learning_rate": 5.669974151588427e-06,
-      "loss": 2.8179,
+      "learning_rate": 2.3233333333333334e-07,
+      "loss": 2.525,
       "step": 1040000
     },
     {
       "epoch": 4.26,
-      "eval_loss": 2.8393590450286865,
-      "eval_runtime": 226.0543,
-      "eval_samples_per_second": 909.118,
-      "eval_steps_per_second": 56.823,
+      "eval_loss": 2.4659371376037598,
+      "eval_runtime": 225.5287,
+      "eval_samples_per_second": 911.237,
+      "eval_steps_per_second": 56.955,
       "step": 1040000
     },
     {
       "epoch": 4.29,
-      "eval_loss": 2.840928077697754,
-      "eval_runtime": 226.338,
-      "eval_samples_per_second": 907.978,
-      "eval_steps_per_second": 56.751,
+      "eval_loss": 2.460909843444824,
+      "eval_runtime": 225.5374,
+      "eval_samples_per_second": 911.201,
+      "eval_steps_per_second": 56.953,
       "step": 1048000
     },
     {
       "epoch": 4.33,
-      "learning_rate": 5.603268573334446e-06,
-      "loss": 2.8212,
+      "learning_rate": 2.2960000000000002e-07,
+      "loss": 2.5318,
       "step": 1056000
     },
     {
       "epoch": 4.33,
-      "eval_loss": 2.834209442138672,
-      "eval_runtime": 225.5106,
-      "eval_samples_per_second": 911.31,
-      "eval_steps_per_second": 56.96,
+      "eval_loss": 2.4571011066436768,
+      "eval_runtime": 225.1138,
+      "eval_samples_per_second": 912.916,
+      "eval_steps_per_second": 57.06,
       "step": 1056000
     },
     {
       "epoch": 4.36,
-      "eval_loss": 2.8416330814361572,
-      "eval_runtime": 226.856,
-      "eval_samples_per_second": 905.905,
-      "eval_steps_per_second": 56.622,
+      "eval_loss": 2.4581968784332275,
+      "eval_runtime": 226.7154,
+      "eval_samples_per_second": 906.467,
+      "eval_steps_per_second": 56.657,
       "step": 1064000
     },
     {
       "epoch": 4.39,
-      "learning_rate": 5.536562995080464e-06,
-      "loss": 2.8236,
+      "learning_rate": 2.2686666666666667e-07,
+      "loss": 2.5332,
       "step": 1072000
     },
     {
       "epoch": 4.39,
-      "eval_loss": 2.839299440383911,
-      "eval_runtime": 228.289,
-      "eval_samples_per_second": 900.218,
-      "eval_steps_per_second": 56.266,
+      "eval_loss": 2.456618547439575,
+      "eval_runtime": 225.422,
+      "eval_samples_per_second": 911.668,
+      "eval_steps_per_second": 56.982,
       "step": 1072000
     },
     {
       "epoch": 4.43,
-      "eval_loss": 2.8458597660064697,
-      "eval_runtime": 228.7211,
-      "eval_samples_per_second": 898.518,
-      "eval_steps_per_second": 56.16,
+      "eval_loss": 2.4587738513946533,
+      "eval_runtime": 226.533,
+      "eval_samples_per_second": 907.197,
+      "eval_steps_per_second": 56.703,
       "step": 1080000
     },
     {
       "epoch": 4.46,
-      "learning_rate": 5.469857416826483e-06,
-      "loss": 2.8134,
+      "learning_rate": 2.2413333333333333e-07,
+      "loss": 2.5168,
       "step": 1088000
     },
     {
       "epoch": 4.46,
-      "eval_loss": 2.844362497329712,
-      "eval_runtime": 227.78,
-      "eval_samples_per_second": 902.23,
-      "eval_steps_per_second": 56.392,
+      "eval_loss": 2.4606146812438965,
+      "eval_runtime": 226.9924,
+      "eval_samples_per_second": 905.361,
+      "eval_steps_per_second": 56.588,
       "step": 1088000
     },
     {
       "epoch": 4.49,
-      "eval_loss": 2.843454360961914,
-      "eval_runtime": 228.7834,
-      "eval_samples_per_second": 898.273,
-      "eval_steps_per_second": 56.145,
+      "eval_loss": 2.4597506523132324,
+      "eval_runtime": 228.0241,
+      "eval_samples_per_second": 901.264,
+      "eval_steps_per_second": 56.332,
       "step": 1096000
     },
     {
       "epoch": 4.52,
-      "learning_rate": 5.403151838572501e-06,
-      "loss": 2.8196,
+      "learning_rate": 2.214e-07,
+      "loss": 2.5181,
       "step": 1104000
     },
     {
       "epoch": 4.52,
-      "eval_loss": 2.848449945449829,
-      "eval_runtime": 228.2467,
-      "eval_samples_per_second": 900.385,
-      "eval_steps_per_second": 56.277,
+      "eval_loss": 2.454252004623413,
+      "eval_runtime": 228.6537,
+      "eval_samples_per_second": 898.783,
+      "eval_steps_per_second": 56.177,
       "step": 1104000
     },
     {
       "epoch": 4.56,
-      "eval_loss": 2.8537204265594482,
-      "eval_runtime": 231.8913,
-      "eval_samples_per_second": 886.234,
-      "eval_steps_per_second": 55.392,
+      "eval_loss": 2.4619953632354736,
+      "eval_runtime": 226.293,
+      "eval_samples_per_second": 908.159,
+      "eval_steps_per_second": 56.763,
       "step": 1112000
     },
     {
       "epoch": 4.59,
-      "learning_rate": 5.33644626031852e-06,
-      "loss": 2.8284,
+      "learning_rate": 2.1866666666666667e-07,
+      "loss": 2.5246,
       "step": 1120000
     },
     {
       "epoch": 4.59,
-      "eval_loss": 2.8540537357330322,
-      "eval_runtime": 230.1324,
-      "eval_samples_per_second": 893.008,
-      "eval_steps_per_second": 55.816,
+      "eval_loss": 2.4638657569885254,
+      "eval_runtime": 228.163,
+      "eval_samples_per_second": 900.716,
+      "eval_steps_per_second": 56.297,
       "step": 1120000
     },
     {
       "epoch": 4.62,
-      "eval_loss": 2.8400697708129883,
-      "eval_runtime": 230.0912,
-      "eval_samples_per_second": 893.167,
-      "eval_steps_per_second": 55.826,
+      "eval_loss": 2.4556171894073486,
+      "eval_runtime": 228.3656,
+      "eval_samples_per_second": 899.917,
+      "eval_steps_per_second": 56.248,
       "step": 1128000
     },
     {
       "epoch": 4.65,
-      "learning_rate": 5.269740682064538e-06,
-      "loss": 2.8349,
+      "learning_rate": 2.1593333333333332e-07,
+      "loss": 2.5318,
       "step": 1136000
     },
     {
       "epoch": 4.65,
-      "eval_loss": 2.847630262374878,
-      "eval_runtime": 227.545,
-      "eval_samples_per_second": 903.162,
-      "eval_steps_per_second": 56.45,
+      "eval_loss": 2.457075595855713,
+      "eval_runtime": 227.349,
+      "eval_samples_per_second": 903.941,
+      "eval_steps_per_second": 56.499,
       "step": 1136000
     },
     {
       "epoch": 4.69,
-      "eval_loss": 2.8476176261901855,
-      "eval_runtime": 227.3305,
-      "eval_samples_per_second": 904.014,
-      "eval_steps_per_second": 56.504,
+      "eval_loss": 2.4636013507843018,
+      "eval_runtime": 228.1517,
+      "eval_samples_per_second": 900.76,
+      "eval_steps_per_second": 56.3,
       "step": 1144000
     },
     {
       "epoch": 4.72,
-      "learning_rate": 5.203035103810556e-06,
-      "loss": 2.8171,
+      "learning_rate": 2.132e-07,
+      "loss": 2.512,
       "step": 1152000
     },
     {
       "epoch": 4.72,
-      "eval_loss": 2.843763589859009,
-      "eval_runtime": 232.1395,
-      "eval_samples_per_second": 885.287,
-      "eval_steps_per_second": 55.333,
+      "eval_loss": 2.4567556381225586,
+      "eval_runtime": 228.1066,
+      "eval_samples_per_second": 900.938,
+      "eval_steps_per_second": 56.311,
       "step": 1152000
     },
     {
       "epoch": 4.75,
-      "eval_loss": 2.853471040725708,
-      "eval_runtime": 229.6265,
-      "eval_samples_per_second": 894.975,
-      "eval_steps_per_second": 55.939,
+      "eval_loss": 2.4644010066986084,
+      "eval_runtime": 228.916,
+      "eval_samples_per_second": 897.753,
+      "eval_steps_per_second": 56.112,
       "step": 1160000
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.136329525556575e-06,
-      "loss": 2.8264,
+      "learning_rate": 2.1046666666666666e-07,
+      "loss": 2.5174,
       "step": 1168000
     },
     {
       "epoch": 4.79,
-      "eval_loss": 2.842829704284668,
-      "eval_runtime": 228.2848,
-      "eval_samples_per_second": 900.235,
-      "eval_steps_per_second": 56.267,
+      "eval_loss": 2.4528720378875732,
+      "eval_runtime": 228.2634,
+      "eval_samples_per_second": 900.32,
+      "eval_steps_per_second": 56.273,
       "step": 1168000
     },
     {
       "epoch": 4.82,
-      "eval_loss": 2.855221748352051,
-      "eval_runtime": 229.9347,
-      "eval_samples_per_second": 893.775,
-      "eval_steps_per_second": 55.864,
+      "eval_loss": 2.4613921642303467,
+      "eval_runtime": 228.3765,
+      "eval_samples_per_second": 899.874,
+      "eval_steps_per_second": 56.245,
       "step": 1176000
     },
     {
       "epoch": 4.85,
-      "learning_rate": 5.0696239473025935e-06,
-      "loss": 2.8335,
+      "learning_rate": 2.0773333333333334e-07,
+      "loss": 2.5196,
       "step": 1184000
     },
     {
       "epoch": 4.85,
-      "eval_loss": 2.857285261154175,
-      "eval_runtime": 229.2467,
-      "eval_samples_per_second": 896.458,
-      "eval_steps_per_second": 56.031,
+      "eval_loss": 2.463758707046509,
+      "eval_runtime": 227.0091,
+      "eval_samples_per_second": 905.294,
+      "eval_steps_per_second": 56.584,
       "step": 1184000
     },
     {
       "epoch": 4.88,
-      "eval_loss": 2.850496292114258,
-      "eval_runtime": 228.4453,
-      "eval_samples_per_second": 899.603,
-      "eval_steps_per_second": 56.228,
+      "eval_loss": 2.453406572341919,
+      "eval_runtime": 227.3538,
+      "eval_samples_per_second": 903.921,
+      "eval_steps_per_second": 56.498,
       "step": 1192000
     },
     {
       "epoch": 4.92,
-      "learning_rate": 5.002918369048611e-06,
-      "loss": 2.8351,
+      "learning_rate": 2.05e-07,
+      "loss": 2.5248,
       "step": 1200000
     },
     {
       "epoch": 4.92,
-      "eval_loss": 2.8511581420898438,
-      "eval_runtime": 228.7507,
-      "eval_samples_per_second": 898.401,
-      "eval_steps_per_second": 56.153,
+      "eval_loss": 2.4553115367889404,
+      "eval_runtime": 227.0142,
+      "eval_samples_per_second": 905.274,
+      "eval_steps_per_second": 56.582,
       "step": 1200000
     },
     {
       "epoch": 4.95,
-      "eval_loss": 2.8499553203582764,
-      "eval_runtime": 230.0891,
-      "eval_samples_per_second": 893.176,
-      "eval_steps_per_second": 55.826,
+      "eval_loss": 2.453683853149414,
+      "eval_runtime": 226.8684,
+      "eval_samples_per_second": 905.855,
+      "eval_steps_per_second": 56.619,
       "step": 1208000
     },
     {
       "epoch": 4.98,
-      "learning_rate": 4.936212790794631e-06,
-      "loss": 2.8366,
+      "learning_rate": 2.0226666666666668e-07,
+      "loss": 2.5201,
       "step": 1216000
     },
     {
       "epoch": 4.98,
-      "eval_loss": 2.8569633960723877,
-      "eval_runtime": 229.1877,
-      "eval_samples_per_second": 896.689,
-      "eval_steps_per_second": 56.046,
+      "eval_loss": 2.4578709602355957,
+      "eval_runtime": 226.9695,
+      "eval_samples_per_second": 905.452,
+      "eval_steps_per_second": 56.594,
       "step": 1216000
     },
     {
       "epoch": 5.02,
-      "eval_loss": 2.846999168395996,
-      "eval_runtime": 228.106,
-      "eval_samples_per_second": 900.941,
-      "eval_steps_per_second": 56.312,
+      "eval_loss": 2.4524765014648438,
+      "eval_runtime": 226.5657,
+      "eval_samples_per_second": 907.066,
+      "eval_steps_per_second": 56.694,
       "step": 1224000
     },
     {
       "epoch": 5.05,
-      "learning_rate": 4.869507212540649e-06,
-      "loss": 2.8257,
+      "learning_rate": 1.9953333333333333e-07,
+      "loss": 2.5164,
       "step": 1232000
     },
     {
       "epoch": 5.05,
-      "eval_loss": 2.8638033866882324,
-      "eval_runtime": 227.5474,
-      "eval_samples_per_second": 903.153,
-      "eval_steps_per_second": 56.45,
+      "eval_loss": 2.4645235538482666,
+      "eval_runtime": 227.3689,
+      "eval_samples_per_second": 903.861,
+      "eval_steps_per_second": 56.494,
       "step": 1232000
     },
     {
       "epoch": 5.08,
-      "eval_loss": 2.8511650562286377,
-      "eval_runtime": 227.331,
-      "eval_samples_per_second": 904.012,
-      "eval_steps_per_second": 56.504,
+      "eval_loss": 2.447993040084839,
+      "eval_runtime": 228.8072,
+      "eval_samples_per_second": 898.18,
+      "eval_steps_per_second": 56.139,
       "step": 1240000
     },
     {
       "epoch": 5.11,
-      "learning_rate": 4.802801634286667e-06,
-      "loss": 2.8351,
+      "learning_rate": 1.968e-07,
+      "loss": 2.5186,
       "step": 1248000
     },
     {
       "epoch": 5.11,
-      "eval_loss": 2.8640658855438232,
-      "eval_runtime": 227.1737,
-      "eval_samples_per_second": 904.638,
-      "eval_steps_per_second": 56.543,
+      "eval_loss": 2.4605581760406494,
+      "eval_runtime": 229.2086,
+      "eval_samples_per_second": 896.607,
+      "eval_steps_per_second": 56.041,
       "step": 1248000
     },
     {
       "epoch": 5.15,
-      "eval_loss": 2.8679511547088623,
-      "eval_runtime": 227.7339,
-      "eval_samples_per_second": 902.413,
-      "eval_steps_per_second": 56.404,
+      "eval_loss": 2.4623043537139893,
+      "eval_runtime": 229.8264,
+      "eval_samples_per_second": 894.197,
+      "eval_steps_per_second": 55.89,
       "step": 1256000
     },
     {
       "epoch": 5.18,
-      "learning_rate": 4.7360960560326865e-06,
-      "loss": 2.8272,
+      "learning_rate": 1.9406666666666667e-07,
+      "loss": 2.5123,
       "step": 1264000
     },
     {
       "epoch": 5.18,
-      "eval_loss": 2.8521032333374023,
-      "eval_runtime": 226.8252,
-      "eval_samples_per_second": 906.028,
-      "eval_steps_per_second": 56.63,
+      "eval_loss": 2.456583261489868,
+      "eval_runtime": 230.427,
+      "eval_samples_per_second": 891.866,
+      "eval_steps_per_second": 55.744,
       "step": 1264000
     },
     {
       "epoch": 5.21,
-      "eval_loss": 2.8616464138031006,
-      "eval_runtime": 226.0986,
-      "eval_samples_per_second": 908.94,
-      "eval_steps_per_second": 56.812,
+      "eval_loss": 2.464402437210083,
+      "eval_runtime": 229.1803,
+      "eval_samples_per_second": 896.717,
+      "eval_steps_per_second": 56.048,
       "step": 1272000
     },
     {
       "epoch": 5.24,
-      "learning_rate": 4.669390477778704e-06,
-      "loss": 2.8346,
+      "learning_rate": 1.9133333333333333e-07,
+      "loss": 2.5233,
       "step": 1280000
     },
     {
       "epoch": 5.24,
-      "eval_loss": 2.8545360565185547,
-      "eval_runtime": 226.6533,
-      "eval_samples_per_second": 906.715,
-      "eval_steps_per_second": 56.672,
+      "eval_loss": 2.457606792449951,
+      "eval_runtime": 227.9824,
+      "eval_samples_per_second": 901.429,
+      "eval_steps_per_second": 56.342,
       "step": 1280000
     },
     {
       "epoch": 5.28,
-      "eval_loss": 2.847703695297241,
-      "eval_runtime": 226.3926,
-      "eval_samples_per_second": 907.759,
-      "eval_steps_per_second": 56.738,
+      "eval_loss": 2.451943874359131,
+      "eval_runtime": 229.374,
+      "eval_samples_per_second": 895.96,
+      "eval_steps_per_second": 56.0,
       "step": 1288000
     },
     {
       "epoch": 5.31,
-      "learning_rate": 4.602684899524723e-06,
-      "loss": 2.8211,
+      "learning_rate": 1.886e-07,
+      "loss": 2.513,
       "step": 1296000
     },
     {
       "epoch": 5.31,
-      "eval_loss": 2.860240936279297,
-      "eval_runtime": 229.1022,
-      "eval_samples_per_second": 897.023,
-      "eval_steps_per_second": 56.067,
+      "eval_loss": 2.456979513168335,
+      "eval_runtime": 228.051,
+      "eval_samples_per_second": 901.158,
+      "eval_steps_per_second": 56.325,
       "step": 1296000
     },
     {
       "epoch": 5.34,
-      "eval_loss": 2.857428550720215,
-      "eval_runtime": 227.3841,
-      "eval_samples_per_second": 903.801,
-      "eval_steps_per_second": 56.49,
+      "eval_loss": 2.462719202041626,
+      "eval_runtime": 228.489,
+      "eval_samples_per_second": 899.43,
+      "eval_steps_per_second": 56.217,
       "step": 1304000
     },
     {
       "epoch": 5.38,
-      "learning_rate": 4.5359793212707415e-06,
-      "loss": 2.8302,
+      "learning_rate": 1.8586666666666666e-07,
+      "loss": 2.5226,
       "step": 1312000
     },
     {
       "epoch": 5.38,
-      "eval_loss": 2.8489556312561035,
-      "eval_runtime": 227.6538,
-      "eval_samples_per_second": 902.73,
-      "eval_steps_per_second": 56.423,
+      "eval_loss": 2.449977397918701,
+      "eval_runtime": 227.9952,
+      "eval_samples_per_second": 901.379,
+      "eval_steps_per_second": 56.339,
       "step": 1312000
     },
     {
       "epoch": 5.41,
-      "eval_loss": 2.8547353744506836,
-      "eval_runtime": 226.9648,
-      "eval_samples_per_second": 905.471,
-      "eval_steps_per_second": 56.595,
+      "eval_loss": 2.4563188552856445,
+      "eval_runtime": 227.2759,
+      "eval_samples_per_second": 904.231,
+      "eval_steps_per_second": 56.517,
       "step": 1320000
     },
     {
       "epoch": 5.44,
-      "learning_rate": 4.46927374301676e-06,
-      "loss": 2.8317,
+      "learning_rate": 1.8313333333333332e-07,
+      "loss": 2.5222,
       "step": 1328000
     },
     {
       "epoch": 5.44,
-      "eval_loss": 2.8535571098327637,
-      "eval_runtime": 228.1487,
-      "eval_samples_per_second": 900.772,
-      "eval_steps_per_second": 56.301,
+      "eval_loss": 2.4521265029907227,
+      "eval_runtime": 226.8418,
+      "eval_samples_per_second": 905.962,
+      "eval_steps_per_second": 56.625,
       "step": 1328000
     },
     {
       "epoch": 5.47,
-      "eval_loss": 2.855344295501709,
-      "eval_runtime": 227.3885,
-      "eval_samples_per_second": 903.783,
-      "eval_steps_per_second": 56.489,
+      "eval_loss": 2.4591453075408936,
+      "eval_runtime": 226.8374,
+      "eval_samples_per_second": 905.98,
+      "eval_steps_per_second": 56.626,
       "step": 1336000
     },
     {
       "epoch": 5.51,
-      "learning_rate": 4.402568164762779e-06,
-      "loss": 2.83,
+      "learning_rate": 1.804e-07,
+      "loss": 2.5191,
       "step": 1344000
     },
     {
       "epoch": 5.51,
-      "eval_loss": 2.853558301925659,
-      "eval_runtime": 226.8054,
-      "eval_samples_per_second": 906.107,
-      "eval_steps_per_second": 56.634,
+      "eval_loss": 2.4508602619171143,
+      "eval_runtime": 228.6931,
+      "eval_samples_per_second": 898.628,
+      "eval_steps_per_second": 56.167,
       "step": 1344000
     },
     {
       "epoch": 5.54,
-      "eval_loss": 2.8564646244049072,
-      "eval_runtime": 227.7928,
-      "eval_samples_per_second": 902.18,
-      "eval_steps_per_second": 56.389,
+      "eval_loss": 2.455850124359131,
+      "eval_runtime": 228.2295,
+      "eval_samples_per_second": 900.453,
+      "eval_steps_per_second": 56.281,
       "step": 1352000
     },
     {
       "epoch": 5.57,
-      "learning_rate": 4.335862586508797e-06,
-      "loss": 2.8347,
+      "learning_rate": 1.7766666666666666e-07,
+      "loss": 2.5243,
       "step": 1360000
     },
     {
       "epoch": 5.57,
-      "eval_loss": 2.8444929122924805,
-      "eval_runtime": 226.9232,
-      "eval_samples_per_second": 905.637,
-      "eval_steps_per_second": 56.605,
+      "eval_loss": 2.4501898288726807,
+      "eval_runtime": 228.7596,
+      "eval_samples_per_second": 898.367,
+      "eval_steps_per_second": 56.151,
       "step": 1360000
     },
     {
       "epoch": 5.61,
-      "eval_loss": 2.854001998901367,
-      "eval_runtime": 227.1661,
-      "eval_samples_per_second": 904.668,
-      "eval_steps_per_second": 56.545,
+      "eval_loss": 2.4514639377593994,
+      "eval_runtime": 227.2719,
+      "eval_samples_per_second": 904.247,
+      "eval_steps_per_second": 56.518,
       "step": 1368000
     },
     {
       "epoch": 5.64,
-      "learning_rate": 4.269157008254816e-06,
-      "loss": 2.8253,
+      "learning_rate": 1.7493333333333334e-07,
+      "loss": 2.5157,
       "step": 1376000
     },
     {
       "epoch": 5.64,
-      "eval_loss": 2.8630056381225586,
-      "eval_runtime": 228.1691,
-      "eval_samples_per_second": 900.692,
-      "eval_steps_per_second": 56.296,
+      "eval_loss": 2.4562854766845703,
+      "eval_runtime": 227.9532,
+      "eval_samples_per_second": 901.545,
+      "eval_steps_per_second": 56.349,
       "step": 1376000
     },
     {
       "epoch": 5.67,
-      "eval_loss": 2.8591771125793457,
-      "eval_runtime": 228.3462,
-      "eval_samples_per_second": 899.993,
-      "eval_steps_per_second": 56.252,
+      "eval_loss": 2.452606678009033,
+      "eval_runtime": 227.4532,
+      "eval_samples_per_second": 903.527,
+      "eval_steps_per_second": 56.473,
       "step": 1384000
     },
     {
       "epoch": 5.7,
-      "learning_rate": 4.202451430000834e-06,
-      "loss": 2.8237,
+      "learning_rate": 1.722e-07,
+      "loss": 2.5162,
       "step": 1392000
     },
     {
       "epoch": 5.7,
-      "eval_loss": 2.8634839057922363,
-      "eval_runtime": 227.2681,
-      "eval_samples_per_second": 904.262,
-      "eval_steps_per_second": 56.519,
+      "eval_loss": 2.458620071411133,
+      "eval_runtime": 228.2374,
+      "eval_samples_per_second": 900.422,
+      "eval_steps_per_second": 56.279,
       "step": 1392000
     },
     {
       "epoch": 5.74,
-      "eval_loss": 2.862060308456421,
-      "eval_runtime": 226.8521,
-      "eval_samples_per_second": 905.921,
-      "eval_steps_per_second": 56.623,
+      "eval_loss": 2.458387613296509,
+      "eval_runtime": 228.0105,
+      "eval_samples_per_second": 901.318,
+      "eval_steps_per_second": 56.335,
       "step": 1400000
     },
     {
       "epoch": 5.77,
-      "learning_rate": 4.135745851746852e-06,
-      "loss": 2.8364,
+      "learning_rate": 1.6946666666666668e-07,
+      "loss": 2.5169,
       "step": 1408000
     },
     {
       "epoch": 5.77,
-      "eval_loss": 2.854534864425659,
-      "eval_runtime": 228.0962,
-      "eval_samples_per_second": 900.98,
-      "eval_steps_per_second": 56.314,
+      "eval_loss": 2.454158067703247,
+      "eval_runtime": 227.4312,
+      "eval_samples_per_second": 903.614,
+      "eval_steps_per_second": 56.479,
       "step": 1408000
     },
     {
       "epoch": 5.8,
-      "eval_loss": 2.8681845664978027,
-      "eval_runtime": 228.9004,
-      "eval_samples_per_second": 897.814,
-      "eval_steps_per_second": 56.116,
+      "eval_loss": 2.460242986679077,
+      "eval_runtime": 228.5958,
+      "eval_samples_per_second": 899.01,
+      "eval_steps_per_second": 56.191,
       "step": 1416000
     },
     {
       "epoch": 5.84,
-      "learning_rate": 4.069040273492872e-06,
-      "loss": 2.8289,
+      "learning_rate": 1.6673333333333333e-07,
+      "loss": 2.5127,
       "step": 1424000
     },
     {
       "epoch": 5.84,
-      "eval_loss": 2.867513418197632,
-      "eval_runtime": 228.9233,
-      "eval_samples_per_second": 897.724,
-      "eval_steps_per_second": 56.111,
+      "eval_loss": 2.458707809448242,
+      "eval_runtime": 228.0452,
+      "eval_samples_per_second": 901.181,
+      "eval_steps_per_second": 56.327,
       "step": 1424000
     },
     {
       "epoch": 5.87,
-      "eval_loss": 2.859713077545166,
-      "eval_runtime": 228.0494,
-      "eval_samples_per_second": 901.164,
-      "eval_steps_per_second": 56.326,
+      "eval_loss": 2.452913284301758,
+      "eval_runtime": 227.4908,
+      "eval_samples_per_second": 903.377,
+      "eval_steps_per_second": 56.464,
       "step": 1432000
     },
     {
       "epoch": 5.9,
-      "learning_rate": 4.0023346952388895e-06,
-      "loss": 2.8327,
+      "learning_rate": 1.64e-07,
+      "loss": 2.5144,
       "step": 1440000
     },
     {
       "epoch": 5.9,
-      "eval_loss": 2.8727521896362305,
-      "eval_runtime": 228.1742,
-      "eval_samples_per_second": 900.672,
-      "eval_steps_per_second": 56.295,
+      "eval_loss": 2.462021827697754,
+      "eval_runtime": 229.4885,
+      "eval_samples_per_second": 895.513,
+      "eval_steps_per_second": 55.972,
       "step": 1440000
     },
     {
       "epoch": 5.93,
-      "eval_loss": 2.8643882274627686,
-      "eval_runtime": 228.4395,
-      "eval_samples_per_second": 899.626,
-      "eval_steps_per_second": 56.229,
+      "eval_loss": 2.450927972793579,
+      "eval_runtime": 227.9748,
+      "eval_samples_per_second": 901.459,
+      "eval_steps_per_second": 56.344,
       "step": 1448000
     },
     {
       "epoch": 5.97,
-      "learning_rate": 3.935629116984908e-06,
-      "loss": 2.8407,
+      "learning_rate": 1.6126666666666667e-07,
+      "loss": 2.5175,
       "step": 1456000
     },
     {
       "epoch": 5.97,
-      "eval_loss": 2.864025354385376,
-      "eval_runtime": 228.8691,
-      "eval_samples_per_second": 897.937,
-      "eval_steps_per_second": 56.124,
+      "eval_loss": 2.4503204822540283,
+      "eval_runtime": 227.5178,
+      "eval_samples_per_second": 903.27,
+      "eval_steps_per_second": 56.457,
       "step": 1456000
     },
     {
       "epoch": 6.0,
-      "eval_loss": 2.8669943809509277,
-      "eval_runtime": 227.4377,
-      "eval_samples_per_second": 903.588,
-      "eval_steps_per_second": 56.477,
+      "eval_loss": 2.4545462131500244,
+      "eval_runtime": 227.7963,
+      "eval_samples_per_second": 902.165,
+      "eval_steps_per_second": 56.388,
       "step": 1464000
     },
     {
       "epoch": 6.03,
-      "learning_rate": 3.868923538730927e-06,
-      "loss": 2.8349,
+      "learning_rate": 1.5853333333333332e-07,
+      "loss": 2.5147,
       "step": 1472000
     },
     {
       "epoch": 6.03,
-      "eval_loss": 2.8555285930633545,
-      "eval_runtime": 228.413,
-      "eval_samples_per_second": 899.73,
-      "eval_steps_per_second": 56.236,
+      "eval_loss": 2.4440090656280518,
+      "eval_runtime": 227.8162,
+      "eval_samples_per_second": 902.087,
+      "eval_steps_per_second": 56.383,
       "step": 1472000
     },
     {
       "epoch": 6.06,
-      "eval_loss": 2.8778481483459473,
-      "eval_runtime": 228.1335,
-      "eval_samples_per_second": 900.832,
-      "eval_steps_per_second": 56.305,
+      "eval_loss": 2.457670211791992,
+      "eval_runtime": 228.5245,
+      "eval_samples_per_second": 899.291,
+      "eval_steps_per_second": 56.208,
       "step": 1480000
     },
     {
       "epoch": 6.1,
-      "learning_rate": 3.8022179604769453e-06,
-      "loss": 2.8395,
+      "learning_rate": 1.558e-07,
+      "loss": 2.5128,
       "step": 1488000
     },
     {
       "epoch": 6.1,
-      "eval_loss": 2.8752517700195312,
-      "eval_runtime": 229.2602,
-      "eval_samples_per_second": 896.405,
-      "eval_steps_per_second": 56.028,
+      "eval_loss": 2.456602096557617,
+      "eval_runtime": 230.1502,
+      "eval_samples_per_second": 892.939,
+      "eval_steps_per_second": 55.811,
       "step": 1488000
     },
     {
       "epoch": 6.13,
-      "eval_loss": 2.865680694580078,
-      "eval_runtime": 228.1396,
-      "eval_samples_per_second": 900.808,
-      "eval_steps_per_second": 56.303,
+      "eval_loss": 2.449889659881592,
+      "eval_runtime": 228.3041,
+      "eval_samples_per_second": 900.159,
+      "eval_steps_per_second": 56.263,
       "step": 1496000
     },
     {
       "epoch": 6.16,
-      "learning_rate": 3.735512382222964e-06,
-      "loss": 2.8364,
+      "learning_rate": 1.5306666666666666e-07,
+      "loss": 2.5168,
       "step": 1504000
     },
     {
       "epoch": 6.16,
-      "eval_loss": 2.864389419555664,
-      "eval_runtime": 228.8163,
-      "eval_samples_per_second": 898.144,
-      "eval_steps_per_second": 56.137,
+      "eval_loss": 2.4480044841766357,
+      "eval_runtime": 228.2508,
+      "eval_samples_per_second": 900.369,
+      "eval_steps_per_second": 56.276,
       "step": 1504000
     },
     {
       "epoch": 6.2,
-      "eval_loss": 2.8669490814208984,
-      "eval_runtime": 230.4793,
-      "eval_samples_per_second": 891.663,
-      "eval_steps_per_second": 55.732,
+      "eval_loss": 2.4436299800872803,
+      "eval_runtime": 229.3638,
+      "eval_samples_per_second": 896.0,
+      "eval_steps_per_second": 56.003,
       "step": 1512000
     },
     {
       "epoch": 6.23,
-      "learning_rate": 3.668806803968982e-06,
-      "loss": 2.85,
+      "learning_rate": 1.5033333333333332e-07,
+      "loss": 2.5225,
       "step": 1520000
     },
     {
       "epoch": 6.23,
-      "eval_loss": 2.863649845123291,
-      "eval_runtime": 228.5034,
-      "eval_samples_per_second": 899.374,
-      "eval_steps_per_second": 56.214,
+      "eval_loss": 2.446739912033081,
+      "eval_runtime": 228.4899,
+      "eval_samples_per_second": 899.427,
+      "eval_steps_per_second": 56.217,
       "step": 1520000
     },
     {
       "epoch": 6.26,
-      "eval_loss": 2.868006944656372,
-      "eval_runtime": 228.7374,
-      "eval_samples_per_second": 898.454,
-      "eval_steps_per_second": 56.156,
+      "eval_loss": 2.4519920349121094,
+      "eval_runtime": 228.2075,
+      "eval_samples_per_second": 900.54,
+      "eval_steps_per_second": 56.286,
       "step": 1528000
     },
     {
       "epoch": 6.29,
-      "learning_rate": 3.6021012257150007e-06,
-      "loss": 2.8359,
+      "learning_rate": 1.476e-07,
+      "loss": 2.5135,
       "step": 1536000
     },
     {
       "epoch": 6.29,
-      "eval_loss": 2.8751797676086426,
-      "eval_runtime": 228.8775,
-      "eval_samples_per_second": 897.904,
-      "eval_steps_per_second": 56.122,
+      "eval_loss": 2.4535210132598877,
+      "eval_runtime": 228.7342,
+      "eval_samples_per_second": 898.466,
+      "eval_steps_per_second": 56.157,
       "step": 1536000
     },
     {
       "epoch": 6.33,
-      "eval_loss": 2.871007204055786,
-      "eval_runtime": 229.7636,
-      "eval_samples_per_second": 894.441,
-      "eval_steps_per_second": 55.905,
+      "eval_loss": 2.4462831020355225,
+      "eval_runtime": 229.9473,
+      "eval_samples_per_second": 893.727,
+      "eval_steps_per_second": 55.861,
       "step": 1544000
     },
     {
       "epoch": 6.36,
-      "learning_rate": 3.535395647461019e-06,
-      "loss": 2.8451,
+      "learning_rate": 1.4486666666666665e-07,
+      "loss": 2.5161,
       "step": 1552000
     },
     {
       "epoch": 6.36,
-      "eval_loss": 2.876699686050415,
-      "eval_runtime": 229.2961,
-      "eval_samples_per_second": 896.265,
-      "eval_steps_per_second": 56.019,
+      "eval_loss": 2.4556400775909424,
+      "eval_runtime": 228.5872,
+      "eval_samples_per_second": 899.044,
+      "eval_steps_per_second": 56.193,
       "step": 1552000
     },
     {
       "epoch": 6.39,
-      "eval_loss": 2.882432699203491,
-      "eval_runtime": 228.5189,
-      "eval_samples_per_second": 899.313,
-      "eval_steps_per_second": 56.21,
+      "eval_loss": 2.4604580402374268,
+      "eval_runtime": 229.1233,
+      "eval_samples_per_second": 896.941,
+      "eval_steps_per_second": 56.062,
       "step": 1560000
     },
     {
       "epoch": 6.43,
-      "learning_rate": 3.468690069207038e-06,
-      "loss": 2.8359,
+      "learning_rate": 1.4213333333333334e-07,
+      "loss": 2.5144,
       "step": 1568000
     },
     {
       "epoch": 6.43,
-      "eval_loss": 2.8723318576812744,
-      "eval_runtime": 229.3117,
-      "eval_samples_per_second": 896.204,
-      "eval_steps_per_second": 56.015,
+      "eval_loss": 2.4516451358795166,
+      "eval_runtime": 229.9726,
+      "eval_samples_per_second": 893.628,
+      "eval_steps_per_second": 55.854,
       "step": 1568000
     },
     {
       "epoch": 6.46,
-      "eval_loss": 2.87728214263916,
-      "eval_runtime": 230.0667,
-      "eval_samples_per_second": 893.263,
-      "eval_steps_per_second": 55.832,
+      "eval_loss": 2.4487648010253906,
+      "eval_runtime": 229.4253,
+      "eval_samples_per_second": 895.76,
+      "eval_steps_per_second": 55.988,
       "step": 1576000
     },
     {
       "epoch": 6.49,
-      "learning_rate": 3.4019844909530565e-06,
-      "loss": 2.8546,
+      "learning_rate": 1.3940000000000002e-07,
+      "loss": 2.5209,
       "step": 1584000
     },
     {
       "epoch": 6.49,
-      "eval_loss": 2.875946044921875,
-      "eval_runtime": 228.6285,
-      "eval_samples_per_second": 898.882,
-      "eval_steps_per_second": 56.183,
+      "eval_loss": 2.4525067806243896,
+      "eval_runtime": 228.8527,
+      "eval_samples_per_second": 898.001,
+      "eval_steps_per_second": 56.128,
       "step": 1584000
     },
     {
       "epoch": 6.52,
-      "eval_loss": 2.8732051849365234,
-      "eval_runtime": 231.6471,
-      "eval_samples_per_second": 887.168,
-      "eval_steps_per_second": 55.451,
+      "eval_loss": 2.450185537338257,
+      "eval_runtime": 230.8087,
+      "eval_samples_per_second": 890.391,
+      "eval_steps_per_second": 55.652,
       "step": 1592000
     },
     {
       "epoch": 6.56,
-      "learning_rate": 3.3352789126990747e-06,
-      "loss": 2.8395,
+      "learning_rate": 1.3666666666666665e-07,
+      "loss": 2.5102,
       "step": 1600000
     },
     {
       "epoch": 6.56,
-      "eval_loss": 2.8803162574768066,
-      "eval_runtime": 230.2291,
-      "eval_samples_per_second": 892.632,
-      "eval_steps_per_second": 55.792,
+      "eval_loss": 2.453780174255371,
+      "eval_runtime": 229.4733,
+      "eval_samples_per_second": 895.573,
+      "eval_steps_per_second": 55.976,
       "step": 1600000
     },
     {
       "epoch": 6.59,
-      "eval_loss": 2.8761353492736816,
-      "eval_runtime": 229.013,
-      "eval_samples_per_second": 897.373,
-      "eval_steps_per_second": 56.089,
+      "eval_loss": 2.4490787982940674,
+      "eval_runtime": 229.059,
+      "eval_samples_per_second": 897.192,
+      "eval_steps_per_second": 56.077,
       "step": 1608000
     },
     {
       "epoch": 6.62,
-      "learning_rate": 3.2685733344450933e-06,
-      "loss": 2.847,
+      "learning_rate": 1.3393333333333333e-07,
+      "loss": 2.5176,
       "step": 1616000
     },
     {
       "epoch": 6.62,
-      "eval_loss": 2.880068302154541,
-      "eval_runtime": 229.0344,
-      "eval_samples_per_second": 897.289,
-      "eval_steps_per_second": 56.083,
+      "eval_loss": 2.452752113342285,
+      "eval_runtime": 228.4962,
+      "eval_samples_per_second": 899.402,
+      "eval_steps_per_second": 56.215,
       "step": 1616000
     },
     {
       "epoch": 6.65,
-      "eval_loss": 2.873690366744995,
-      "eval_runtime": 228.9162,
-      "eval_samples_per_second": 897.752,
-      "eval_steps_per_second": 56.112,
+      "eval_loss": 2.44599986076355,
+      "eval_runtime": 228.5114,
+      "eval_samples_per_second": 899.342,
+      "eval_steps_per_second": 56.212,
       "step": 1624000
     },
     {
       "epoch": 6.69,
-      "learning_rate": 3.2018677561911115e-06,
-      "loss": 2.8555,
+      "learning_rate": 1.312e-07,
+      "loss": 2.5208,
       "step": 1632000
     },
     {
       "epoch": 6.69,
-      "eval_loss": 2.8797101974487305,
-      "eval_runtime": 228.8191,
-      "eval_samples_per_second": 898.133,
-      "eval_steps_per_second": 56.136,
+      "eval_loss": 2.4484992027282715,
+      "eval_runtime": 230.1605,
+      "eval_samples_per_second": 892.899,
+      "eval_steps_per_second": 55.809,
       "step": 1632000
     },
     {
       "epoch": 6.72,
-      "eval_loss": 2.8781728744506836,
-      "eval_runtime": 229.3564,
-      "eval_samples_per_second": 896.029,
-      "eval_steps_per_second": 56.005,
+      "eval_loss": 2.451284646987915,
+      "eval_runtime": 229.1401,
+      "eval_samples_per_second": 896.875,
+      "eval_steps_per_second": 56.057,
       "step": 1640000
     },
     {
       "epoch": 6.75,
-      "learning_rate": 3.1351621779371306e-06,
-      "loss": 2.8377,
+      "learning_rate": 1.2846666666666667e-07,
+      "loss": 2.5064,
       "step": 1648000
     },
     {
       "epoch": 6.75,
-      "eval_loss": 2.8825597763061523,
-      "eval_runtime": 229.3988,
-      "eval_samples_per_second": 895.863,
-      "eval_steps_per_second": 55.994,
+      "eval_loss": 2.451927900314331,
+      "eval_runtime": 229.3071,
+      "eval_samples_per_second": 896.222,
+      "eval_steps_per_second": 56.017,
       "step": 1648000
     },
     {
       "epoch": 6.79,
-      "eval_loss": 2.879844903945923,
-      "eval_runtime": 230.5279,
-      "eval_samples_per_second": 891.476,
-      "eval_steps_per_second": 55.72,
+      "eval_loss": 2.449305295944214,
+      "eval_runtime": 231.2204,
+      "eval_samples_per_second": 888.806,
+      "eval_steps_per_second": 55.553,
       "step": 1656000
     },
     {
       "epoch": 6.82,
-      "learning_rate": 3.0684565996831487e-06,
-      "loss": 2.8517,
+      "learning_rate": 1.2573333333333332e-07,
+      "loss": 2.5111,
       "step": 1664000
     },
     {
       "epoch": 6.82,
-      "eval_loss": 2.8799245357513428,
-      "eval_runtime": 230.1674,
-      "eval_samples_per_second": 892.872,
-      "eval_steps_per_second": 55.807,
+      "eval_loss": 2.4505178928375244,
+      "eval_runtime": 230.2462,
+      "eval_samples_per_second": 892.566,
+      "eval_steps_per_second": 55.788,
       "step": 1664000
     },
     {
       "epoch": 6.85,
-      "eval_loss": 2.8835196495056152,
-      "eval_runtime": 229.3852,
-      "eval_samples_per_second": 895.917,
-      "eval_steps_per_second": 55.998,
+      "eval_loss": 2.4501988887786865,
+      "eval_runtime": 229.973,
+      "eval_samples_per_second": 893.627,
+      "eval_steps_per_second": 55.854,
       "step": 1672000
     },
     {
       "epoch": 6.88,
-      "learning_rate": 3.0017510214291673e-06,
-      "loss": 2.8526,
+      "learning_rate": 1.23e-07,
+      "loss": 2.5141,
       "step": 1680000
     },
     {
       "epoch": 6.88,
-      "eval_loss": 2.8874735832214355,
-      "eval_runtime": 229.3731,
-      "eval_samples_per_second": 895.964,
-      "eval_steps_per_second": 56.0,
+      "eval_loss": 2.4560253620147705,
+      "eval_runtime": 229.6465,
+      "eval_samples_per_second": 894.897,
+      "eval_steps_per_second": 55.934,
       "step": 1680000
     },
     {
       "epoch": 6.92,
-      "eval_loss": 2.882905960083008,
-      "eval_runtime": 231.0041,
-      "eval_samples_per_second": 889.638,
-      "eval_steps_per_second": 55.605,
+      "eval_loss": 2.4499940872192383,
+      "eval_runtime": 229.0726,
+      "eval_samples_per_second": 897.139,
+      "eval_steps_per_second": 56.074,
       "step": 1688000
     },
     {
       "epoch": 6.95,
-      "learning_rate": 2.9350454431751855e-06,
-      "loss": 2.8511,
+      "learning_rate": 1.2026666666666666e-07,
+      "loss": 2.5089,
       "step": 1696000
     },
     {
       "epoch": 6.95,
-      "eval_loss": 2.8908419609069824,
-      "eval_runtime": 229.2389,
-      "eval_samples_per_second": 896.488,
-      "eval_steps_per_second": 56.033,
+      "eval_loss": 2.4512550830841064,
+      "eval_runtime": 228.4897,
+      "eval_samples_per_second": 899.428,
+      "eval_steps_per_second": 56.217,
       "step": 1696000
     },
     {
       "epoch": 6.98,
-      "eval_loss": 2.8755762577056885,
-      "eval_runtime": 230.9816,
-      "eval_samples_per_second": 889.724,
-      "eval_steps_per_second": 55.61,
+      "eval_loss": 2.4418201446533203,
+      "eval_runtime": 229.377,
+      "eval_samples_per_second": 895.949,
+      "eval_steps_per_second": 56.0,
       "step": 1704000
     },
     {
       "epoch": 7.02,
-      "learning_rate": 2.868339864921204e-06,
-      "loss": 2.8606,
+      "learning_rate": 1.1753333333333334e-07,
+      "loss": 2.5174,
       "step": 1712000
     },
     {
       "epoch": 7.02,
-      "eval_loss": 2.882680654525757,
-      "eval_runtime": 230.7858,
-      "eval_samples_per_second": 890.479,
-      "eval_steps_per_second": 55.658,
+      "eval_loss": 2.447690010070801,
+      "eval_runtime": 231.0137,
+      "eval_samples_per_second": 889.601,
+      "eval_steps_per_second": 55.603,
       "step": 1712000
     },
     {
       "epoch": 7.05,
-      "eval_loss": 2.884394407272339,
-      "eval_runtime": 230.0457,
-      "eval_samples_per_second": 893.344,
-      "eval_steps_per_second": 55.837,
+      "eval_loss": 2.450817584991455,
+      "eval_runtime": 231.6212,
+      "eval_samples_per_second": 887.268,
+      "eval_steps_per_second": 55.457,
       "step": 1720000
     },
     {
       "epoch": 7.08,
-      "learning_rate": 2.801634286667223e-06,
-      "loss": 2.852,
+      "learning_rate": 1.1480000000000001e-07,
+      "loss": 2.5198,
       "step": 1728000
     },
     {
       "epoch": 7.08,
-      "eval_loss": 2.8865275382995605,
-      "eval_runtime": 229.7673,
-      "eval_samples_per_second": 894.427,
-      "eval_steps_per_second": 55.904,
+      "eval_loss": 2.448648691177368,
+      "eval_runtime": 230.9308,
+      "eval_samples_per_second": 889.92,
+      "eval_steps_per_second": 55.623,
       "step": 1728000
     },
     {
       "epoch": 7.11,
-      "eval_loss": 2.8909904956817627,
-      "eval_runtime": 230.4684,
-      "eval_samples_per_second": 891.706,
-      "eval_steps_per_second": 55.734,
+      "eval_loss": 2.4577322006225586,
+      "eval_runtime": 230.4865,
+      "eval_samples_per_second": 891.636,
+      "eval_steps_per_second": 55.73,
       "step": 1736000
     },
     {
       "epoch": 7.15,
-      "learning_rate": 2.7349287084132413e-06,
-      "loss": 2.8319,
+      "learning_rate": 1.1206666666666666e-07,
+      "loss": 2.4974,
       "step": 1744000
     },
     {
       "epoch": 7.15,
-      "eval_loss": 2.884788751602173,
-      "eval_runtime": 229.622,
-      "eval_samples_per_second": 894.993,
-      "eval_steps_per_second": 55.94,
+      "eval_loss": 2.4416255950927734,
+      "eval_runtime": 229.8237,
+      "eval_samples_per_second": 894.207,
+      "eval_steps_per_second": 55.891,
       "step": 1744000
     },
     {
       "epoch": 7.18,
-      "eval_loss": 2.8916478157043457,
-      "eval_runtime": 231.4898,
-      "eval_samples_per_second": 887.771,
-      "eval_steps_per_second": 55.488,
+      "eval_loss": 2.4549336433410645,
+      "eval_runtime": 229.7571,
+      "eval_samples_per_second": 894.466,
+      "eval_steps_per_second": 55.907,
       "step": 1752000
     },
     {
       "epoch": 7.21,
-      "learning_rate": 2.66822313015926e-06,
-      "loss": 2.842,
+      "learning_rate": 1.0933333333333333e-07,
+      "loss": 2.5016,
       "step": 1760000
     },
     {
       "epoch": 7.21,
-      "eval_loss": 2.8830392360687256,
-      "eval_runtime": 230.1781,
-      "eval_samples_per_second": 892.83,
-      "eval_steps_per_second": 55.805,
+      "eval_loss": 2.455679416656494,
+      "eval_runtime": 230.1335,
+      "eval_samples_per_second": 893.003,
+      "eval_steps_per_second": 55.815,
       "step": 1760000
     },
     {
       "epoch": 7.24,
-      "eval_loss": 2.8850438594818115,
-      "eval_runtime": 230.145,
-      "eval_samples_per_second": 892.959,
-      "eval_steps_per_second": 55.813,
+      "eval_loss": 2.4531571865081787,
+      "eval_runtime": 231.3847,
+      "eval_samples_per_second": 888.175,
+      "eval_steps_per_second": 55.514,
       "step": 1768000
     },
     {
       "epoch": 7.28,
-      "learning_rate": 2.601517551905278e-06,
-      "loss": 2.8421,
+      "learning_rate": 1.066e-07,
+      "loss": 2.5112,
       "step": 1776000
     },
     {
       "epoch": 7.28,
-      "eval_loss": 2.875260591506958,
-      "eval_runtime": 230.4814,
-      "eval_samples_per_second": 891.655,
-      "eval_steps_per_second": 55.731,
+      "eval_loss": 2.445054531097412,
+      "eval_runtime": 231.2999,
+      "eval_samples_per_second": 888.5,
+      "eval_steps_per_second": 55.534,
       "step": 1776000
     },
     {
       "epoch": 7.31,
-      "eval_loss": 2.8958091735839844,
-      "eval_runtime": 230.6404,
-      "eval_samples_per_second": 891.041,
-      "eval_steps_per_second": 55.693,
+      "eval_loss": 2.460723638534546,
+      "eval_runtime": 230.196,
+      "eval_samples_per_second": 892.761,
+      "eval_steps_per_second": 55.8,
       "step": 1784000
     },
     {
       "epoch": 7.34,
-      "learning_rate": 2.5348119736512967e-06,
-      "loss": 2.8558,
+      "learning_rate": 1.0386666666666667e-07,
+      "loss": 2.5172,
       "step": 1792000
     },
     {
       "epoch": 7.34,
-      "eval_loss": 2.8712708950042725,
-      "eval_runtime": 229.8418,
-      "eval_samples_per_second": 894.137,
-      "eval_steps_per_second": 55.886,
+      "eval_loss": 2.4451537132263184,
+      "eval_runtime": 231.1406,
+      "eval_samples_per_second": 889.112,
+      "eval_steps_per_second": 55.572,
       "step": 1792000
     },
     {
       "epoch": 7.38,
-      "eval_loss": 2.8743996620178223,
-      "eval_runtime": 231.7487,
-      "eval_samples_per_second": 886.779,
-      "eval_steps_per_second": 55.426,
+      "eval_loss": 2.4426777362823486,
+      "eval_runtime": 230.7159,
+      "eval_samples_per_second": 890.749,
+      "eval_steps_per_second": 55.675,
       "step": 1800000
     },
     {
       "epoch": 7.41,
-      "learning_rate": 2.4681063953973154e-06,
-      "loss": 2.8382,
+      "learning_rate": 1.0113333333333334e-07,
+      "loss": 2.5089,
       "step": 1808000
     },
     {
       "epoch": 7.41,
-      "eval_loss": 2.890810966491699,
-      "eval_runtime": 230.3071,
-      "eval_samples_per_second": 892.33,
-      "eval_steps_per_second": 55.773,
+      "eval_loss": 2.4511077404022217,
+      "eval_runtime": 231.5975,
+      "eval_samples_per_second": 887.359,
+      "eval_steps_per_second": 55.463,
       "step": 1808000
     },
     {
       "epoch": 7.44,
-      "eval_loss": 2.8748860359191895,
-      "eval_runtime": 230.3829,
-      "eval_samples_per_second": 892.037,
-      "eval_steps_per_second": 55.755,
+      "eval_loss": 2.4440526962280273,
+      "eval_runtime": 231.4447,
+      "eval_samples_per_second": 887.944,
+      "eval_steps_per_second": 55.499,
       "step": 1816000
     },
     {
       "epoch": 7.47,
-      "learning_rate": 2.4014008171433335e-06,
-      "loss": 2.8508,
+      "learning_rate": 9.84e-08,
+      "loss": 2.5136,
       "step": 1824000
     },
     {
       "epoch": 7.47,
-      "eval_loss": 2.878951072692871,
-      "eval_runtime": 230.3626,
-      "eval_samples_per_second": 892.115,
-      "eval_steps_per_second": 55.76,
+      "eval_loss": 2.4492361545562744,
+      "eval_runtime": 231.7181,
+      "eval_samples_per_second": 886.896,
+      "eval_steps_per_second": 55.434,
       "step": 1824000
     },
     {
       "epoch": 7.51,
-      "eval_loss": 2.886600971221924,
-      "eval_runtime": 230.5556,
-      "eval_samples_per_second": 891.368,
-      "eval_steps_per_second": 55.713,
+      "eval_loss": 2.4523823261260986,
+      "eval_runtime": 231.3659,
+      "eval_samples_per_second": 888.247,
+      "eval_steps_per_second": 55.518,
       "step": 1832000
     },
     {
       "epoch": 7.54,
-      "learning_rate": 2.334695238889352e-06,
-      "loss": 2.8477,
+      "learning_rate": 9.566666666666666e-08,
+      "loss": 2.509,
       "step": 1840000
     },
     {
       "epoch": 7.54,
-      "eval_loss": 2.8806040287017822,
-      "eval_runtime": 230.3727,
-      "eval_samples_per_second": 892.076,
-      "eval_steps_per_second": 55.757,
+      "eval_loss": 2.451181411743164,
+      "eval_runtime": 230.8127,
+      "eval_samples_per_second": 890.376,
+      "eval_steps_per_second": 55.651,
       "step": 1840000
     },
     {
       "epoch": 7.57,
-      "eval_loss": 2.8820905685424805,
-      "eval_runtime": 230.4476,
-      "eval_samples_per_second": 891.786,
-      "eval_steps_per_second": 55.739,
+      "eval_loss": 2.4528069496154785,
+      "eval_runtime": 230.6096,
+      "eval_samples_per_second": 891.16,
+      "eval_steps_per_second": 55.7,
       "step": 1848000
     },
     {
       "epoch": 7.61,
-      "learning_rate": 2.2679896606353707e-06,
-      "loss": 2.8497,
+      "learning_rate": 9.293333333333333e-08,
+      "loss": 2.5157,
       "step": 1856000
     },
     {
       "epoch": 7.61,
-      "eval_loss": 2.877049684524536,
-      "eval_runtime": 230.7875,
-      "eval_samples_per_second": 890.473,
-      "eval_steps_per_second": 55.657,
+      "eval_loss": 2.4439537525177,
+      "eval_runtime": 233.382,
+      "eval_samples_per_second": 880.573,
+      "eval_steps_per_second": 55.039,
       "step": 1856000
     },
     {
       "epoch": 7.64,
-      "eval_loss": 2.873206377029419,
-      "eval_runtime": 230.6269,
-      "eval_samples_per_second": 891.093,
-      "eval_steps_per_second": 55.696,
+      "eval_loss": 2.4401602745056152,
+      "eval_runtime": 231.584,
+      "eval_samples_per_second": 887.41,
+      "eval_steps_per_second": 55.466,
       "step": 1864000
     },
     {
       "epoch": 7.67,
-      "learning_rate": 2.2012840823813894e-06,
-      "loss": 2.8566,
+      "learning_rate": 9.02e-08,
+      "loss": 2.5181,
       "step": 1872000
     },
     {
       "epoch": 7.67,
-      "eval_loss": 2.887915849685669,
-      "eval_runtime": 230.6589,
-      "eval_samples_per_second": 890.969,
-      "eval_steps_per_second": 55.688,
+      "eval_loss": 2.4537830352783203,
+      "eval_runtime": 230.4518,
+      "eval_samples_per_second": 891.77,
+      "eval_steps_per_second": 55.738,
       "step": 1872000
     },
     {
       "epoch": 7.7,
-      "eval_loss": 2.876028537750244,
-      "eval_runtime": 230.4372,
-      "eval_samples_per_second": 891.826,
-      "eval_steps_per_second": 55.742,
+      "eval_loss": 2.4480724334716797,
+      "eval_runtime": 229.9532,
+      "eval_samples_per_second": 893.703,
+      "eval_steps_per_second": 55.859,
       "step": 1880000
     },
     {
       "epoch": 7.74,
-      "learning_rate": 2.134578504127408e-06,
-      "loss": 2.8527,
+      "learning_rate": 8.746666666666667e-08,
+      "loss": 2.5145,
       "step": 1888000
     },
     {
       "epoch": 7.74,
-      "eval_loss": 2.8763537406921387,
-      "eval_runtime": 232.1508,
-      "eval_samples_per_second": 885.244,
-      "eval_steps_per_second": 55.33,
+      "eval_loss": 2.4417428970336914,
+      "eval_runtime": 231.464,
+      "eval_samples_per_second": 887.87,
+      "eval_steps_per_second": 55.495,
       "step": 1888000
     },
     {
       "epoch": 7.77,
-      "eval_loss": 2.8837802410125732,
-      "eval_runtime": 231.7421,
-      "eval_samples_per_second": 886.805,
-      "eval_steps_per_second": 55.428,
+      "eval_loss": 2.4512147903442383,
+      "eval_runtime": 231.0711,
+      "eval_samples_per_second": 889.38,
+      "eval_steps_per_second": 55.589,
       "step": 1896000
     },
     {
       "epoch": 7.8,
-      "learning_rate": 2.067872925873426e-06,
-      "loss": 2.8438,
+      "learning_rate": 8.473333333333334e-08,
+      "loss": 2.5013,
       "step": 1904000
     },
     {
       "epoch": 7.8,
-      "eval_loss": 2.8954691886901855,
-      "eval_runtime": 231.1122,
-      "eval_samples_per_second": 889.222,
-      "eval_steps_per_second": 55.579,
+      "eval_loss": 2.45603084564209,
+      "eval_runtime": 231.877,
+      "eval_samples_per_second": 886.289,
+      "eval_steps_per_second": 55.396,
       "step": 1904000
     },
     {
       "epoch": 7.83,
-      "eval_loss": 2.88919997215271,
-      "eval_runtime": 232.0892,
-      "eval_samples_per_second": 885.478,
-      "eval_steps_per_second": 55.345,
+      "eval_loss": 2.4508955478668213,
+      "eval_runtime": 230.4147,
+      "eval_samples_per_second": 891.913,
+      "eval_steps_per_second": 55.747,
       "step": 1912000
     },
     {
       "epoch": 7.87,
-      "learning_rate": 2.0011673476194448e-06,
-      "loss": 2.8422,
+      "learning_rate": 8.2e-08,
+      "loss": 2.5064,
       "step": 1920000
     },
     {
       "epoch": 7.87,
-      "eval_loss": 2.883701801300049,
-      "eval_runtime": 231.4957,
-      "eval_samples_per_second": 887.749,
-      "eval_steps_per_second": 55.487,
+      "eval_loss": 2.447256565093994,
+      "eval_runtime": 231.4505,
+      "eval_samples_per_second": 887.922,
+      "eval_steps_per_second": 55.498,
       "step": 1920000
     },
     {
       "epoch": 7.9,
-      "eval_loss": 2.89695143699646,
-      "eval_runtime": 230.9788,
-      "eval_samples_per_second": 889.735,
-      "eval_steps_per_second": 55.611,
+      "eval_loss": 2.457575559616089,
+      "eval_runtime": 232.2387,
+      "eval_samples_per_second": 884.908,
+      "eval_steps_per_second": 55.309,
       "step": 1928000
     },
     {
       "epoch": 7.93,
-      "learning_rate": 1.9344617693654634e-06,
-      "loss": 2.8521,
+      "learning_rate": 7.926666666666666e-08,
+      "loss": 2.5068,
       "step": 1936000
     },
     {
       "epoch": 7.93,
-      "eval_loss": 2.8805034160614014,
-      "eval_runtime": 231.7434,
-      "eval_samples_per_second": 886.8,
-      "eval_steps_per_second": 55.428,
+      "eval_loss": 2.4460949897766113,
+      "eval_runtime": 230.6448,
+      "eval_samples_per_second": 891.024,
+      "eval_steps_per_second": 55.692,
       "step": 1936000
     },
     {
       "epoch": 7.97,
-      "eval_loss": 2.8818843364715576,
-      "eval_runtime": 230.9357,
-      "eval_samples_per_second": 889.901,
-      "eval_steps_per_second": 55.622,
+      "eval_loss": 2.4451067447662354,
+      "eval_runtime": 231.6713,
+      "eval_samples_per_second": 887.076,
+      "eval_steps_per_second": 55.445,
       "step": 1944000
     },
     {
       "epoch": 8.0,
-      "learning_rate": 1.867756191111482e-06,
-      "loss": 2.8562,
+      "learning_rate": 7.653333333333333e-08,
+      "loss": 2.5152,
       "step": 1952000
     },
     {
       "epoch": 8.0,
-      "eval_loss": 2.877063751220703,
-      "eval_runtime": 230.6656,
-      "eval_samples_per_second": 890.943,
-      "eval_steps_per_second": 55.687,
+      "eval_loss": 2.442117214202881,
+      "eval_runtime": 231.4315,
+      "eval_samples_per_second": 887.995,
+      "eval_steps_per_second": 55.502,
       "step": 1952000
     },
     {
       "epoch": 8.03,
-      "eval_loss": 2.881901502609253,
-      "eval_runtime": 231.9508,
-      "eval_samples_per_second": 886.007,
-      "eval_steps_per_second": 55.378,
+      "eval_loss": 2.4458179473876953,
+      "eval_runtime": 230.6413,
+      "eval_samples_per_second": 891.037,
+      "eval_steps_per_second": 55.693,
       "step": 1960000
     },
     {
       "epoch": 8.06,
-      "learning_rate": 1.8010506128575004e-06,
-      "loss": 2.8417,
+      "learning_rate": 7.38e-08,
+      "loss": 2.5025,
       "step": 1968000
     },
     {
       "epoch": 8.06,
-      "eval_loss": 2.883180618286133,
-      "eval_runtime": 231.1061,
-      "eval_samples_per_second": 889.245,
-      "eval_steps_per_second": 55.581,
+      "eval_loss": 2.4532368183135986,
+      "eval_runtime": 230.9812,
+      "eval_samples_per_second": 889.726,
+      "eval_steps_per_second": 55.611,
       "step": 1968000
     },
     {
       "epoch": 8.1,
-      "eval_loss": 2.8927578926086426,
-      "eval_runtime": 232.292,
-      "eval_samples_per_second": 884.705,
-      "eval_steps_per_second": 55.297,
+      "eval_loss": 2.4541139602661133,
+      "eval_runtime": 231.1965,
+      "eval_samples_per_second": 888.898,
+      "eval_steps_per_second": 55.559,
       "step": 1976000
     },
     {
       "epoch": 8.13,
-      "learning_rate": 1.734345034603519e-06,
-      "loss": 2.8493,
+      "learning_rate": 7.106666666666667e-08,
+      "loss": 2.5151,
       "step": 1984000
     },
     {
       "epoch": 8.13,
-      "eval_loss": 2.8891210556030273,
-      "eval_runtime": 232.2356,
-      "eval_samples_per_second": 884.92,
-      "eval_steps_per_second": 55.31,
+      "eval_loss": 2.4499058723449707,
+      "eval_runtime": 231.2124,
+      "eval_samples_per_second": 888.836,
+      "eval_steps_per_second": 55.555,
       "step": 1984000
     },
     {
       "epoch": 8.16,
-      "eval_loss": 2.8862810134887695,
-      "eval_runtime": 233.5203,
-      "eval_samples_per_second": 880.052,
-      "eval_steps_per_second": 55.006,
+      "eval_loss": 2.4501264095306396,
+      "eval_runtime": 231.2241,
+      "eval_samples_per_second": 888.791,
+      "eval_steps_per_second": 55.552,
       "step": 1992000
     },
     {
       "epoch": 8.2,
-      "learning_rate": 1.6676394563495374e-06,
-      "loss": 2.8549,
+      "learning_rate": 6.833333333333332e-08,
+      "loss": 2.5138,
       "step": 2000000
     },
     {
       "epoch": 8.2,
-      "eval_loss": 2.876460313796997,
-      "eval_runtime": 232.4349,
-      "eval_samples_per_second": 884.162,
-      "eval_steps_per_second": 55.263,
+      "eval_loss": 2.444784641265869,
+      "eval_runtime": 231.6831,
+      "eval_samples_per_second": 887.031,
+      "eval_steps_per_second": 55.442,
       "step": 2000000
     },
     {
       "epoch": 8.23,
-      "eval_loss": 2.892146110534668,
-      "eval_runtime": 232.392,
-      "eval_samples_per_second": 884.325,
-      "eval_steps_per_second": 55.273,
+      "eval_loss": 2.4562456607818604,
+      "eval_runtime": 231.974,
+      "eval_samples_per_second": 885.918,
+      "eval_steps_per_second": 55.373,
       "step": 2008000
     },
     {
       "epoch": 8.26,
-      "learning_rate": 1.6009338780955558e-06,
-      "loss": 2.8421,
+      "learning_rate": 6.56e-08,
+      "loss": 2.5039,
       "step": 2016000
     },
     {
       "epoch": 8.26,
-      "eval_loss": 2.897282600402832,
-      "eval_runtime": 234.5662,
-      "eval_samples_per_second": 876.128,
-      "eval_steps_per_second": 54.761,
+      "eval_loss": 2.4612646102905273,
+      "eval_runtime": 234.4229,
+      "eval_samples_per_second": 876.663,
+      "eval_steps_per_second": 54.794,
       "step": 2016000
     },
     {
       "epoch": 8.29,
-      "eval_loss": 2.884657621383667,
-      "eval_runtime": 232.2059,
-      "eval_samples_per_second": 885.033,
-      "eval_steps_per_second": 55.317,
+      "eval_loss": 2.4471163749694824,
+      "eval_runtime": 233.3806,
+      "eval_samples_per_second": 880.579,
+      "eval_steps_per_second": 55.039,
       "step": 2024000
     },
     {
       "epoch": 8.33,
-      "learning_rate": 1.5342282998415744e-06,
-      "loss": 2.8451,
+      "learning_rate": 6.286666666666666e-08,
+      "loss": 2.5055,
       "step": 2032000
     },
     {
       "epoch": 8.33,
-      "eval_loss": 2.885942220687866,
-      "eval_runtime": 236.2429,
-      "eval_samples_per_second": 869.91,
-      "eval_steps_per_second": 54.372,
+      "eval_loss": 2.445026159286499,
+      "eval_runtime": 233.3418,
+      "eval_samples_per_second": 880.725,
+      "eval_steps_per_second": 55.048,
       "step": 2032000
     },
     {
       "epoch": 8.36,
-      "eval_loss": 2.8866655826568604,
-      "eval_runtime": 233.4019,
-      "eval_samples_per_second": 880.498,
-      "eval_steps_per_second": 55.034,
+      "eval_loss": 2.4492921829223633,
+      "eval_runtime": 232.3875,
+      "eval_samples_per_second": 884.342,
+      "eval_steps_per_second": 55.274,
       "step": 2040000
     },
     {
       "epoch": 8.39,
-      "learning_rate": 1.4675227215875928e-06,
-      "loss": 2.8465,
+      "learning_rate": 6.013333333333333e-08,
+      "loss": 2.5085,
       "step": 2048000
     },
     {
       "epoch": 8.39,
-      "eval_loss": 2.8853070735931396,
-      "eval_runtime": 233.5685,
-      "eval_samples_per_second": 879.87,
-      "eval_steps_per_second": 54.995,
+      "eval_loss": 2.448164224624634,
+      "eval_runtime": 233.4578,
+      "eval_samples_per_second": 880.288,
+      "eval_steps_per_second": 55.021,
       "step": 2048000
     },
     {
       "epoch": 8.42,
-      "eval_loss": 2.885338306427002,
-      "eval_runtime": 234.5164,
-      "eval_samples_per_second": 876.314,
-      "eval_steps_per_second": 54.772,
+      "eval_loss": 2.4571895599365234,
+      "eval_runtime": 235.4355,
+      "eval_samples_per_second": 872.893,
+      "eval_steps_per_second": 54.558,
       "step": 2056000
     },
     {
       "epoch": 8.46,
-      "learning_rate": 1.4008171433336116e-06,
-      "loss": 2.8516,
+      "learning_rate": 5.7400000000000004e-08,
+      "loss": 2.5114,
       "step": 2064000
     },
     {
       "epoch": 8.46,
-      "eval_loss": 2.8797264099121094,
-      "eval_runtime": 232.9203,
-      "eval_samples_per_second": 882.319,
-      "eval_steps_per_second": 55.148,
+      "eval_loss": 2.444307804107666,
+      "eval_runtime": 234.4924,
+      "eval_samples_per_second": 876.404,
+      "eval_steps_per_second": 54.778,
       "step": 2064000
     },
     {
       "epoch": 8.49,
-      "eval_loss": 2.882495880126953,
-      "eval_runtime": 232.5052,
-      "eval_samples_per_second": 883.894,
-      "eval_steps_per_second": 55.246,
+      "eval_loss": 2.445603132247925,
+      "eval_runtime": 234.6223,
+      "eval_samples_per_second": 875.919,
+      "eval_steps_per_second": 54.748,
       "step": 2072000
     },
     {
       "epoch": 8.52,
-      "learning_rate": 1.33411156507963e-06,
-      "loss": 2.8519,
+      "learning_rate": 5.4666666666666666e-08,
+      "loss": 2.5132,
       "step": 2080000
     },
     {
       "epoch": 8.52,
-      "eval_loss": 2.8863439559936523,
-      "eval_runtime": 232.936,
-      "eval_samples_per_second": 882.26,
-      "eval_steps_per_second": 55.144,
+      "eval_loss": 2.4528441429138184,
+      "eval_runtime": 234.3887,
+      "eval_samples_per_second": 876.791,
+      "eval_steps_per_second": 54.802,
       "step": 2080000
     },
     {
       "epoch": 8.56,
-      "eval_loss": 2.8823108673095703,
-      "eval_runtime": 234.3106,
-      "eval_samples_per_second": 877.084,
-      "eval_steps_per_second": 54.82,
+      "eval_loss": 2.449744939804077,
+      "eval_runtime": 233.1003,
+      "eval_samples_per_second": 881.638,
+      "eval_steps_per_second": 55.105,
       "step": 2088000
     },
     {
       "epoch": 8.59,
-      "learning_rate": 1.2674059868256484e-06,
-      "loss": 2.8454,
+      "learning_rate": 5.1933333333333335e-08,
+      "loss": 2.5072,
       "step": 2096000
     },
     {
       "epoch": 8.59,
-      "eval_loss": 2.887014865875244,
-      "eval_runtime": 236.1124,
-      "eval_samples_per_second": 870.391,
-      "eval_steps_per_second": 54.402,
+      "eval_loss": 2.4547877311706543,
+      "eval_runtime": 232.2237,
+      "eval_samples_per_second": 884.966,
+      "eval_steps_per_second": 55.313,
       "step": 2096000
     },
     {
       "epoch": 8.62,
-      "eval_loss": 2.8897511959075928,
-      "eval_runtime": 234.9051,
-      "eval_samples_per_second": 874.864,
-      "eval_steps_per_second": 54.682,
+      "eval_loss": 2.4547617435455322,
+      "eval_runtime": 232.0067,
+      "eval_samples_per_second": 885.794,
+      "eval_steps_per_second": 55.365,
       "step": 2104000
     },
     {
       "epoch": 8.65,
-      "learning_rate": 1.2007004085716668e-06,
-      "loss": 2.8428,
+      "learning_rate": 4.92e-08,
+      "loss": 2.504,
       "step": 2112000
     },
     {
       "epoch": 8.65,
-      "eval_loss": 2.8753533363342285,
-      "eval_runtime": 235.2777,
-      "eval_samples_per_second": 873.479,
-      "eval_steps_per_second": 54.595,
+      "eval_loss": 2.444261312484741,
+      "eval_runtime": 232.4079,
+      "eval_samples_per_second": 884.264,
+      "eval_steps_per_second": 55.269,
       "step": 2112000
     },
     {
       "epoch": 8.69,
-      "eval_loss": 2.8771519660949707,
-      "eval_runtime": 234.9372,
-      "eval_samples_per_second": 874.744,
-      "eval_steps_per_second": 54.674,
+      "eval_loss": 2.445204734802246,
+      "eval_runtime": 233.2645,
+      "eval_samples_per_second": 881.017,
+      "eval_steps_per_second": 55.066,
       "step": 2120000
     },
     {
       "epoch": 8.72,
-      "learning_rate": 1.1339948303176854e-06,
-      "loss": 2.85,
+      "learning_rate": 4.6466666666666666e-08,
+      "loss": 2.5128,
       "step": 2128000
     },
     {
       "epoch": 8.72,
-      "eval_loss": 2.8815581798553467,
-      "eval_runtime": 233.3269,
-      "eval_samples_per_second": 880.781,
-      "eval_steps_per_second": 55.052,
+      "eval_loss": 2.4509565830230713,
+      "eval_runtime": 233.1857,
+      "eval_samples_per_second": 881.315,
+      "eval_steps_per_second": 55.085,
       "step": 2128000
     },
     {
       "epoch": 8.75,
-      "eval_loss": 2.8723196983337402,
-      "eval_runtime": 233.3963,
-      "eval_samples_per_second": 880.52,
-      "eval_steps_per_second": 55.035,
+      "eval_loss": 2.447999954223633,
+      "eval_runtime": 233.2452,
+      "eval_samples_per_second": 881.09,
+      "eval_steps_per_second": 55.071,
       "step": 2136000
     },
     {
       "epoch": 8.79,
-      "learning_rate": 1.067289252063704e-06,
-      "loss": 2.8482,
+      "learning_rate": 4.3733333333333335e-08,
+      "loss": 2.5133,
       "step": 2144000
     },
     {
       "epoch": 8.79,
-      "eval_loss": 2.8833649158477783,
-      "eval_runtime": 233.2426,
-      "eval_samples_per_second": 881.1,
-      "eval_steps_per_second": 55.071,
+      "eval_loss": 2.4470479488372803,
+      "eval_runtime": 234.9529,
+      "eval_samples_per_second": 874.686,
+      "eval_steps_per_second": 54.671,
       "step": 2144000
     },
     {
       "epoch": 8.82,
-      "eval_loss": 2.8784103393554688,
-      "eval_runtime": 232.8953,
-      "eval_samples_per_second": 882.414,
-      "eval_steps_per_second": 55.154,
+      "eval_loss": 2.4436631202697754,
+      "eval_runtime": 234.9836,
+      "eval_samples_per_second": 874.572,
+      "eval_steps_per_second": 54.663,
       "step": 2152000
     },
     {
       "epoch": 8.85,
-      "learning_rate": 1.0005836738097224e-06,
-      "loss": 2.8417,
+      "learning_rate": 4.1e-08,
+      "loss": 2.5067,
       "step": 2160000
     },
     {
       "epoch": 8.85,
-      "eval_loss": 2.8758938312530518,
-      "eval_runtime": 233.0635,
-      "eval_samples_per_second": 881.777,
-      "eval_steps_per_second": 55.114,
+      "eval_loss": 2.444672107696533,
+      "eval_runtime": 234.3233,
+      "eval_samples_per_second": 877.036,
+      "eval_steps_per_second": 54.817,
       "step": 2160000
     },
     {
       "epoch": 8.88,
-      "eval_loss": 2.881685256958008,
-      "eval_runtime": 232.8737,
-      "eval_samples_per_second": 882.495,
-      "eval_steps_per_second": 55.159,
+      "eval_loss": 2.453118085861206,
+      "eval_runtime": 233.5384,
+      "eval_samples_per_second": 879.984,
+      "eval_steps_per_second": 55.002,
       "step": 2168000
     },
     {
       "epoch": 8.92,
-      "learning_rate": 9.33878095555741e-07,
-      "loss": 2.8336,
+      "learning_rate": 3.8266666666666665e-08,
+      "loss": 2.4996,
       "step": 2176000
     },
     {
       "epoch": 8.92,
-      "eval_loss": 2.8810818195343018,
-      "eval_runtime": 233.1976,
-      "eval_samples_per_second": 881.27,
-      "eval_steps_per_second": 55.082,
+      "eval_loss": 2.447479009628296,
+      "eval_runtime": 235.7844,
+      "eval_samples_per_second": 871.601,
+      "eval_steps_per_second": 54.478,
       "step": 2176000
     },
     {
       "epoch": 8.95,
-      "eval_loss": 2.8726937770843506,
-      "eval_runtime": 232.1542,
-      "eval_samples_per_second": 885.231,
-      "eval_steps_per_second": 55.33,
+      "eval_loss": 2.4438347816467285,
+      "eval_runtime": 233.6193,
+      "eval_samples_per_second": 879.679,
+      "eval_steps_per_second": 54.983,
       "step": 2184000
     },
     {
       "epoch": 8.98,
-      "learning_rate": 8.671725173017595e-07,
-      "loss": 2.8514,
+      "learning_rate": 3.5533333333333334e-08,
+      "loss": 2.5123,
       "step": 2192000
     },
     {
       "epoch": 8.98,
-      "eval_loss": 2.889397621154785,
-      "eval_runtime": 232.9383,
-      "eval_samples_per_second": 882.251,
-      "eval_steps_per_second": 55.143,
+      "eval_loss": 2.4552195072174072,
+      "eval_runtime": 235.201,
+      "eval_samples_per_second": 873.763,
+      "eval_steps_per_second": 54.613,
       "step": 2192000
     },
     {
       "epoch": 9.01,
-      "eval_loss": 2.8750553131103516,
-      "eval_runtime": 234.1088,
-      "eval_samples_per_second": 877.84,
-      "eval_steps_per_second": 54.868,
+      "eval_loss": 2.4441311359405518,
+      "eval_runtime": 234.6948,
+      "eval_samples_per_second": 875.648,
+      "eval_steps_per_second": 54.731,
       "step": 2200000
     },
     {
       "epoch": 9.05,
-      "learning_rate": 8.004669390477779e-07,
-      "loss": 2.8312,
+      "learning_rate": 3.28e-08,
+      "loss": 2.5044,
       "step": 2208000
     },
     {
       "epoch": 9.05,
-      "eval_loss": 2.877980947494507,
-      "eval_runtime": 234.0757,
-      "eval_samples_per_second": 877.964,
-      "eval_steps_per_second": 54.875,
+      "eval_loss": 2.4438366889953613,
+      "eval_runtime": 233.1145,
+      "eval_samples_per_second": 881.584,
+      "eval_steps_per_second": 55.102,
       "step": 2208000
     },
     {
       "epoch": 9.08,
-      "eval_loss": 2.8862829208374023,
-      "eval_runtime": 232.0198,
-      "eval_samples_per_second": 885.743,
-      "eval_steps_per_second": 55.362,
+      "eval_loss": 2.453371286392212,
+      "eval_runtime": 234.9783,
+      "eval_samples_per_second": 874.592,
+      "eval_steps_per_second": 54.665,
       "step": 2216000
     },
     {
       "epoch": 9.11,
-      "learning_rate": 7.337613607937964e-07,
-      "loss": 2.8315,
+      "learning_rate": 3.0066666666666665e-08,
+      "loss": 2.5068,
       "step": 2224000
     },
     {
       "epoch": 9.11,
-      "eval_loss": 2.881166696548462,
-      "eval_runtime": 232.6004,
-      "eval_samples_per_second": 883.533,
-      "eval_steps_per_second": 55.223,
+      "eval_loss": 2.449671745300293,
+      "eval_runtime": 232.7881,
+      "eval_samples_per_second": 882.82,
+      "eval_steps_per_second": 55.179,
       "step": 2224000
     },
     {
       "epoch": 9.15,
-      "eval_loss": 2.871530532836914,
-      "eval_runtime": 233.3854,
-      "eval_samples_per_second": 880.561,
-      "eval_steps_per_second": 55.038,
+      "eval_loss": 2.444044828414917,
+      "eval_runtime": 233.2255,
+      "eval_samples_per_second": 881.164,
+      "eval_steps_per_second": 55.075,
       "step": 2232000
     },
     {
       "epoch": 9.18,
-      "learning_rate": 6.67055782539815e-07,
-      "loss": 2.8509,
+      "learning_rate": 2.7333333333333333e-08,
+      "loss": 2.5165,
       "step": 2240000
     },
     {
       "epoch": 9.18,
-      "eval_loss": 2.890825033187866,
-      "eval_runtime": 233.6047,
-      "eval_samples_per_second": 879.734,
-      "eval_steps_per_second": 54.986,
+      "eval_loss": 2.457695722579956,
+      "eval_runtime": 234.2621,
+      "eval_samples_per_second": 877.265,
+      "eval_steps_per_second": 54.832,
       "step": 2240000
     },
     {
       "epoch": 9.21,
-      "eval_loss": 2.880777359008789,
-      "eval_runtime": 232.8121,
-      "eval_samples_per_second": 882.729,
-      "eval_steps_per_second": 55.173,
+      "eval_loss": 2.4506990909576416,
+      "eval_runtime": 232.4001,
+      "eval_samples_per_second": 884.294,
+      "eval_steps_per_second": 55.271,
       "step": 2248000
     },
     {
       "epoch": 9.24,
-      "learning_rate": 6.003502042858334e-07,
-      "loss": 2.8394,
+      "learning_rate": 2.46e-08,
+      "loss": 2.5087,
       "step": 2256000
     },
     {
       "epoch": 9.24,
-      "eval_loss": 2.8801651000976562,
-      "eval_runtime": 235.058,
-      "eval_samples_per_second": 874.295,
-      "eval_steps_per_second": 54.646,
+      "eval_loss": 2.4494166374206543,
+      "eval_runtime": 233.0606,
+      "eval_samples_per_second": 881.788,
+      "eval_steps_per_second": 55.114,
       "step": 2256000
     },
     {
       "epoch": 9.28,
-      "eval_loss": 2.8692407608032227,
-      "eval_runtime": 234.8843,
-      "eval_samples_per_second": 874.941,
-      "eval_steps_per_second": 54.686,
+      "eval_loss": 2.4393150806427,
+      "eval_runtime": 234.1923,
+      "eval_samples_per_second": 877.527,
+      "eval_steps_per_second": 54.848,
       "step": 2264000
     },
     {
       "epoch": 9.31,
-      "learning_rate": 5.33644626031852e-07,
-      "loss": 2.8332,
+      "learning_rate": 2.1866666666666667e-08,
+      "loss": 2.5036,
       "step": 2272000
     },
     {
       "epoch": 9.31,
-      "eval_loss": 2.871225357055664,
-      "eval_runtime": 233.9349,
-      "eval_samples_per_second": 878.492,
-      "eval_steps_per_second": 54.908,
+      "eval_loss": 2.4486756324768066,
+      "eval_runtime": 233.5876,
+      "eval_samples_per_second": 879.798,
+      "eval_steps_per_second": 54.99,
       "step": 2272000
     },
     {
       "epoch": 9.34,
-      "eval_loss": 2.8688416481018066,
-      "eval_runtime": 233.3243,
-      "eval_samples_per_second": 880.791,
-      "eval_steps_per_second": 55.052,
+      "eval_loss": 2.442298173904419,
+      "eval_runtime": 233.1053,
+      "eval_samples_per_second": 881.619,
+      "eval_steps_per_second": 55.104,
       "step": 2280000
     },
     {
       "epoch": 9.38,
-      "learning_rate": 4.669390477778705e-07,
-      "loss": 2.837,
+      "learning_rate": 1.9133333333333333e-08,
+      "loss": 2.5086,
       "step": 2288000
     },
     {
       "epoch": 9.38,
-      "eval_loss": 2.8778717517852783,
-      "eval_runtime": 233.117,
-      "eval_samples_per_second": 881.574,
-      "eval_steps_per_second": 55.101,
+      "eval_loss": 2.4455623626708984,
+      "eval_runtime": 232.7856,
+      "eval_samples_per_second": 882.83,
+      "eval_steps_per_second": 55.18,
       "step": 2288000
     },
     {
       "epoch": 9.41,
-      "eval_loss": 2.879408597946167,
-      "eval_runtime": 233.2294,
-      "eval_samples_per_second": 881.149,
-      "eval_steps_per_second": 55.075,
+      "eval_loss": 2.449575185775757,
+      "eval_runtime": 234.5471,
+      "eval_samples_per_second": 876.199,
+      "eval_steps_per_second": 54.765,
       "step": 2296000
     },
     {
       "epoch": 9.44,
-      "learning_rate": 4.0023346952388894e-07,
-      "loss": 2.8344,
+      "learning_rate": 1.64e-08,
+      "loss": 2.5034,
       "step": 2304000
     },
     {
       "epoch": 9.44,
-      "eval_loss": 2.8751115798950195,
-      "eval_runtime": 233.3516,
-      "eval_samples_per_second": 880.688,
-      "eval_steps_per_second": 55.046,
+      "eval_loss": 2.4498891830444336,
+      "eval_runtime": 232.7935,
+      "eval_samples_per_second": 882.8,
+      "eval_steps_per_second": 55.178,
       "step": 2304000
     },
     {
       "epoch": 9.47,
-      "eval_loss": 2.874969482421875,
-      "eval_runtime": 233.1427,
-      "eval_samples_per_second": 881.477,
-      "eval_steps_per_second": 55.095,
+      "eval_loss": 2.4432790279388428,
+      "eval_runtime": 233.6332,
+      "eval_samples_per_second": 879.627,
+      "eval_steps_per_second": 54.979,
       "step": 2312000
     },
     {
       "epoch": 9.51,
-      "learning_rate": 3.335278912699075e-07,
-      "loss": 2.8376,
+      "learning_rate": 1.3666666666666667e-08,
+      "loss": 2.5099,
       "step": 2320000
     },
     {
       "epoch": 9.51,
-      "eval_loss": 2.8837876319885254,
-      "eval_runtime": 233.3471,
-      "eval_samples_per_second": 880.705,
-      "eval_steps_per_second": 55.047,
+      "eval_loss": 2.4534084796905518,
+      "eval_runtime": 233.6601,
+      "eval_samples_per_second": 879.525,
+      "eval_steps_per_second": 54.973,
       "step": 2320000
     },
     {
       "epoch": 9.54,
-      "eval_loss": 2.8825130462646484,
-      "eval_runtime": 233.8169,
-      "eval_samples_per_second": 878.935,
-      "eval_steps_per_second": 54.936,
+      "eval_loss": 2.4494857788085938,
+      "eval_runtime": 233.4959,
+      "eval_samples_per_second": 880.144,
+      "eval_steps_per_second": 55.012,
       "step": 2328000
     },
     {
       "epoch": 9.57,
-      "learning_rate": 2.66822313015926e-07,
-      "loss": 2.8329,
+      "learning_rate": 1.0933333333333334e-08,
+      "loss": 2.5065,
       "step": 2336000
     },
     {
       "epoch": 9.57,
-      "eval_loss": 2.880941390991211,
-      "eval_runtime": 234.5491,
-      "eval_samples_per_second": 876.192,
-      "eval_steps_per_second": 54.765,
+      "eval_loss": 2.4510202407836914,
+      "eval_runtime": 233.7041,
+      "eval_samples_per_second": 879.36,
+      "eval_steps_per_second": 54.963,
       "step": 2336000
     },
     {
       "epoch": 9.6,
-      "eval_loss": 2.8842968940734863,
-      "eval_runtime": 233.344,
-      "eval_samples_per_second": 880.717,
-      "eval_steps_per_second": 55.047,
+      "eval_loss": 2.4512877464294434,
+      "eval_runtime": 237.5888,
+      "eval_samples_per_second": 864.982,
+      "eval_steps_per_second": 54.064,
       "step": 2344000
     },
     {
       "epoch": 9.64,
-      "learning_rate": 2.0011673476194447e-07,
-      "loss": 2.8264,
+      "learning_rate": 8.2e-09,
+      "loss": 2.502,
       "step": 2352000
     },
     {
       "epoch": 9.64,
-      "eval_loss": 2.8784091472625732,
-      "eval_runtime": 233.7157,
-      "eval_samples_per_second": 879.316,
-      "eval_steps_per_second": 54.96,
+      "eval_loss": 2.451225996017456,
+      "eval_runtime": 233.4207,
+      "eval_samples_per_second": 880.427,
+      "eval_steps_per_second": 55.029,
       "step": 2352000
     },
     {
       "epoch": 9.67,
-      "eval_loss": 2.8687503337860107,
-      "eval_runtime": 233.7026,
-      "eval_samples_per_second": 879.365,
-      "eval_steps_per_second": 54.963,
+      "eval_loss": 2.4469268321990967,
+      "eval_runtime": 234.7938,
+      "eval_samples_per_second": 875.279,
+      "eval_steps_per_second": 54.708,
       "step": 2360000
     },
     {
       "epoch": 9.7,
-      "learning_rate": 1.33411156507963e-07,
-      "loss": 2.8317,
+      "learning_rate": 5.466666666666667e-09,
+      "loss": 2.5043,
       "step": 2368000
     },
     {
       "epoch": 9.7,
-      "eval_loss": 2.8792617321014404,
-      "eval_runtime": 235.256,
-      "eval_samples_per_second": 873.559,
-      "eval_steps_per_second": 54.6,
+      "eval_loss": 2.4544479846954346,
+      "eval_runtime": 234.2869,
+      "eval_samples_per_second": 877.173,
+      "eval_steps_per_second": 54.826,
       "step": 2368000
     },
     {
       "epoch": 9.74,
-      "eval_loss": 2.8814611434936523,
-      "eval_runtime": 235.5823,
-      "eval_samples_per_second": 872.349,
-      "eval_steps_per_second": 54.524,
+      "eval_loss": 2.4492740631103516,
+      "eval_runtime": 234.1805,
+      "eval_samples_per_second": 877.571,
+      "eval_steps_per_second": 54.851,
       "step": 2376000
     },
     {
       "epoch": 9.77,
-      "learning_rate": 6.67055782539815e-08,
-      "loss": 2.8328,
+      "learning_rate": 2.7333333333333334e-09,
+      "loss": 2.5068,
       "step": 2384000
     },
     {
       "epoch": 9.77,
-      "eval_loss": 2.8755900859832764,
-      "eval_runtime": 233.6052,
-      "eval_samples_per_second": 879.732,
-      "eval_steps_per_second": 54.986,
+      "eval_loss": 2.453711748123169,
+      "eval_runtime": 233.0608,
+      "eval_samples_per_second": 881.787,
+      "eval_steps_per_second": 55.114,
       "step": 2384000
     },
     {
       "epoch": 9.8,
-      "eval_loss": 2.869050979614258,
-      "eval_runtime": 233.5136,
-      "eval_samples_per_second": 880.077,
-      "eval_steps_per_second": 55.007,
+      "eval_loss": 2.4386837482452393,
+      "eval_runtime": 234.1662,
+      "eval_samples_per_second": 877.625,
+      "eval_steps_per_second": 54.854,
       "step": 2392000
     },
     {
       "epoch": 9.83,
       "learning_rate": 0.0,
-      "loss": 2.841,
+      "loss": 2.5118,
       "step": 2400000
     },
     {
       "epoch": 9.83,
-      "eval_loss": 2.867439031600952,
-      "eval_runtime": 233.911,
-      "eval_samples_per_second": 878.582,
-      "eval_steps_per_second": 54.914,
+      "eval_loss": 2.4494030475616455,
+      "eval_runtime": 234.0187,
+      "eval_samples_per_second": 878.178,
+      "eval_steps_per_second": 54.889,
       "step": 2400000
     },
     {
       "epoch": 9.83,
       "step": 2400000,
       "total_flos": 7.305293129309786e+17,
-      "train_loss": 2.773840421549479,
-      "train_runtime": 219744.0708,
-      "train_samples_per_second": 174.749,
-      "train_steps_per_second": 10.922
+      "train_loss": 2.5438934391276042,
+      "train_runtime": 220778.1092,
+      "train_samples_per_second": 173.93,
+      "train_steps_per_second": 10.871
     }
   ],
   "logging_steps": 16000,