diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20822 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.970873786407767,
+  "global_step": 80000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0863903706529285e-10,
+      "loss": 2.0937,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.431951853264643e-09,
+      "loss": 2.0753,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 13.1146,
+      "eval_samples_per_second": 381.253,
+      "eval_steps_per_second": 0.763,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0863903706529285e-08,
+      "loss": 2.0633,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.8758,
+      "eval_samples_per_second": 388.324,
+      "eval_steps_per_second": 0.777,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.629585555979393e-08,
+      "loss": 2.0687,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.9552,
+      "eval_samples_per_second": 385.945,
+      "eval_steps_per_second": 0.772,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.172780741305857e-08,
+      "loss": 2.0719,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.6634,
+      "eval_samples_per_second": 394.839,
+      "eval_steps_per_second": 0.79,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7159759266323215e-08,
+      "loss": 2.0687,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.9371,
+      "eval_samples_per_second": 386.484,
+      "eval_steps_per_second": 0.773,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.259171111958786e-08,
+      "loss": 2.0745,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.7773,
+      "eval_samples_per_second": 391.32,
+      "eval_steps_per_second": 0.783,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.80236629728525e-08,
+      "loss": 2.0643,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.8646,
+      "eval_samples_per_second": 388.662,
+      "eval_steps_per_second": 0.777,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.345561482611714e-08,
+      "loss": 2.0673,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "eval_runtime": 12.7924,
+      "eval_samples_per_second": 390.858,
+      "eval_steps_per_second": 0.782,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.888756667938178e-08,
+      "loss": 2.0688,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.8652,
+      "eval_samples_per_second": 388.647,
+      "eval_steps_per_second": 0.777,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.431951853264643e-08,
+      "loss": 2.0696,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.4944,
+      "eval_samples_per_second": 400.179,
+      "eval_steps_per_second": 0.8,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.975147038591107e-08,
+      "loss": 2.0682,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.7396,
+      "eval_samples_per_second": 392.478,
+      "eval_steps_per_second": 0.785,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.518342223917572e-08,
+      "loss": 2.0657,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.8663,
+      "eval_samples_per_second": 388.613,
+      "eval_steps_per_second": 0.777,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.061537409244035e-08,
+      "loss": 2.0662,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.8854,
+      "eval_samples_per_second": 388.036,
+      "eval_steps_per_second": 0.776,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6047325945705e-08,
+      "loss": 2.062,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.8237,
+      "eval_samples_per_second": 389.904,
+      "eval_steps_per_second": 0.78,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.147927779896965e-08,
+      "loss": 2.0636,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.8137,
+      "eval_samples_per_second": 390.207,
+      "eval_steps_per_second": 0.78,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.691122965223428e-08,
+      "loss": 2.0665,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.7934,
+      "eval_samples_per_second": 390.827,
+      "eval_steps_per_second": 0.782,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.234318150549893e-08,
+      "loss": 2.0626,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.9429,
+      "eval_samples_per_second": 386.312,
+      "eval_steps_per_second": 0.773,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.777513335876356e-08,
+      "loss": 2.0616,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.9661,
+      "eval_samples_per_second": 385.621,
+      "eval_steps_per_second": 0.771,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0320708521202821e-07,
+      "loss": 2.0603,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.9465,
+      "eval_samples_per_second": 386.206,
+      "eval_steps_per_second": 0.772,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0863903706529286e-07,
+      "loss": 2.0595,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.8367,
+      "eval_samples_per_second": 389.509,
+      "eval_steps_per_second": 0.779,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.140709889185575e-07,
+      "loss": 2.0612,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.5993,
+      "eval_samples_per_second": 396.848,
+      "eval_steps_per_second": 0.794,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1950294077182214e-07,
+      "loss": 2.0527,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.9757,
+      "eval_samples_per_second": 385.335,
+      "eval_steps_per_second": 0.771,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.249348926250868e-07,
+      "loss": 2.058,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.7369,
+      "eval_samples_per_second": 392.561,
+      "eval_steps_per_second": 0.785,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3036684447835144e-07,
+      "loss": 2.0572,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "eval_runtime": 12.9968,
+      "eval_samples_per_second": 384.709,
+      "eval_steps_per_second": 0.769,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3579879633161608e-07,
+      "loss": 2.059,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.9405,
+      "eval_samples_per_second": 386.383,
+      "eval_steps_per_second": 0.773,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.412307481848807e-07,
+      "loss": 2.0582,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.6726,
+      "eval_samples_per_second": 394.551,
+      "eval_steps_per_second": 0.789,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4666270003814535e-07,
+      "loss": 2.0542,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.9181,
+      "eval_samples_per_second": 387.054,
+      "eval_steps_per_second": 0.774,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5209465189141e-07,
+      "loss": 2.0473,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8785,
+      "eval_samples_per_second": 388.245,
+      "eval_steps_per_second": 0.776,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5752660374467465e-07,
+      "loss": 2.0479,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8183,
+      "eval_samples_per_second": 390.066,
+      "eval_steps_per_second": 0.78,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.629585555979393e-07,
+      "loss": 2.0579,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.5964,
+      "eval_samples_per_second": 396.94,
+      "eval_steps_per_second": 0.794,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6839050745120392e-07,
+      "loss": 2.0526,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8734,
+      "eval_samples_per_second": 388.398,
+      "eval_steps_per_second": 0.777,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7382245930446856e-07,
+      "loss": 2.0578,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.6763,
+      "eval_samples_per_second": 394.436,
+      "eval_steps_per_second": 0.789,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.792544111577332e-07,
+      "loss": 2.0498,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.4737,
+      "eval_samples_per_second": 400.845,
+      "eval_steps_per_second": 0.802,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8468636301099786e-07,
+      "loss": 2.0483,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.5369,
+      "eval_samples_per_second": 398.823,
+      "eval_steps_per_second": 0.798,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.901183148642625e-07,
+      "loss": 2.0512,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8116,
+      "eval_samples_per_second": 390.272,
+      "eval_steps_per_second": 0.781,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9555026671752713e-07,
+      "loss": 2.0449,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8638,
+      "eval_samples_per_second": 388.688,
+      "eval_steps_per_second": 0.777,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.009822185707918e-07,
+      "loss": 2.0507,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.6794,
+      "eval_samples_per_second": 394.339,
+      "eval_steps_per_second": 0.789,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0641417042405642e-07,
+      "loss": 2.0448,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8009,
+      "eval_samples_per_second": 390.597,
+      "eval_steps_per_second": 0.781,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.118461222773211e-07,
+      "loss": 2.0463,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 13.0074,
+      "eval_samples_per_second": 384.395,
+      "eval_steps_per_second": 0.769,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1727807413058572e-07,
+      "loss": 2.0498,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8075,
+      "eval_samples_per_second": 390.397,
+      "eval_steps_per_second": 0.781,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2271002598385034e-07,
+      "loss": 2.0399,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "eval_runtime": 12.8001,
+      "eval_samples_per_second": 390.623,
+      "eval_steps_per_second": 0.781,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.28141977837115e-07,
+      "loss": 2.0392,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.8138,
+      "eval_samples_per_second": 390.203,
+      "eval_steps_per_second": 0.78,
+      "step": 2100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.3357392969037963e-07,
+      "loss": 2.032,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.5356,
+      "eval_samples_per_second": 398.865,
+      "eval_steps_per_second": 0.798,
+      "step": 2150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.390058815436443e-07,
+      "loss": 2.0471,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.8145,
+      "eval_samples_per_second": 390.182,
+      "eval_steps_per_second": 0.78,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.444378333969089e-07,
+      "loss": 2.0342,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.6096,
+      "eval_samples_per_second": 396.523,
+      "eval_steps_per_second": 0.793,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.498697852501736e-07,
+      "loss": 2.0376,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.823,
+      "eval_samples_per_second": 389.923,
+      "eval_steps_per_second": 0.78,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.553017371034382e-07,
+      "loss": 2.0293,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.7771,
+      "eval_samples_per_second": 391.325,
+      "eval_steps_per_second": 0.783,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6073368895670287e-07,
+      "loss": 2.0309,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.6592,
+      "eval_samples_per_second": 394.97,
+      "eval_steps_per_second": 0.79,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.661656408099675e-07,
+      "loss": 2.0371,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.4525,
+      "eval_samples_per_second": 401.526,
+      "eval_steps_per_second": 0.803,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7159759266323217e-07,
+      "loss": 2.0298,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.8201,
+      "eval_samples_per_second": 390.014,
+      "eval_steps_per_second": 0.78,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.770295445164968e-07,
+      "loss": 2.0326,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.693,
+      "eval_samples_per_second": 393.918,
+      "eval_steps_per_second": 0.788,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.824614963697614e-07,
+      "loss": 2.0285,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.6954,
+      "eval_samples_per_second": 393.843,
+      "eval_steps_per_second": 0.788,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.878934482230261e-07,
+      "loss": 2.0262,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.9737,
+      "eval_samples_per_second": 385.396,
+      "eval_steps_per_second": 0.771,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.933254000762907e-07,
+      "loss": 2.0201,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.7938,
+      "eval_samples_per_second": 390.815,
+      "eval_steps_per_second": 0.782,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.987573519295554e-07,
+      "loss": 2.0215,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.4133,
+      "eval_samples_per_second": 402.794,
+      "eval_steps_per_second": 0.806,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0418930378282e-07,
+      "loss": 2.0214,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.9042,
+      "eval_samples_per_second": 387.472,
+      "eval_steps_per_second": 0.775,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.096212556360847e-07,
+      "loss": 2.0268,
+      "step": 2850
+    },
+    {
+      "epoch": 0.03,
+      "eval_runtime": 12.8414,
+      "eval_samples_per_second": 389.367,
+      "eval_steps_per_second": 0.779,
+      "step": 2850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.150532074893493e-07,
+      "loss": 2.0098,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.8198,
+      "eval_samples_per_second": 390.021,
+      "eval_steps_per_second": 0.78,
+      "step": 2900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.204851593426139e-07,
+      "loss": 2.0138,
+      "step": 2950
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.5135,
+      "eval_samples_per_second": 399.569,
+      "eval_steps_per_second": 0.799,
+      "step": 2950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.259171111958786e-07,
+      "loss": 2.0223,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.4443,
+      "eval_samples_per_second": 401.79,
+      "eval_steps_per_second": 0.804,
+      "step": 3000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.313490630491432e-07,
+      "loss": 2.0097,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.826,
+      "eval_samples_per_second": 389.834,
+      "eval_steps_per_second": 0.78,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3678101490240783e-07,
+      "loss": 2.0145,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.4887,
+      "eval_samples_per_second": 400.362,
+      "eval_steps_per_second": 0.801,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4221296675567245e-07,
+      "loss": 2.0065,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.8072,
+      "eval_samples_per_second": 390.404,
+      "eval_steps_per_second": 0.781,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4764491860893713e-07,
+      "loss": 2.0105,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.4497,
+      "eval_samples_per_second": 401.618,
+      "eval_steps_per_second": 0.803,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5307687046220175e-07,
+      "loss": 2.0009,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.6583,
+      "eval_samples_per_second": 394.996,
+      "eval_steps_per_second": 0.79,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.585088223154664e-07,
+      "loss": 2.001,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.9553,
+      "eval_samples_per_second": 385.943,
+      "eval_steps_per_second": 0.772,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6394077416873104e-07,
+      "loss": 2.0015,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.8551,
+      "eval_samples_per_second": 388.951,
+      "eval_steps_per_second": 0.778,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.693727260219957e-07,
+      "loss": 2.0019,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.5288,
+      "eval_samples_per_second": 399.08,
+      "eval_steps_per_second": 0.798,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7480467787526034e-07,
+      "loss": 1.993,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.9187,
+      "eval_samples_per_second": 387.035,
+      "eval_steps_per_second": 0.774,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.80236629728525e-07,
+      "loss": 1.9987,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.4572,
+      "eval_samples_per_second": 401.374,
+      "eval_steps_per_second": 0.803,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8566858158178964e-07,
+      "loss": 2.0014,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.8266,
+      "eval_samples_per_second": 389.815,
+      "eval_steps_per_second": 0.78,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9110053343505426e-07,
+      "loss": 1.9944,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.7885,
+      "eval_samples_per_second": 390.977,
+      "eval_steps_per_second": 0.782,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.965324852883189e-07,
+      "loss": 1.9897,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.5305,
+      "eval_samples_per_second": 399.028,
+      "eval_steps_per_second": 0.798,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.019644371415836e-07,
+      "loss": 1.9898,
+      "step": 3700
+    },
+    {
+      "epoch": 0.04,
+      "eval_runtime": 12.7895,
+      "eval_samples_per_second": 390.944,
+      "eval_steps_per_second": 0.782,
+      "step": 3700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.073963889948482e-07,
+      "loss": 1.985,
+      "step": 3750
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.5029,
+      "eval_samples_per_second": 399.908,
+      "eval_steps_per_second": 0.8,
+      "step": 3750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1282834084811285e-07,
+      "loss": 1.9835,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.9241,
+      "eval_samples_per_second": 386.874,
+      "eval_steps_per_second": 0.774,
+      "step": 3800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1826029270137747e-07,
+      "loss": 1.9931,
+      "step": 3850
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.8566,
+      "eval_samples_per_second": 388.905,
+      "eval_steps_per_second": 0.778,
+      "step": 3850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.236922445546422e-07,
+      "loss": 1.9855,
+      "step": 3900
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.7181,
+      "eval_samples_per_second": 393.139,
+      "eval_steps_per_second": 0.786,
+      "step": 3900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.291241964079068e-07,
+      "loss": 1.982,
+      "step": 3950
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.9707,
+      "eval_samples_per_second": 385.484,
+      "eval_steps_per_second": 0.771,
+      "step": 3950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3455614826117144e-07,
+      "loss": 1.9744,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.9989,
+      "eval_samples_per_second": 384.648,
+      "eval_steps_per_second": 0.769,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3998810011443606e-07,
+      "loss": 1.9752,
+      "step": 4050
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.6837,
+      "eval_samples_per_second": 394.206,
+      "eval_steps_per_second": 0.788,
+      "step": 4050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.454200519677007e-07,
+      "loss": 1.9682,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.6586,
+      "eval_samples_per_second": 394.988,
+      "eval_steps_per_second": 0.79,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5085200382096535e-07,
+      "loss": 1.9711,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 13.0304,
+      "eval_samples_per_second": 383.717,
+      "eval_steps_per_second": 0.767,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5628395567423e-07,
+      "loss": 1.9643,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.8061,
+      "eval_samples_per_second": 390.438,
+      "eval_steps_per_second": 0.781,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6171590752749465e-07,
+      "loss": 1.9699,
+      "step": 4250
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.5304,
+      "eval_samples_per_second": 399.03,
+      "eval_steps_per_second": 0.798,
+      "step": 4250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6714785938075927e-07,
+      "loss": 1.9598,
+      "step": 4300
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.9546,
+      "eval_samples_per_second": 385.962,
+      "eval_steps_per_second": 0.772,
+      "step": 4300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7257981123402394e-07,
+      "loss": 1.9613,
+      "step": 4350
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.8121,
+      "eval_samples_per_second": 390.257,
+      "eval_steps_per_second": 0.781,
+      "step": 4350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.780117630872886e-07,
+      "loss": 1.963,
+      "step": 4400
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.608,
+      "eval_samples_per_second": 396.575,
+      "eval_steps_per_second": 0.793,
+      "step": 4400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.834437149405532e-07,
+      "loss": 1.9577,
+      "step": 4450
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.4736,
+      "eval_samples_per_second": 400.846,
+      "eval_steps_per_second": 0.802,
+      "step": 4450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.888756667938178e-07,
+      "loss": 1.9561,
+      "step": 4500
+    },
+    {
+      "epoch": 0.05,
+      "eval_runtime": 12.5079,
+      "eval_samples_per_second": 399.748,
+      "eval_steps_per_second": 0.799,
+      "step": 4500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.943076186470824e-07,
+      "loss": 1.9556,
+      "step": 4550
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8026,
+      "eval_samples_per_second": 390.546,
+      "eval_steps_per_second": 0.781,
+      "step": 4550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.997395705003472e-07,
+      "loss": 1.9521,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8801,
+      "eval_samples_per_second": 388.194,
+      "eval_steps_per_second": 0.776,
+      "step": 4600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.051715223536118e-07,
+      "loss": 1.9459,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8116,
+      "eval_samples_per_second": 390.272,
+      "eval_steps_per_second": 0.781,
+      "step": 4650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.106034742068764e-07,
+      "loss": 1.9421,
+      "step": 4700
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.7543,
+      "eval_samples_per_second": 392.023,
+      "eval_steps_per_second": 0.784,
+      "step": 4700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.16035426060141e-07,
+      "loss": 1.9474,
+      "step": 4750
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.7756,
+      "eval_samples_per_second": 391.37,
+      "eval_steps_per_second": 0.783,
+      "step": 4750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.214673779134057e-07,
+      "loss": 1.9384,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.9102,
+      "eval_samples_per_second": 387.29,
+      "eval_steps_per_second": 0.775,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.268993297666704e-07,
+      "loss": 1.9391,
+      "step": 4850
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8433,
+      "eval_samples_per_second": 389.307,
+      "eval_steps_per_second": 0.779,
+      "step": 4850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.32331281619935e-07,
+      "loss": 1.9383,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.9188,
+      "eval_samples_per_second": 387.033,
+      "eval_steps_per_second": 0.774,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.377632334731996e-07,
+      "loss": 1.9413,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 13.0356,
+      "eval_samples_per_second": 383.564,
+      "eval_steps_per_second": 0.767,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.431951853264643e-07,
+      "loss": 1.9345,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8803,
+      "eval_samples_per_second": 388.191,
+      "eval_steps_per_second": 0.776,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.48627137179729e-07,
+      "loss": 1.934,
+      "step": 5050
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.6092,
+      "eval_samples_per_second": 396.537,
+      "eval_steps_per_second": 0.793,
+      "step": 5050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.540590890329936e-07,
+      "loss": 1.926,
+      "step": 5100
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.9586,
+      "eval_samples_per_second": 385.845,
+      "eval_steps_per_second": 0.772,
+      "step": 5100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.594910408862582e-07,
+      "loss": 1.9242,
+      "step": 5150
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8364,
+      "eval_samples_per_second": 389.517,
+      "eval_steps_per_second": 0.779,
+      "step": 5150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.649229927395228e-07,
+      "loss": 1.9221,
+      "step": 5200
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.4889,
+      "eval_samples_per_second": 400.357,
+      "eval_steps_per_second": 0.801,
+      "step": 5200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.703549445927875e-07,
+      "loss": 1.9295,
+      "step": 5250
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.8361,
+      "eval_samples_per_second": 389.526,
+      "eval_steps_per_second": 0.779,
+      "step": 5250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.757868964460522e-07,
+      "loss": 1.9235,
+      "step": 5300
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.6992,
+      "eval_samples_per_second": 393.725,
+      "eval_steps_per_second": 0.787,
+      "step": 5300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.812188482993168e-07,
+      "loss": 1.9224,
+      "step": 5350
+    },
+    {
+      "epoch": 0.06,
+      "eval_runtime": 12.9042,
+      "eval_samples_per_second": 387.47,
+      "eval_steps_per_second": 0.775,
+      "step": 5350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.866508001525814e-07,
+      "loss": 1.9212,
+      "step": 5400
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.9229,
+      "eval_samples_per_second": 386.91,
+      "eval_steps_per_second": 0.774,
+      "step": 5400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.920827520058461e-07,
+      "loss": 1.9172,
+      "step": 5450
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.7181,
+      "eval_samples_per_second": 393.142,
+      "eval_steps_per_second": 0.786,
+      "step": 5450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.975147038591108e-07,
+      "loss": 1.9201,
+      "step": 5500
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.6381,
+      "eval_samples_per_second": 395.63,
+      "eval_steps_per_second": 0.791,
+      "step": 5500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.029466557123754e-07,
+      "loss": 1.9039,
+      "step": 5550
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.812,
+      "eval_samples_per_second": 390.259,
+      "eval_steps_per_second": 0.781,
+      "step": 5550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.0837860756564e-07,
+      "loss": 1.9114,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.9609,
+      "eval_samples_per_second": 385.774,
+      "eval_steps_per_second": 0.772,
+      "step": 5600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.138105594189046e-07,
+      "loss": 1.916,
+      "step": 5650
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.6556,
+      "eval_samples_per_second": 395.082,
+      "eval_steps_per_second": 0.79,
+      "step": 5650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.192425112721693e-07,
+      "loss": 1.9077,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.8791,
+      "eval_samples_per_second": 388.226,
+      "eval_steps_per_second": 0.776,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.24674463125434e-07,
+      "loss": 1.8925,
+      "step": 5750
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.5794,
+      "eval_samples_per_second": 397.475,
+      "eval_steps_per_second": 0.795,
+      "step": 5750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.301064149786986e-07,
+      "loss": 1.9015,
+      "step": 5800
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.7644,
+      "eval_samples_per_second": 391.715,
+      "eval_steps_per_second": 0.783,
+      "step": 5800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.355383668319632e-07,
+      "loss": 1.896,
+      "step": 5850
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.7095,
+      "eval_samples_per_second": 393.407,
+      "eval_steps_per_second": 0.787,
+      "step": 5850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.409703186852278e-07,
+      "loss": 1.9001,
+      "step": 5900
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.5448,
+      "eval_samples_per_second": 398.571,
+      "eval_steps_per_second": 0.797,
+      "step": 5900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.464022705384926e-07,
+      "loss": 1.8975,
+      "step": 5950
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.7943,
+      "eval_samples_per_second": 390.799,
+      "eval_steps_per_second": 0.782,
+      "step": 5950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.518342223917572e-07,
+      "loss": 1.8957,
+      "step": 6000
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.4392,
+      "eval_samples_per_second": 401.954,
+      "eval_steps_per_second": 0.804,
+      "step": 6000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.572661742450218e-07,
+      "loss": 1.8916,
+      "step": 6050
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.7765,
+      "eval_samples_per_second": 391.343,
+      "eval_steps_per_second": 0.783,
+      "step": 6050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.626981260982864e-07,
+      "loss": 1.8893,
+      "step": 6100
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.8711,
+      "eval_samples_per_second": 388.468,
+      "eval_steps_per_second": 0.777,
+      "step": 6100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.68130077951551e-07,
+      "loss": 1.8851,
+      "step": 6150
+    },
+    {
+      "epoch": 0.07,
+      "eval_runtime": 12.8369,
+      "eval_samples_per_second": 389.502,
+      "eval_steps_per_second": 0.779,
+      "step": 6150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.735620298048157e-07,
+      "loss": 1.8834,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.7801,
+      "eval_samples_per_second": 391.234,
+      "eval_steps_per_second": 0.782,
+      "step": 6200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.789939816580803e-07,
+      "loss": 1.8826,
+      "step": 6250
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.45,
+      "eval_samples_per_second": 401.607,
+      "eval_steps_per_second": 0.803,
+      "step": 6250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.844259335113449e-07,
+      "loss": 1.8896,
+      "step": 6300
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.8465,
+      "eval_samples_per_second": 389.21,
+      "eval_steps_per_second": 0.778,
+      "step": 6300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.898578853646096e-07,
+      "loss": 1.877,
+      "step": 6350
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.9531,
+      "eval_samples_per_second": 386.008,
+      "eval_steps_per_second": 0.772,
+      "step": 6350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.952898372178743e-07,
+      "loss": 1.8742,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.7805,
+      "eval_samples_per_second": 391.221,
+      "eval_steps_per_second": 0.782,
+      "step": 6400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.007217890711389e-07,
+      "loss": 1.8684,
+      "step": 6450
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.8735,
+      "eval_samples_per_second": 388.393,
+      "eval_steps_per_second": 0.777,
+      "step": 6450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.061537409244035e-07,
+      "loss": 1.871,
+      "step": 6500
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.9728,
+      "eval_samples_per_second": 385.423,
+      "eval_steps_per_second": 0.771,
+      "step": 6500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.115856927776681e-07,
+      "loss": 1.8676,
+      "step": 6550
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.9365,
+      "eval_samples_per_second": 386.505,
+      "eval_steps_per_second": 0.773,
+      "step": 6550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.170176446309328e-07,
+      "loss": 1.8617,
+      "step": 6600
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.7791,
+      "eval_samples_per_second": 391.263,
+      "eval_steps_per_second": 0.783,
+      "step": 6600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.224495964841975e-07,
+      "loss": 1.8555,
+      "step": 6650
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.8206,
+      "eval_samples_per_second": 389.997,
+      "eval_steps_per_second": 0.78,
+      "step": 6650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.278815483374621e-07,
+      "loss": 1.8654,
+      "step": 6700
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.9573,
+      "eval_samples_per_second": 385.884,
+      "eval_steps_per_second": 0.772,
+      "step": 6700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.333135001907267e-07,
+      "loss": 1.8674,
+      "step": 6750
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.9202,
+      "eval_samples_per_second": 386.991,
+      "eval_steps_per_second": 0.774,
+      "step": 6750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.387454520439914e-07,
+      "loss": 1.8582,
+      "step": 6800
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 13.0048,
+      "eval_samples_per_second": 384.474,
+      "eval_steps_per_second": 0.769,
+      "step": 6800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.441774038972561e-07,
+      "loss": 1.8561,
+      "step": 6850
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.8387,
+      "eval_samples_per_second": 389.447,
+      "eval_steps_per_second": 0.779,
+      "step": 6850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.496093557505207e-07,
+      "loss": 1.8612,
+      "step": 6900
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.9925,
+      "eval_samples_per_second": 384.837,
+      "eval_steps_per_second": 0.77,
+      "step": 6900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.550413076037853e-07,
+      "loss": 1.8592,
+      "step": 6950
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.8615,
+      "eval_samples_per_second": 388.756,
+      "eval_steps_per_second": 0.778,
+      "step": 6950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.6047325945705e-07,
+      "loss": 1.8505,
+      "step": 7000
+    },
+    {
+      "epoch": 0.08,
+      "eval_runtime": 12.5402,
+      "eval_samples_per_second": 398.717,
+      "eval_steps_per_second": 0.797,
+      "step": 7000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.659052113103146e-07,
+      "loss": 1.8511,
+      "step": 7050
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8746,
+      "eval_samples_per_second": 388.363,
+      "eval_steps_per_second": 0.777,
+      "step": 7050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.713371631635793e-07,
+      "loss": 1.8494,
+      "step": 7100
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8633,
+      "eval_samples_per_second": 388.702,
+      "eval_steps_per_second": 0.777,
+      "step": 7100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.767691150168439e-07,
+      "loss": 1.8482,
+      "step": 7150
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.9012,
+      "eval_samples_per_second": 387.562,
+      "eval_steps_per_second": 0.775,
+      "step": 7150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.822010668701085e-07,
+      "loss": 1.8431,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.5337,
+      "eval_samples_per_second": 398.923,
+      "eval_steps_per_second": 0.798,
+      "step": 7200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.876330187233731e-07,
+      "loss": 1.8442,
+      "step": 7250
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8287,
+      "eval_samples_per_second": 389.75,
+      "eval_steps_per_second": 0.779,
+      "step": 7250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.930649705766378e-07,
+      "loss": 1.8458,
+      "step": 7300
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8502,
+      "eval_samples_per_second": 389.1,
+      "eval_steps_per_second": 0.778,
+      "step": 7300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.984969224299026e-07,
+      "loss": 1.8369,
+      "step": 7350
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.5085,
+      "eval_samples_per_second": 399.728,
+      "eval_steps_per_second": 0.799,
+      "step": 7350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.039288742831672e-07,
+      "loss": 1.8405,
+      "step": 7400
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.7767,
+      "eval_samples_per_second": 391.338,
+      "eval_steps_per_second": 0.783,
+      "step": 7400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.093608261364318e-07,
+      "loss": 1.8379,
+      "step": 7450
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8557,
+      "eval_samples_per_second": 388.931,
+      "eval_steps_per_second": 0.778,
+      "step": 7450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.147927779896965e-07,
+      "loss": 1.8338,
+      "step": 7500
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.976,
+      "eval_samples_per_second": 385.326,
+      "eval_steps_per_second": 0.771,
+      "step": 7500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.202247298429611e-07,
+      "loss": 1.8316,
+      "step": 7550
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8238,
+      "eval_samples_per_second": 389.899,
+      "eval_steps_per_second": 0.78,
+      "step": 7550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.256566816962257e-07,
+      "loss": 1.8299,
+      "step": 7600
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.9423,
+      "eval_samples_per_second": 386.33,
+      "eval_steps_per_second": 0.773,
+      "step": 7600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.310886335494903e-07,
+      "loss": 1.8279,
+      "step": 7650
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.7314,
+      "eval_samples_per_second": 392.73,
+      "eval_steps_per_second": 0.785,
+      "step": 7650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.365205854027549e-07,
+      "loss": 1.8281,
+      "step": 7700
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.9448,
+      "eval_samples_per_second": 386.256,
+      "eval_steps_per_second": 0.773,
+      "step": 7700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.419525372560196e-07,
+      "loss": 1.8233,
+      "step": 7750
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.8039,
+      "eval_samples_per_second": 390.505,
+      "eval_steps_per_second": 0.781,
+      "step": 7750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.473844891092844e-07,
+      "loss": 1.8306,
+      "step": 7800
+    },
+    {
+      "epoch": 0.09,
+      "eval_runtime": 12.6932,
+      "eval_samples_per_second": 393.911,
+      "eval_steps_per_second": 0.788,
+      "step": 7800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.52816440962549e-07,
+      "loss": 1.818,
+      "step": 7850
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.4575,
+      "eval_samples_per_second": 401.365,
+      "eval_steps_per_second": 0.803,
+      "step": 7850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.582483928158136e-07,
+      "loss": 1.8141,
+      "step": 7900
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.7841,
+      "eval_samples_per_second": 391.111,
+      "eval_steps_per_second": 0.782,
+      "step": 7900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.636803446690783e-07,
+      "loss": 1.8195,
+      "step": 7950
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.7337,
+      "eval_samples_per_second": 392.658,
+      "eval_steps_per_second": 0.785,
+      "step": 7950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.691122965223429e-07,
+      "loss": 1.8174,
+      "step": 8000
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.8197,
+      "eval_samples_per_second": 390.026,
+      "eval_steps_per_second": 0.78,
+      "step": 8000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.745442483756075e-07,
+      "loss": 1.8082,
+      "step": 8050
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.7691,
+      "eval_samples_per_second": 391.57,
+      "eval_steps_per_second": 0.783,
+      "step": 8050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.799762002288721e-07,
+      "loss": 1.8107,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.9388,
+      "eval_samples_per_second": 386.435,
+      "eval_steps_per_second": 0.773,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.854081520821367e-07,
+      "loss": 1.8087,
+      "step": 8150
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.8758,
+      "eval_samples_per_second": 388.324,
+      "eval_steps_per_second": 0.777,
+      "step": 8150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.908401039354014e-07,
+      "loss": 1.8149,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.9367,
+      "eval_samples_per_second": 386.498,
+      "eval_steps_per_second": 0.773,
+      "step": 8200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.962720557886661e-07,
+      "loss": 1.8054,
+      "step": 8250
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.7709,
+      "eval_samples_per_second": 391.517,
+      "eval_steps_per_second": 0.783,
+      "step": 8250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.017040076419307e-07,
+      "loss": 1.8054,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.5096,
+      "eval_samples_per_second": 399.694,
+      "eval_steps_per_second": 0.799,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.071359594951953e-07,
+      "loss": 1.8074,
+      "step": 8350
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.8824,
+      "eval_samples_per_second": 388.127,
+      "eval_steps_per_second": 0.776,
+      "step": 8350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.1256791134846e-07,
+      "loss": 1.8053,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.5588,
+      "eval_samples_per_second": 398.127,
+      "eval_steps_per_second": 0.796,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.179998632017247e-07,
+      "loss": 1.7963,
+      "step": 8450
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.6912,
+      "eval_samples_per_second": 393.975,
+      "eval_steps_per_second": 0.788,
+      "step": 8450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.234318150549893e-07,
+      "loss": 1.8007,
+      "step": 8500
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.8738,
+      "eval_samples_per_second": 388.387,
+      "eval_steps_per_second": 0.777,
+      "step": 8500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.288637669082539e-07,
+      "loss": 1.7921,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.6413,
+      "eval_samples_per_second": 395.529,
+      "eval_steps_per_second": 0.791,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.342957187615185e-07,
+      "loss": 1.7978,
+      "step": 8600
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.795,
+      "eval_samples_per_second": 390.777,
+      "eval_steps_per_second": 0.782,
+      "step": 8600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.397276706147832e-07,
+      "loss": 1.7942,
+      "step": 8650
+    },
+    {
+      "epoch": 0.1,
+      "eval_runtime": 12.4796,
+      "eval_samples_per_second": 400.653,
+      "eval_steps_per_second": 0.801,
+      "step": 8650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.451596224680479e-07,
+      "loss": 1.7924,
+      "step": 8700
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.9002,
+      "eval_samples_per_second": 387.591,
+      "eval_steps_per_second": 0.775,
+      "step": 8700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.505915743213125e-07,
+      "loss": 1.7914,
+      "step": 8750
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8042,
+      "eval_samples_per_second": 390.498,
+      "eval_steps_per_second": 0.781,
+      "step": 8750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.560235261745771e-07,
+      "loss": 1.7886,
+      "step": 8800
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8312,
+      "eval_samples_per_second": 389.676,
+      "eval_steps_per_second": 0.779,
+      "step": 8800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.614554780278419e-07,
+      "loss": 1.7877,
+      "step": 8850
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.6466,
+      "eval_samples_per_second": 395.364,
+      "eval_steps_per_second": 0.791,
+      "step": 8850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.668874298811064e-07,
+      "loss": 1.7888,
+      "step": 8900
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8579,
+      "eval_samples_per_second": 388.866,
+      "eval_steps_per_second": 0.778,
+      "step": 8900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72319381734371e-07,
+      "loss": 1.7813,
+      "step": 8950
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8434,
+      "eval_samples_per_second": 389.305,
+      "eval_steps_per_second": 0.779,
+      "step": 8950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.777513335876356e-07,
+      "loss": 1.7826,
+      "step": 9000
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8754,
+      "eval_samples_per_second": 388.337,
+      "eval_steps_per_second": 0.777,
+      "step": 9000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831832854409003e-07,
+      "loss": 1.7787,
+      "step": 9050
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.653,
+      "eval_samples_per_second": 395.164,
+      "eval_steps_per_second": 0.79,
+      "step": 9050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.886152372941649e-07,
+      "loss": 1.7771,
+      "step": 9100
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8241,
+      "eval_samples_per_second": 389.892,
+      "eval_steps_per_second": 0.78,
+      "step": 9100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940471891474298e-07,
+      "loss": 1.7753,
+      "step": 9150
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.9237,
+      "eval_samples_per_second": 386.887,
+      "eval_steps_per_second": 0.774,
+      "step": 9150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.994791410006943e-07,
+      "loss": 1.7785,
+      "step": 9200
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8616,
+      "eval_samples_per_second": 388.753,
+      "eval_steps_per_second": 0.778,
+      "step": 9200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.004911092853959e-06,
+      "loss": 1.7707,
+      "step": 9250
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.9012,
+      "eval_samples_per_second": 387.56,
+      "eval_steps_per_second": 0.775,
+      "step": 9250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0103430447072236e-06,
+      "loss": 1.7721,
+      "step": 9300
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.8849,
+      "eval_samples_per_second": 388.05,
+      "eval_steps_per_second": 0.776,
+      "step": 9300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0157749965604883e-06,
+      "loss": 1.76,
+      "step": 9350
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.6868,
+      "eval_samples_per_second": 394.111,
+      "eval_steps_per_second": 0.788,
+      "step": 9350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0212069484137528e-06,
+      "loss": 1.7675,
+      "step": 9400
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.9205,
+      "eval_samples_per_second": 386.983,
+      "eval_steps_per_second": 0.774,
+      "step": 9400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0266389002670175e-06,
+      "loss": 1.7613,
+      "step": 9450
+    },
+    {
+      "epoch": 0.11,
+      "eval_runtime": 12.6546,
+      "eval_samples_per_second": 395.112,
+      "eval_steps_per_second": 0.79,
+      "step": 9450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.032070852120282e-06,
+      "loss": 1.7637,
+      "step": 9500
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.945,
+      "eval_samples_per_second": 386.248,
+      "eval_steps_per_second": 0.772,
+      "step": 9500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0375028039735468e-06,
+      "loss": 1.7607,
+      "step": 9550
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.9489,
+      "eval_samples_per_second": 386.132,
+      "eval_steps_per_second": 0.772,
+      "step": 9550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0429347558268115e-06,
+      "loss": 1.7491,
+      "step": 9600
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.835,
+      "eval_samples_per_second": 389.56,
+      "eval_steps_per_second": 0.779,
+      "step": 9600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.048366707680076e-06,
+      "loss": 1.7595,
+      "step": 9650
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.9166,
+      "eval_samples_per_second": 387.099,
+      "eval_steps_per_second": 0.774,
+      "step": 9650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0537986595333407e-06,
+      "loss": 1.7582,
+      "step": 9700
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.874,
+      "eval_samples_per_second": 388.379,
+      "eval_steps_per_second": 0.777,
+      "step": 9700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0592306113866052e-06,
+      "loss": 1.7532,
+      "step": 9750
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.813,
+      "eval_samples_per_second": 390.228,
+      "eval_steps_per_second": 0.78,
+      "step": 9750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.06466256323987e-06,
+      "loss": 1.7525,
+      "step": 9800
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.6742,
+      "eval_samples_per_second": 394.504,
+      "eval_steps_per_second": 0.789,
+      "step": 9800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0700945150931345e-06,
+      "loss": 1.7575,
+      "step": 9850
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.6613,
+      "eval_samples_per_second": 394.905,
+      "eval_steps_per_second": 0.79,
+      "step": 9850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0755264669463992e-06,
+      "loss": 1.7432,
+      "step": 9900
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.693,
+      "eval_samples_per_second": 393.919,
+      "eval_steps_per_second": 0.788,
+      "step": 9900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.080958418799664e-06,
+      "loss": 1.7508,
+      "step": 9950
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.534,
+      "eval_samples_per_second": 398.915,
+      "eval_steps_per_second": 0.798,
+      "step": 9950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0863903706529287e-06,
+      "loss": 1.7475,
+      "step": 10000
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.6087,
+      "eval_samples_per_second": 396.552,
+      "eval_steps_per_second": 0.793,
+      "step": 10000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.0918223225061932e-06,
+      "loss": 1.745,
+      "step": 10050
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.9129,
+      "eval_samples_per_second": 387.209,
+      "eval_steps_per_second": 0.774,
+      "step": 10050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.097254274359458e-06,
+      "loss": 1.7486,
+      "step": 10100
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.6957,
+      "eval_samples_per_second": 393.833,
+      "eval_steps_per_second": 0.788,
+      "step": 10100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1026862262127224e-06,
+      "loss": 1.7382,
+      "step": 10150
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.8424,
+      "eval_samples_per_second": 389.336,
+      "eval_steps_per_second": 0.779,
+      "step": 10150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1081181780659872e-06,
+      "loss": 1.7444,
+      "step": 10200
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.8126,
+      "eval_samples_per_second": 390.242,
+      "eval_steps_per_second": 0.78,
+      "step": 10200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1135501299192517e-06,
+      "loss": 1.7428,
+      "step": 10250
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.8807,
+      "eval_samples_per_second": 388.179,
+      "eval_steps_per_second": 0.776,
+      "step": 10250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1189820817725164e-06,
+      "loss": 1.7376,
+      "step": 10300
+    },
+    {
+      "epoch": 0.12,
+      "eval_runtime": 12.4784,
+      "eval_samples_per_second": 400.693,
+      "eval_steps_per_second": 0.801,
+      "step": 10300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.124414033625781e-06,
+      "loss": 1.7357,
+      "step": 10350
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.8663,
+      "eval_samples_per_second": 388.612,
+      "eval_steps_per_second": 0.777,
+      "step": 10350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1298459854790456e-06,
+      "loss": 1.7342,
+      "step": 10400
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.906,
+      "eval_samples_per_second": 387.415,
+      "eval_steps_per_second": 0.775,
+      "step": 10400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1352779373323104e-06,
+      "loss": 1.7385,
+      "step": 10450
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.9325,
+      "eval_samples_per_second": 386.622,
+      "eval_steps_per_second": 0.773,
+      "step": 10450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.140709889185575e-06,
+      "loss": 1.7308,
+      "step": 10500
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.8102,
+      "eval_samples_per_second": 390.313,
+      "eval_steps_per_second": 0.781,
+      "step": 10500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1461418410388396e-06,
+      "loss": 1.7307,
+      "step": 10550
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.7134,
+      "eval_samples_per_second": 393.286,
+      "eval_steps_per_second": 0.787,
+      "step": 10550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1515737928921043e-06,
+      "loss": 1.7342,
+      "step": 10600
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.8299,
+      "eval_samples_per_second": 389.713,
+      "eval_steps_per_second": 0.779,
+      "step": 10600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1570057447453689e-06,
+      "loss": 1.7315,
+      "step": 10650
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.5177,
+      "eval_samples_per_second": 399.433,
+      "eval_steps_per_second": 0.799,
+      "step": 10650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1624376965986336e-06,
+      "loss": 1.7302,
+      "step": 10700
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.9224,
+      "eval_samples_per_second": 386.926,
+      "eval_steps_per_second": 0.774,
+      "step": 10700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.167869648451898e-06,
+      "loss": 1.7208,
+      "step": 10750
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.7543,
+      "eval_samples_per_second": 392.024,
+      "eval_steps_per_second": 0.784,
+      "step": 10750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1733016003051628e-06,
+      "loss": 1.7272,
+      "step": 10800
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.9451,
+      "eval_samples_per_second": 386.245,
+      "eval_steps_per_second": 0.772,
+      "step": 10800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1787335521584273e-06,
+      "loss": 1.719,
+      "step": 10850
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.5593,
+      "eval_samples_per_second": 398.113,
+      "eval_steps_per_second": 0.796,
+      "step": 10850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1841655040116923e-06,
+      "loss": 1.7152,
+      "step": 10900
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.7161,
+      "eval_samples_per_second": 393.202,
+      "eval_steps_per_second": 0.786,
+      "step": 10900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1895974558649568e-06,
+      "loss": 1.7115,
+      "step": 10950
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.9306,
+      "eval_samples_per_second": 386.681,
+      "eval_steps_per_second": 0.773,
+      "step": 10950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1950294077182215e-06,
+      "loss": 1.7198,
+      "step": 11000
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.5481,
+      "eval_samples_per_second": 398.465,
+      "eval_steps_per_second": 0.797,
+      "step": 11000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.200461359571486e-06,
+      "loss": 1.7163,
+      "step": 11050
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.6056,
+      "eval_samples_per_second": 396.648,
+      "eval_steps_per_second": 0.793,
+      "step": 11050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2058933114247508e-06,
+      "loss": 1.7092,
+      "step": 11100
+    },
+    {
+      "epoch": 0.13,
+      "eval_runtime": 12.626,
+      "eval_samples_per_second": 396.007,
+      "eval_steps_per_second": 0.792,
+      "step": 11100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2113252632780153e-06,
+      "loss": 1.7079,
+      "step": 11150
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.8023,
+      "eval_samples_per_second": 390.554,
+      "eval_steps_per_second": 0.781,
+      "step": 11150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.21675721513128e-06,
+      "loss": 1.7174,
+      "step": 11200
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.9331,
+      "eval_samples_per_second": 386.604,
+      "eval_steps_per_second": 0.773,
+      "step": 11200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2221891669845445e-06,
+      "loss": 1.7017,
+      "step": 11250
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.948,
+      "eval_samples_per_second": 386.161,
+      "eval_steps_per_second": 0.772,
+      "step": 11250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2276211188378092e-06,
+      "loss": 1.7052,
+      "step": 11300
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.8735,
+      "eval_samples_per_second": 388.396,
+      "eval_steps_per_second": 0.777,
+      "step": 11300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.233053070691074e-06,
+      "loss": 1.7072,
+      "step": 11350
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.6953,
+      "eval_samples_per_second": 393.846,
+      "eval_steps_per_second": 0.788,
+      "step": 11350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2384850225443387e-06,
+      "loss": 1.7052,
+      "step": 11400
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.6324,
+      "eval_samples_per_second": 395.807,
+      "eval_steps_per_second": 0.792,
+      "step": 11400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2439169743976032e-06,
+      "loss": 1.7069,
+      "step": 11450
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 13.0238,
+      "eval_samples_per_second": 383.912,
+      "eval_steps_per_second": 0.768,
+      "step": 11450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.249348926250868e-06,
+      "loss": 1.7033,
+      "step": 11500
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.8025,
+      "eval_samples_per_second": 390.55,
+      "eval_steps_per_second": 0.781,
+      "step": 11500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2547808781041325e-06,
+      "loss": 1.6999,
+      "step": 11550
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.6142,
+      "eval_samples_per_second": 396.378,
+      "eval_steps_per_second": 0.793,
+      "step": 11550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2602128299573972e-06,
+      "loss": 1.6993,
+      "step": 11600
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.8637,
+      "eval_samples_per_second": 388.69,
+      "eval_steps_per_second": 0.777,
+      "step": 11600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2656447818106617e-06,
+      "loss": 1.693,
+      "step": 11650
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 13.007,
+      "eval_samples_per_second": 384.407,
+      "eval_steps_per_second": 0.769,
+      "step": 11650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2710767336639264e-06,
+      "loss": 1.6913,
+      "step": 11700
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.4913,
+      "eval_samples_per_second": 400.28,
+      "eval_steps_per_second": 0.801,
+      "step": 11700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.276508685517191e-06,
+      "loss": 1.6857,
+      "step": 11750
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.972,
+      "eval_samples_per_second": 385.446,
+      "eval_steps_per_second": 0.771,
+      "step": 11750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2819406373704557e-06,
+      "loss": 1.6917,
+      "step": 11800
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.9325,
+      "eval_samples_per_second": 386.624,
+      "eval_steps_per_second": 0.773,
+      "step": 11800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2873725892237204e-06,
+      "loss": 1.6813,
+      "step": 11850
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.8605,
+      "eval_samples_per_second": 388.786,
+      "eval_steps_per_second": 0.778,
+      "step": 11850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2928045410769851e-06,
+      "loss": 1.6845,
+      "step": 11900
+    },
+    {
+      "epoch": 0.14,
+      "eval_runtime": 12.8963,
+      "eval_samples_per_second": 387.709,
+      "eval_steps_per_second": 0.775,
+      "step": 11900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.2982364929302496e-06,
+      "loss": 1.6884,
+      "step": 11950
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.9379,
+      "eval_samples_per_second": 386.462,
+      "eval_steps_per_second": 0.773,
+      "step": 11950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3036684447835144e-06,
+      "loss": 1.6929,
+      "step": 12000
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.7264,
+      "eval_samples_per_second": 392.883,
+      "eval_steps_per_second": 0.786,
+      "step": 12000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3091003966367789e-06,
+      "loss": 1.6827,
+      "step": 12050
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.8986,
+      "eval_samples_per_second": 387.64,
+      "eval_steps_per_second": 0.775,
+      "step": 12050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3145323484900436e-06,
+      "loss": 1.6803,
+      "step": 12100
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.9039,
+      "eval_samples_per_second": 387.48,
+      "eval_steps_per_second": 0.775,
+      "step": 12100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3199643003433081e-06,
+      "loss": 1.6803,
+      "step": 12150
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.8763,
+      "eval_samples_per_second": 388.31,
+      "eval_steps_per_second": 0.777,
+      "step": 12150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3253962521965728e-06,
+      "loss": 1.6781,
+      "step": 12200
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.8049,
+      "eval_samples_per_second": 390.474,
+      "eval_steps_per_second": 0.781,
+      "step": 12200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3308282040498376e-06,
+      "loss": 1.6773,
+      "step": 12250
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.4806,
+      "eval_samples_per_second": 400.623,
+      "eval_steps_per_second": 0.801,
+      "step": 12250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.336260155903102e-06,
+      "loss": 1.6739,
+      "step": 12300
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.655,
+      "eval_samples_per_second": 395.102,
+      "eval_steps_per_second": 0.79,
+      "step": 12300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3416921077563668e-06,
+      "loss": 1.6697,
+      "step": 12350
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.9066,
+      "eval_samples_per_second": 387.399,
+      "eval_steps_per_second": 0.775,
+      "step": 12350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3471240596096313e-06,
+      "loss": 1.664,
+      "step": 12400
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.8185,
+      "eval_samples_per_second": 390.062,
+      "eval_steps_per_second": 0.78,
+      "step": 12400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.352556011462896e-06,
+      "loss": 1.6707,
+      "step": 12450
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 13.0289,
+      "eval_samples_per_second": 383.763,
+      "eval_steps_per_second": 0.768,
+      "step": 12450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3579879633161606e-06,
+      "loss": 1.667,
+      "step": 12500
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.5155,
+      "eval_samples_per_second": 399.506,
+      "eval_steps_per_second": 0.799,
+      "step": 12500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3634199151694253e-06,
+      "loss": 1.6627,
+      "step": 12550
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.8428,
+      "eval_samples_per_second": 389.325,
+      "eval_steps_per_second": 0.779,
+      "step": 12550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3688518670226898e-06,
+      "loss": 1.6656,
+      "step": 12600
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.9159,
+      "eval_samples_per_second": 387.118,
+      "eval_steps_per_second": 0.774,
+      "step": 12600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3742838188759545e-06,
+      "loss": 1.6617,
+      "step": 12650
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.8472,
+      "eval_samples_per_second": 389.19,
+      "eval_steps_per_second": 0.778,
+      "step": 12650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3797157707292193e-06,
+      "loss": 1.6582,
+      "step": 12700
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.7659,
+      "eval_samples_per_second": 391.669,
+      "eval_steps_per_second": 0.783,
+      "step": 12700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.385147722582484e-06,
+      "loss": 1.6546,
+      "step": 12750
+    },
+    {
+      "epoch": 0.15,
+      "eval_runtime": 12.6414,
+      "eval_samples_per_second": 395.527,
+      "eval_steps_per_second": 0.791,
+      "step": 12750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3905796744357485e-06,
+      "loss": 1.6527,
+      "step": 12800
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.944,
+      "eval_samples_per_second": 386.279,
+      "eval_steps_per_second": 0.773,
+      "step": 12800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.3960116262890132e-06,
+      "loss": 1.6578,
+      "step": 12850
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.6668,
+      "eval_samples_per_second": 394.734,
+      "eval_steps_per_second": 0.789,
+      "step": 12850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4014435781422778e-06,
+      "loss": 1.6568,
+      "step": 12900
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.8172,
+      "eval_samples_per_second": 390.102,
+      "eval_steps_per_second": 0.78,
+      "step": 12900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4068755299955425e-06,
+      "loss": 1.6503,
+      "step": 12950
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.8171,
+      "eval_samples_per_second": 390.105,
+      "eval_steps_per_second": 0.78,
+      "step": 12950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.412307481848807e-06,
+      "loss": 1.6442,
+      "step": 13000
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.7149,
+      "eval_samples_per_second": 393.239,
+      "eval_steps_per_second": 0.786,
+      "step": 13000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4177394337020717e-06,
+      "loss": 1.639,
+      "step": 13050
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 13.0433,
+      "eval_samples_per_second": 383.338,
+      "eval_steps_per_second": 0.767,
+      "step": 13050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4231713855553362e-06,
+      "loss": 1.6435,
+      "step": 13100
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.7704,
+      "eval_samples_per_second": 391.531,
+      "eval_steps_per_second": 0.783,
+      "step": 13100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4286033374086012e-06,
+      "loss": 1.6453,
+      "step": 13150
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.5385,
+      "eval_samples_per_second": 398.77,
+      "eval_steps_per_second": 0.798,
+      "step": 13150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4340352892618657e-06,
+      "loss": 1.6378,
+      "step": 13200
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.977,
+      "eval_samples_per_second": 385.296,
+      "eval_steps_per_second": 0.771,
+      "step": 13200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4394672411151304e-06,
+      "loss": 1.6314,
+      "step": 13250
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 13.0367,
+      "eval_samples_per_second": 383.533,
+      "eval_steps_per_second": 0.767,
+      "step": 13250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.444899192968395e-06,
+      "loss": 1.6318,
+      "step": 13300
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.7365,
+      "eval_samples_per_second": 392.572,
+      "eval_steps_per_second": 0.785,
+      "step": 13300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4503311448216597e-06,
+      "loss": 1.6378,
+      "step": 13350
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.8723,
+      "eval_samples_per_second": 388.43,
+      "eval_steps_per_second": 0.777,
+      "step": 13350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4557630966749242e-06,
+      "loss": 1.626,
+      "step": 13400
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.8605,
+      "eval_samples_per_second": 388.786,
+      "eval_steps_per_second": 0.778,
+      "step": 13400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.461195048528189e-06,
+      "loss": 1.6235,
+      "step": 13450
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.5123,
+      "eval_samples_per_second": 399.607,
+      "eval_steps_per_second": 0.799,
+      "step": 13450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4666270003814534e-06,
+      "loss": 1.62,
+      "step": 13500
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.5399,
+      "eval_samples_per_second": 398.727,
+      "eval_steps_per_second": 0.797,
+      "step": 13500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4720589522347181e-06,
+      "loss": 1.6224,
+      "step": 13550
+    },
+    {
+      "epoch": 0.16,
+      "eval_runtime": 12.7693,
+      "eval_samples_per_second": 391.564,
+      "eval_steps_per_second": 0.783,
+      "step": 13550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4774909040879829e-06,
+      "loss": 1.6213,
+      "step": 13600
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8972,
+      "eval_samples_per_second": 387.68,
+      "eval_steps_per_second": 0.775,
+      "step": 13600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4829228559412476e-06,
+      "loss": 1.6212,
+      "step": 13650
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.9608,
+      "eval_samples_per_second": 385.777,
+      "eval_steps_per_second": 0.772,
+      "step": 13650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4883548077945121e-06,
+      "loss": 1.6158,
+      "step": 13700
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.7918,
+      "eval_samples_per_second": 390.877,
+      "eval_steps_per_second": 0.782,
+      "step": 13700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4937867596477768e-06,
+      "loss": 1.6039,
+      "step": 13750
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.9565,
+      "eval_samples_per_second": 385.906,
+      "eval_steps_per_second": 0.772,
+      "step": 13750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.4992187115010414e-06,
+      "loss": 1.6176,
+      "step": 13800
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8965,
+      "eval_samples_per_second": 387.703,
+      "eval_steps_per_second": 0.775,
+      "step": 13800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.504650663354306e-06,
+      "loss": 1.6094,
+      "step": 13850
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8222,
+      "eval_samples_per_second": 389.947,
+      "eval_steps_per_second": 0.78,
+      "step": 13850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5100826152075706e-06,
+      "loss": 1.6057,
+      "step": 13900
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.9112,
+      "eval_samples_per_second": 387.259,
+      "eval_steps_per_second": 0.775,
+      "step": 13900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5155145670608353e-06,
+      "loss": 1.5997,
+      "step": 13950
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8616,
+      "eval_samples_per_second": 388.753,
+      "eval_steps_per_second": 0.778,
+      "step": 13950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5209465189141e-06,
+      "loss": 1.6034,
+      "step": 14000
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8553,
+      "eval_samples_per_second": 388.945,
+      "eval_steps_per_second": 0.778,
+      "step": 14000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5263784707673648e-06,
+      "loss": 1.5999,
+      "step": 14050
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.5349,
+      "eval_samples_per_second": 398.887,
+      "eval_steps_per_second": 0.798,
+      "step": 14050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5318104226206293e-06,
+      "loss": 1.6002,
+      "step": 14100
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.7457,
+      "eval_samples_per_second": 392.29,
+      "eval_steps_per_second": 0.785,
+      "step": 14100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.537242374473894e-06,
+      "loss": 1.5967,
+      "step": 14150
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.9136,
+      "eval_samples_per_second": 387.187,
+      "eval_steps_per_second": 0.774,
+      "step": 14150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5426743263271585e-06,
+      "loss": 1.59,
+      "step": 14200
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8547,
+      "eval_samples_per_second": 388.962,
+      "eval_steps_per_second": 0.778,
+      "step": 14200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5481062781804233e-06,
+      "loss": 1.5947,
+      "step": 14250
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.7235,
+      "eval_samples_per_second": 392.973,
+      "eval_steps_per_second": 0.786,
+      "step": 14250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5535382300336878e-06,
+      "loss": 1.5901,
+      "step": 14300
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8529,
+      "eval_samples_per_second": 389.018,
+      "eval_steps_per_second": 0.778,
+      "step": 14300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5589701818869525e-06,
+      "loss": 1.5912,
+      "step": 14350
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.9685,
+      "eval_samples_per_second": 385.549,
+      "eval_steps_per_second": 0.771,
+      "step": 14350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.564402133740217e-06,
+      "loss": 1.5884,
+      "step": 14400
+    },
+    {
+      "epoch": 0.17,
+      "eval_runtime": 12.8015,
+      "eval_samples_per_second": 390.58,
+      "eval_steps_per_second": 0.781,
+      "step": 14400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5698340855934818e-06,
+      "loss": 1.5867,
+      "step": 14450
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8316,
+      "eval_samples_per_second": 389.663,
+      "eval_steps_per_second": 0.779,
+      "step": 14450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5752660374467463e-06,
+      "loss": 1.5844,
+      "step": 14500
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8909,
+      "eval_samples_per_second": 387.87,
+      "eval_steps_per_second": 0.776,
+      "step": 14500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.580697989300011e-06,
+      "loss": 1.5855,
+      "step": 14550
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8702,
+      "eval_samples_per_second": 388.495,
+      "eval_steps_per_second": 0.777,
+      "step": 14550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5861299411532755e-06,
+      "loss": 1.5816,
+      "step": 14600
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8723,
+      "eval_samples_per_second": 388.431,
+      "eval_steps_per_second": 0.777,
+      "step": 14600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5915618930065402e-06,
+      "loss": 1.5846,
+      "step": 14650
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.6187,
+      "eval_samples_per_second": 396.239,
+      "eval_steps_per_second": 0.792,
+      "step": 14650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5969938448598052e-06,
+      "loss": 1.583,
+      "step": 14700
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8265,
+      "eval_samples_per_second": 389.817,
+      "eval_steps_per_second": 0.78,
+      "step": 14700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6024257967130695e-06,
+      "loss": 1.5773,
+      "step": 14750
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8807,
+      "eval_samples_per_second": 388.177,
+      "eval_steps_per_second": 0.776,
+      "step": 14750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6078577485663344e-06,
+      "loss": 1.582,
+      "step": 14800
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.6598,
+      "eval_samples_per_second": 394.95,
+      "eval_steps_per_second": 0.79,
+      "step": 14800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.613289700419599e-06,
+      "loss": 1.5722,
+      "step": 14850
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8407,
+      "eval_samples_per_second": 389.386,
+      "eval_steps_per_second": 0.779,
+      "step": 14850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6187216522728637e-06,
+      "loss": 1.5742,
+      "step": 14900
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.9551,
+      "eval_samples_per_second": 385.947,
+      "eval_steps_per_second": 0.772,
+      "step": 14900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6241536041261282e-06,
+      "loss": 1.5706,
+      "step": 14950
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.774,
+      "eval_samples_per_second": 391.42,
+      "eval_steps_per_second": 0.783,
+      "step": 14950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.629585555979393e-06,
+      "loss": 1.5747,
+      "step": 15000
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8291,
+      "eval_samples_per_second": 389.74,
+      "eval_steps_per_second": 0.779,
+      "step": 15000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6350175078326574e-06,
+      "loss": 1.5655,
+      "step": 15050
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.6131,
+      "eval_samples_per_second": 396.413,
+      "eval_steps_per_second": 0.793,
+      "step": 15050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6404494596859221e-06,
+      "loss": 1.5685,
+      "step": 15100
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 13.0103,
+      "eval_samples_per_second": 384.31,
+      "eval_steps_per_second": 0.769,
+      "step": 15100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6458814115391869e-06,
+      "loss": 1.5699,
+      "step": 15150
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.8694,
+      "eval_samples_per_second": 388.519,
+      "eval_steps_per_second": 0.777,
+      "step": 15150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6513133633924514e-06,
+      "loss": 1.5651,
+      "step": 15200
+    },
+    {
+      "epoch": 0.18,
+      "eval_runtime": 12.7699,
+      "eval_samples_per_second": 391.546,
+      "eval_steps_per_second": 0.783,
+      "step": 15200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6567453152457161e-06,
+      "loss": 1.5692,
+      "step": 15250
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.9094,
+      "eval_samples_per_second": 387.315,
+      "eval_steps_per_second": 0.775,
+      "step": 15250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6621772670989806e-06,
+      "loss": 1.567,
+      "step": 15300
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.7347,
+      "eval_samples_per_second": 392.629,
+      "eval_steps_per_second": 0.785,
+      "step": 15300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6676092189522454e-06,
+      "loss": 1.5624,
+      "step": 15350
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.7426,
+      "eval_samples_per_second": 392.384,
+      "eval_steps_per_second": 0.785,
+      "step": 15350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6730411708055099e-06,
+      "loss": 1.5583,
+      "step": 15400
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.8432,
+      "eval_samples_per_second": 389.311,
+      "eval_steps_per_second": 0.779,
+      "step": 15400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6784731226587746e-06,
+      "loss": 1.564,
+      "step": 15450
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.7444,
+      "eval_samples_per_second": 392.33,
+      "eval_steps_per_second": 0.785,
+      "step": 15450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6839050745120391e-06,
+      "loss": 1.562,
+      "step": 15500
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.9181,
+      "eval_samples_per_second": 387.053,
+      "eval_steps_per_second": 0.774,
+      "step": 15500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6893370263653038e-06,
+      "loss": 1.5623,
+      "step": 15550
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.911,
+      "eval_samples_per_second": 387.266,
+      "eval_steps_per_second": 0.775,
+      "step": 15550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6947689782185688e-06,
+      "loss": 1.5582,
+      "step": 15600
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.859,
+      "eval_samples_per_second": 388.833,
+      "eval_steps_per_second": 0.778,
+      "step": 15600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.700200930071833e-06,
+      "loss": 1.5552,
+      "step": 15650
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.9335,
+      "eval_samples_per_second": 386.594,
+      "eval_steps_per_second": 0.773,
+      "step": 15650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.705632881925098e-06,
+      "loss": 1.5575,
+      "step": 15700
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.8439,
+      "eval_samples_per_second": 389.29,
+      "eval_steps_per_second": 0.779,
+      "step": 15700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7110648337783623e-06,
+      "loss": 1.5564,
+      "step": 15750
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.9273,
+      "eval_samples_per_second": 386.778,
+      "eval_steps_per_second": 0.774,
+      "step": 15750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7164967856316273e-06,
+      "loss": 1.5501,
+      "step": 15800
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.8957,
+      "eval_samples_per_second": 387.725,
+      "eval_steps_per_second": 0.775,
+      "step": 15800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7219287374848916e-06,
+      "loss": 1.5555,
+      "step": 15850
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 13.0823,
+      "eval_samples_per_second": 382.194,
+      "eval_steps_per_second": 0.764,
+      "step": 15850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7273606893381565e-06,
+      "loss": 1.5469,
+      "step": 15900
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 13.1484,
+      "eval_samples_per_second": 380.274,
+      "eval_steps_per_second": 0.761,
+      "step": 15900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7327926411914208e-06,
+      "loss": 1.5462,
+      "step": 15950
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.915,
+      "eval_samples_per_second": 387.146,
+      "eval_steps_per_second": 0.774,
+      "step": 15950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7382245930446857e-06,
+      "loss": 1.55,
+      "step": 16000
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.5475,
+      "eval_samples_per_second": 398.485,
+      "eval_steps_per_second": 0.797,
+      "step": 16000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7436565448979505e-06,
+      "loss": 1.5551,
+      "step": 16050
+    },
+    {
+      "epoch": 0.19,
+      "eval_runtime": 12.5545,
+      "eval_samples_per_second": 398.263,
+      "eval_steps_per_second": 0.797,
+      "step": 16050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.749088496751215e-06,
+      "loss": 1.5494,
+      "step": 16100
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.4707,
+      "eval_samples_per_second": 400.94,
+      "eval_steps_per_second": 0.802,
+      "step": 16100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7545204486044797e-06,
+      "loss": 1.5396,
+      "step": 16150
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.7359,
+      "eval_samples_per_second": 392.59,
+      "eval_steps_per_second": 0.785,
+      "step": 16150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7599524004577442e-06,
+      "loss": 1.54,
+      "step": 16200
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.5065,
+      "eval_samples_per_second": 399.792,
+      "eval_steps_per_second": 0.8,
+      "step": 16200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.765384352311009e-06,
+      "loss": 1.5408,
+      "step": 16250
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.6874,
+      "eval_samples_per_second": 394.091,
+      "eval_steps_per_second": 0.788,
+      "step": 16250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7708163041642735e-06,
+      "loss": 1.5348,
+      "step": 16300
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.7825,
+      "eval_samples_per_second": 391.159,
+      "eval_steps_per_second": 0.782,
+      "step": 16300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7762482560175382e-06,
+      "loss": 1.5421,
+      "step": 16350
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.8036,
+      "eval_samples_per_second": 390.514,
+      "eval_steps_per_second": 0.781,
+      "step": 16350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7816802078708027e-06,
+      "loss": 1.5385,
+      "step": 16400
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.8978,
+      "eval_samples_per_second": 387.664,
+      "eval_steps_per_second": 0.775,
+      "step": 16400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7871121597240674e-06,
+      "loss": 1.5398,
+      "step": 16450
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.7173,
+      "eval_samples_per_second": 393.164,
+      "eval_steps_per_second": 0.786,
+      "step": 16450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7925441115773322e-06,
+      "loss": 1.5397,
+      "step": 16500
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.5256,
+      "eval_samples_per_second": 399.182,
+      "eval_steps_per_second": 0.798,
+      "step": 16500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7979760634305967e-06,
+      "loss": 1.5316,
+      "step": 16550
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.9375,
+      "eval_samples_per_second": 386.474,
+      "eval_steps_per_second": 0.773,
+      "step": 16550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8034080152838614e-06,
+      "loss": 1.5341,
+      "step": 16600
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.8737,
+      "eval_samples_per_second": 388.387,
+      "eval_steps_per_second": 0.777,
+      "step": 16600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808839967137126e-06,
+      "loss": 1.5395,
+      "step": 16650
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.6886,
+      "eval_samples_per_second": 394.053,
+      "eval_steps_per_second": 0.788,
+      "step": 16650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8142719189903907e-06,
+      "loss": 1.5336,
+      "step": 16700
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.783,
+      "eval_samples_per_second": 391.145,
+      "eval_steps_per_second": 0.782,
+      "step": 16700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8197038708436552e-06,
+      "loss": 1.5345,
+      "step": 16750
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.5486,
+      "eval_samples_per_second": 398.45,
+      "eval_steps_per_second": 0.797,
+      "step": 16750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82513582269692e-06,
+      "loss": 1.5275,
+      "step": 16800
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.8087,
+      "eval_samples_per_second": 390.359,
+      "eval_steps_per_second": 0.781,
+      "step": 16800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8305677745501844e-06,
+      "loss": 1.5236,
+      "step": 16850
+    },
+    {
+      "epoch": 0.2,
+      "eval_runtime": 12.7855,
+      "eval_samples_per_second": 391.067,
+      "eval_steps_per_second": 0.782,
+      "step": 16850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359997264034494e-06,
+      "loss": 1.529,
+      "step": 16900
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.9,
+      "eval_samples_per_second": 387.595,
+      "eval_steps_per_second": 0.775,
+      "step": 16900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841431678256714e-06,
+      "loss": 1.5331,
+      "step": 16950
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.5117,
+      "eval_samples_per_second": 399.626,
+      "eval_steps_per_second": 0.799,
+      "step": 16950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8468636301099786e-06,
+      "loss": 1.5314,
+      "step": 17000
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.7376,
+      "eval_samples_per_second": 392.54,
+      "eval_steps_per_second": 0.785,
+      "step": 17000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8522955819632433e-06,
+      "loss": 1.5204,
+      "step": 17050
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.908,
+      "eval_samples_per_second": 387.356,
+      "eval_steps_per_second": 0.775,
+      "step": 17050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8577275338165078e-06,
+      "loss": 1.5198,
+      "step": 17100
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.8759,
+      "eval_samples_per_second": 388.323,
+      "eval_steps_per_second": 0.777,
+      "step": 17100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8631594856697726e-06,
+      "loss": 1.5198,
+      "step": 17150
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.9771,
+      "eval_samples_per_second": 385.295,
+      "eval_steps_per_second": 0.771,
+      "step": 17150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.868591437523037e-06,
+      "loss": 1.5208,
+      "step": 17200
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.8206,
+      "eval_samples_per_second": 389.998,
+      "eval_steps_per_second": 0.78,
+      "step": 17200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8740233893763018e-06,
+      "loss": 1.5198,
+      "step": 17250
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.9268,
+      "eval_samples_per_second": 386.792,
+      "eval_steps_per_second": 0.774,
+      "step": 17250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8794553412295663e-06,
+      "loss": 1.5208,
+      "step": 17300
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.5278,
+      "eval_samples_per_second": 399.111,
+      "eval_steps_per_second": 0.798,
+      "step": 17300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.884887293082831e-06,
+      "loss": 1.5232,
+      "step": 17350
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.7362,
+      "eval_samples_per_second": 392.583,
+      "eval_steps_per_second": 0.785,
+      "step": 17350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8903192449360958e-06,
+      "loss": 1.5174,
+      "step": 17400
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.8689,
+      "eval_samples_per_second": 388.533,
+      "eval_steps_per_second": 0.777,
+      "step": 17400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8957511967893603e-06,
+      "loss": 1.5195,
+      "step": 17450
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.6658,
+      "eval_samples_per_second": 394.763,
+      "eval_steps_per_second": 0.79,
+      "step": 17450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.901183148642625e-06,
+      "loss": 1.5173,
+      "step": 17500
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.8491,
+      "eval_samples_per_second": 389.132,
+      "eval_steps_per_second": 0.778,
+      "step": 17500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9066151004958895e-06,
+      "loss": 1.52,
+      "step": 17550
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.8899,
+      "eval_samples_per_second": 387.899,
+      "eval_steps_per_second": 0.776,
+      "step": 17550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9120470523491543e-06,
+      "loss": 1.5136,
+      "step": 17600
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.7068,
+      "eval_samples_per_second": 393.489,
+      "eval_steps_per_second": 0.787,
+      "step": 17600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9174790042024188e-06,
+      "loss": 1.5101,
+      "step": 17650
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.9537,
+      "eval_samples_per_second": 385.991,
+      "eval_steps_per_second": 0.772,
+      "step": 17650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9229109560556837e-06,
+      "loss": 1.5071,
+      "step": 17700
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 12.6865,
+      "eval_samples_per_second": 394.12,
+      "eval_steps_per_second": 0.788,
+      "step": 17700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9283429079089482e-06,
+      "loss": 1.5161,
+      "step": 17750
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.5339,
+      "eval_samples_per_second": 398.917,
+      "eval_steps_per_second": 0.798,
+      "step": 17750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9337748597622127e-06,
+      "loss": 1.5068,
+      "step": 17800
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.9759,
+      "eval_samples_per_second": 385.329,
+      "eval_steps_per_second": 0.771,
+      "step": 17800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9392068116154777e-06,
+      "loss": 1.5059,
+      "step": 17850
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.9011,
+      "eval_samples_per_second": 387.565,
+      "eval_steps_per_second": 0.775,
+      "step": 17850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.944638763468742e-06,
+      "loss": 1.5053,
+      "step": 17900
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.8419,
+      "eval_samples_per_second": 389.351,
+      "eval_steps_per_second": 0.779,
+      "step": 17900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9500707153220067e-06,
+      "loss": 1.5058,
+      "step": 17950
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.7588,
+      "eval_samples_per_second": 391.885,
+      "eval_steps_per_second": 0.784,
+      "step": 17950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9555026671752712e-06,
+      "loss": 1.5092,
+      "step": 18000
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.6321,
+      "eval_samples_per_second": 395.818,
+      "eval_steps_per_second": 0.792,
+      "step": 18000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.960934619028536e-06,
+      "loss": 1.5052,
+      "step": 18050
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.903,
+      "eval_samples_per_second": 387.508,
+      "eval_steps_per_second": 0.775,
+      "step": 18050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9663665708818007e-06,
+      "loss": 1.5004,
+      "step": 18100
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.9507,
+      "eval_samples_per_second": 386.08,
+      "eval_steps_per_second": 0.772,
+      "step": 18100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.971798522735065e-06,
+      "loss": 1.4985,
+      "step": 18150
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.9273,
+      "eval_samples_per_second": 386.779,
+      "eval_steps_per_second": 0.774,
+      "step": 18150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9772304745883297e-06,
+      "loss": 1.498,
+      "step": 18200
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.8977,
+      "eval_samples_per_second": 387.666,
+      "eval_steps_per_second": 0.775,
+      "step": 18200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9826624264415947e-06,
+      "loss": 1.506,
+      "step": 18250
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.9673,
+      "eval_samples_per_second": 385.586,
+      "eval_steps_per_second": 0.771,
+      "step": 18250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9880943782948596e-06,
+      "loss": 1.5007,
+      "step": 18300
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.9311,
+      "eval_samples_per_second": 386.664,
+      "eval_steps_per_second": 0.773,
+      "step": 18300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9935263301481237e-06,
+      "loss": 1.5028,
+      "step": 18350
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.551,
+      "eval_samples_per_second": 398.374,
+      "eval_steps_per_second": 0.797,
+      "step": 18350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989582820013886e-06,
+      "loss": 1.4993,
+      "step": 18400
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.809,
+      "eval_samples_per_second": 390.351,
+      "eval_steps_per_second": 0.781,
+      "step": 18400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.004390233854653e-06,
+      "loss": 1.5006,
+      "step": 18450
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.8445,
+      "eval_samples_per_second": 389.27,
+      "eval_steps_per_second": 0.779,
+      "step": 18450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009822185707918e-06,
+      "loss": 1.4994,
+      "step": 18500
+    },
+    {
+      "epoch": 0.22,
+      "eval_runtime": 12.8438,
+      "eval_samples_per_second": 389.294,
+      "eval_steps_per_second": 0.779,
+      "step": 18500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.015254137561182e-06,
+      "loss": 1.4966,
+      "step": 18550
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.7679,
+      "eval_samples_per_second": 391.608,
+      "eval_steps_per_second": 0.783,
+      "step": 18550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.020686089414447e-06,
+      "loss": 1.4964,
+      "step": 18600
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.903,
+      "eval_samples_per_second": 387.507,
+      "eval_steps_per_second": 0.775,
+      "step": 18600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0261180412677116e-06,
+      "loss": 1.4915,
+      "step": 18650
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8995,
+      "eval_samples_per_second": 387.612,
+      "eval_steps_per_second": 0.775,
+      "step": 18650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0315499931209766e-06,
+      "loss": 1.4926,
+      "step": 18700
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 13.0444,
+      "eval_samples_per_second": 383.306,
+      "eval_steps_per_second": 0.767,
+      "step": 18700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.036981944974241e-06,
+      "loss": 1.4915,
+      "step": 18750
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.5252,
+      "eval_samples_per_second": 399.197,
+      "eval_steps_per_second": 0.798,
+      "step": 18750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0424138968275056e-06,
+      "loss": 1.4921,
+      "step": 18800
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8903,
+      "eval_samples_per_second": 387.89,
+      "eval_steps_per_second": 0.776,
+      "step": 18800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0478458486807705e-06,
+      "loss": 1.4912,
+      "step": 18850
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.6586,
+      "eval_samples_per_second": 394.989,
+      "eval_steps_per_second": 0.79,
+      "step": 18850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.053277800534035e-06,
+      "loss": 1.4881,
+      "step": 18900
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.728,
+      "eval_samples_per_second": 392.834,
+      "eval_steps_per_second": 0.786,
+      "step": 18900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0587097523872996e-06,
+      "loss": 1.4911,
+      "step": 18950
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8643,
+      "eval_samples_per_second": 388.673,
+      "eval_steps_per_second": 0.777,
+      "step": 18950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.064141704240564e-06,
+      "loss": 1.4922,
+      "step": 19000
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8066,
+      "eval_samples_per_second": 390.425,
+      "eval_steps_per_second": 0.781,
+      "step": 19000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.069573656093829e-06,
+      "loss": 1.4869,
+      "step": 19050
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8314,
+      "eval_samples_per_second": 389.668,
+      "eval_steps_per_second": 0.779,
+      "step": 19050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0750056079470935e-06,
+      "loss": 1.4899,
+      "step": 19100
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.5924,
+      "eval_samples_per_second": 397.066,
+      "eval_steps_per_second": 0.794,
+      "step": 19100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.080437559800358e-06,
+      "loss": 1.4889,
+      "step": 19150
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.6686,
+      "eval_samples_per_second": 394.678,
+      "eval_steps_per_second": 0.789,
+      "step": 19150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.085869511653623e-06,
+      "loss": 1.4849,
+      "step": 19200
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.7731,
+      "eval_samples_per_second": 391.448,
+      "eval_steps_per_second": 0.783,
+      "step": 19200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0913014635068875e-06,
+      "loss": 1.4877,
+      "step": 19250
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8462,
+      "eval_samples_per_second": 389.222,
+      "eval_steps_per_second": 0.778,
+      "step": 19250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.096733415360152e-06,
+      "loss": 1.4785,
+      "step": 19300
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.9461,
+      "eval_samples_per_second": 386.216,
+      "eval_steps_per_second": 0.772,
+      "step": 19300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.1021653672134165e-06,
+      "loss": 1.4819,
+      "step": 19350
+    },
+    {
+      "epoch": 0.23,
+      "eval_runtime": 12.8044,
+      "eval_samples_per_second": 390.49,
+      "eval_steps_per_second": 0.781,
+      "step": 19350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1075973190666815e-06,
+      "loss": 1.4805,
+      "step": 19400
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.8737,
+      "eval_samples_per_second": 388.389,
+      "eval_steps_per_second": 0.777,
+      "step": 19400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.113029270919946e-06,
+      "loss": 1.4746,
+      "step": 19450
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.7035,
+      "eval_samples_per_second": 393.593,
+      "eval_steps_per_second": 0.787,
+      "step": 19450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1184612227732105e-06,
+      "loss": 1.4801,
+      "step": 19500
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.9047,
+      "eval_samples_per_second": 387.454,
+      "eval_steps_per_second": 0.775,
+      "step": 19500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.123893174626475e-06,
+      "loss": 1.4815,
+      "step": 19550
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.9074,
+      "eval_samples_per_second": 387.375,
+      "eval_steps_per_second": 0.775,
+      "step": 19550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.12932512647974e-06,
+      "loss": 1.4728,
+      "step": 19600
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 13.1034,
+      "eval_samples_per_second": 381.58,
+      "eval_steps_per_second": 0.763,
+      "step": 19600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.134757078333005e-06,
+      "loss": 1.4797,
+      "step": 19650
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.8377,
+      "eval_samples_per_second": 389.477,
+      "eval_steps_per_second": 0.779,
+      "step": 19650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.140189030186269e-06,
+      "loss": 1.4702,
+      "step": 19700
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.8905,
+      "eval_samples_per_second": 387.882,
+      "eval_steps_per_second": 0.776,
+      "step": 19700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.145620982039534e-06,
+      "loss": 1.4776,
+      "step": 19750
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.9452,
+      "eval_samples_per_second": 386.243,
+      "eval_steps_per_second": 0.772,
+      "step": 19750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1510529338927984e-06,
+      "loss": 1.4743,
+      "step": 19800
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.7748,
+      "eval_samples_per_second": 391.395,
+      "eval_steps_per_second": 0.783,
+      "step": 19800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1564848857460634e-06,
+      "loss": 1.4741,
+      "step": 19850
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.9063,
+      "eval_samples_per_second": 387.407,
+      "eval_steps_per_second": 0.775,
+      "step": 19850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.161916837599328e-06,
+      "loss": 1.4688,
+      "step": 19900
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.8625,
+      "eval_samples_per_second": 388.725,
+      "eval_steps_per_second": 0.777,
+      "step": 19900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1673487894525924e-06,
+      "loss": 1.4717,
+      "step": 19950
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.8448,
+      "eval_samples_per_second": 389.264,
+      "eval_steps_per_second": 0.779,
+      "step": 19950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1727807413058573e-06,
+      "loss": 1.4738,
+      "step": 20000
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.7434,
+      "eval_samples_per_second": 392.36,
+      "eval_steps_per_second": 0.785,
+      "step": 20000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.178212693159122e-06,
+      "loss": 1.4711,
+      "step": 20050
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.8673,
+      "eval_samples_per_second": 388.581,
+      "eval_steps_per_second": 0.777,
+      "step": 20050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.1836446450123864e-06,
+      "loss": 1.4715,
+      "step": 20100
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.638,
+      "eval_samples_per_second": 395.632,
+      "eval_steps_per_second": 0.791,
+      "step": 20100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.189076596865651e-06,
+      "loss": 1.4623,
+      "step": 20150
+    },
+    {
+      "epoch": 0.24,
+      "eval_runtime": 12.7061,
+      "eval_samples_per_second": 393.513,
+      "eval_steps_per_second": 0.787,
+      "step": 20150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.194508548718916e-06,
+      "loss": 1.467,
+      "step": 20200
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9782,
+      "eval_samples_per_second": 385.26,
+      "eval_steps_per_second": 0.771,
+      "step": 20200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.1999405005721803e-06,
+      "loss": 1.469,
+      "step": 20250
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9596,
+      "eval_samples_per_second": 385.815,
+      "eval_steps_per_second": 0.772,
+      "step": 20250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.205372452425445e-06,
+      "loss": 1.4695,
+      "step": 20300
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.5352,
+      "eval_samples_per_second": 398.877,
+      "eval_steps_per_second": 0.798,
+      "step": 20300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2108044042787094e-06,
+      "loss": 1.4611,
+      "step": 20350
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.5528,
+      "eval_samples_per_second": 398.317,
+      "eval_steps_per_second": 0.797,
+      "step": 20350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2162363561319743e-06,
+      "loss": 1.4726,
+      "step": 20400
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.7809,
+      "eval_samples_per_second": 391.209,
+      "eval_steps_per_second": 0.782,
+      "step": 20400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2216683079852393e-06,
+      "loss": 1.4593,
+      "step": 20450
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.5223,
+      "eval_samples_per_second": 399.288,
+      "eval_steps_per_second": 0.799,
+      "step": 20450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2271002598385033e-06,
+      "loss": 1.4645,
+      "step": 20500
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.7526,
+      "eval_samples_per_second": 392.078,
+      "eval_steps_per_second": 0.784,
+      "step": 20500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2325322116917683e-06,
+      "loss": 1.4617,
+      "step": 20550
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.6525,
+      "eval_samples_per_second": 395.18,
+      "eval_steps_per_second": 0.79,
+      "step": 20550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.237964163545033e-06,
+      "loss": 1.4611,
+      "step": 20600
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.5607,
+      "eval_samples_per_second": 398.066,
+      "eval_steps_per_second": 0.796,
+      "step": 20600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2433961153982977e-06,
+      "loss": 1.4628,
+      "step": 20650
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.7265,
+      "eval_samples_per_second": 392.881,
+      "eval_steps_per_second": 0.786,
+      "step": 20650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.248828067251562e-06,
+      "loss": 1.4591,
+      "step": 20700
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9462,
+      "eval_samples_per_second": 386.214,
+      "eval_steps_per_second": 0.772,
+      "step": 20700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2542600191048268e-06,
+      "loss": 1.4545,
+      "step": 20750
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.7316,
+      "eval_samples_per_second": 392.723,
+      "eval_steps_per_second": 0.785,
+      "step": 20750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2596919709580913e-06,
+      "loss": 1.4541,
+      "step": 20800
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9901,
+      "eval_samples_per_second": 384.908,
+      "eval_steps_per_second": 0.77,
+      "step": 20800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2651239228113562e-06,
+      "loss": 1.4545,
+      "step": 20850
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9445,
+      "eval_samples_per_second": 386.264,
+      "eval_steps_per_second": 0.773,
+      "step": 20850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2705558746646207e-06,
+      "loss": 1.4495,
+      "step": 20900
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9118,
+      "eval_samples_per_second": 387.243,
+      "eval_steps_per_second": 0.774,
+      "step": 20900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2759878265178853e-06,
+      "loss": 1.4545,
+      "step": 20950
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9117,
+      "eval_samples_per_second": 387.247,
+      "eval_steps_per_second": 0.774,
+      "step": 20950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.28141977837115e-06,
+      "loss": 1.46,
+      "step": 21000
+    },
+    {
+      "epoch": 0.25,
+      "eval_runtime": 12.9255,
+      "eval_samples_per_second": 386.833,
+      "eval_steps_per_second": 0.774,
+      "step": 21000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2868517302244147e-06,
+      "loss": 1.448,
+      "step": 21050
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.9426,
+      "eval_samples_per_second": 386.321,
+      "eval_steps_per_second": 0.773,
+      "step": 21050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2922836820776792e-06,
+      "loss": 1.4489,
+      "step": 21100
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.7477,
+      "eval_samples_per_second": 392.226,
+      "eval_steps_per_second": 0.784,
+      "step": 21100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2977156339309437e-06,
+      "loss": 1.4501,
+      "step": 21150
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.8406,
+      "eval_samples_per_second": 389.389,
+      "eval_steps_per_second": 0.779,
+      "step": 21150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3031475857842087e-06,
+      "loss": 1.4539,
+      "step": 21200
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 13.0893,
+      "eval_samples_per_second": 381.99,
+      "eval_steps_per_second": 0.764,
+      "step": 21200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.308579537637473e-06,
+      "loss": 1.4539,
+      "step": 21250
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.9304,
+      "eval_samples_per_second": 386.685,
+      "eval_steps_per_second": 0.773,
+      "step": 21250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3140114894907377e-06,
+      "loss": 1.4521,
+      "step": 21300
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.5063,
+      "eval_samples_per_second": 399.798,
+      "eval_steps_per_second": 0.8,
+      "step": 21300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3194434413440026e-06,
+      "loss": 1.4448,
+      "step": 21350
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.9983,
+      "eval_samples_per_second": 384.667,
+      "eval_steps_per_second": 0.769,
+      "step": 21350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.324875393197267e-06,
+      "loss": 1.4383,
+      "step": 21400
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.7125,
+      "eval_samples_per_second": 393.313,
+      "eval_steps_per_second": 0.787,
+      "step": 21400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3303073450505317e-06,
+      "loss": 1.4428,
+      "step": 21450
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.6463,
+      "eval_samples_per_second": 395.373,
+      "eval_steps_per_second": 0.791,
+      "step": 21450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.335739296903796e-06,
+      "loss": 1.4401,
+      "step": 21500
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.8016,
+      "eval_samples_per_second": 390.577,
+      "eval_steps_per_second": 0.781,
+      "step": 21500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.341171248757061e-06,
+      "loss": 1.4485,
+      "step": 21550
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.8297,
+      "eval_samples_per_second": 389.719,
+      "eval_steps_per_second": 0.779,
+      "step": 21550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3466032006103256e-06,
+      "loss": 1.4406,
+      "step": 21600
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.7994,
+      "eval_samples_per_second": 390.642,
+      "eval_steps_per_second": 0.781,
+      "step": 21600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.35203515246359e-06,
+      "loss": 1.4437,
+      "step": 21650
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.9689,
+      "eval_samples_per_second": 385.539,
+      "eval_steps_per_second": 0.771,
+      "step": 21650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3574671043168547e-06,
+      "loss": 1.4374,
+      "step": 21700
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.8464,
+      "eval_samples_per_second": 389.215,
+      "eval_steps_per_second": 0.778,
+      "step": 21700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3628990561701196e-06,
+      "loss": 1.4417,
+      "step": 21750
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 13.015,
+      "eval_samples_per_second": 384.171,
+      "eval_steps_per_second": 0.768,
+      "step": 21750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.3683310080233846e-06,
+      "loss": 1.4377,
+      "step": 21800
+    },
+    {
+      "epoch": 0.26,
+      "eval_runtime": 12.8975,
+      "eval_samples_per_second": 387.672,
+      "eval_steps_per_second": 0.775,
+      "step": 21800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.373762959876649e-06,
+      "loss": 1.4352,
+      "step": 21850
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8499,
+      "eval_samples_per_second": 389.109,
+      "eval_steps_per_second": 0.778,
+      "step": 21850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.3791949117299136e-06,
+      "loss": 1.4342,
+      "step": 21900
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8368,
+      "eval_samples_per_second": 389.505,
+      "eval_steps_per_second": 0.779,
+      "step": 21900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.384626863583178e-06,
+      "loss": 1.4355,
+      "step": 21950
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8632,
+      "eval_samples_per_second": 388.707,
+      "eval_steps_per_second": 0.777,
+      "step": 21950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.390058815436443e-06,
+      "loss": 1.4369,
+      "step": 22000
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8134,
+      "eval_samples_per_second": 390.216,
+      "eval_steps_per_second": 0.78,
+      "step": 22000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.3954907672897076e-06,
+      "loss": 1.4352,
+      "step": 22050
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.9725,
+      "eval_samples_per_second": 385.43,
+      "eval_steps_per_second": 0.771,
+      "step": 22050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.400922719142972e-06,
+      "loss": 1.4326,
+      "step": 22100
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.919,
+      "eval_samples_per_second": 387.026,
+      "eval_steps_per_second": 0.774,
+      "step": 22100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4063546709962366e-06,
+      "loss": 1.4304,
+      "step": 22150
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 13.0075,
+      "eval_samples_per_second": 384.394,
+      "eval_steps_per_second": 0.769,
+      "step": 22150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4117866228495015e-06,
+      "loss": 1.4308,
+      "step": 22200
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.9617,
+      "eval_samples_per_second": 385.751,
+      "eval_steps_per_second": 0.772,
+      "step": 22200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.417218574702766e-06,
+      "loss": 1.4321,
+      "step": 22250
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.7503,
+      "eval_samples_per_second": 392.148,
+      "eval_steps_per_second": 0.784,
+      "step": 22250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4226505265560306e-06,
+      "loss": 1.4283,
+      "step": 22300
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 13.0589,
+      "eval_samples_per_second": 382.879,
+      "eval_steps_per_second": 0.766,
+      "step": 22300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4280824784092955e-06,
+      "loss": 1.4285,
+      "step": 22350
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.7381,
+      "eval_samples_per_second": 392.522,
+      "eval_steps_per_second": 0.785,
+      "step": 22350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.43351443026256e-06,
+      "loss": 1.4244,
+      "step": 22400
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8914,
+      "eval_samples_per_second": 387.855,
+      "eval_steps_per_second": 0.776,
+      "step": 22400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4389463821158245e-06,
+      "loss": 1.4343,
+      "step": 22450
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.918,
+      "eval_samples_per_second": 387.058,
+      "eval_steps_per_second": 0.774,
+      "step": 22450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.444378333969089e-06,
+      "loss": 1.4252,
+      "step": 22500
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8776,
+      "eval_samples_per_second": 388.27,
+      "eval_steps_per_second": 0.777,
+      "step": 22500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.449810285822354e-06,
+      "loss": 1.4264,
+      "step": 22550
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.7353,
+      "eval_samples_per_second": 392.609,
+      "eval_steps_per_second": 0.785,
+      "step": 22550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.4552422376756185e-06,
+      "loss": 1.4252,
+      "step": 22600
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.6902,
+      "eval_samples_per_second": 394.005,
+      "eval_steps_per_second": 0.788,
+      "step": 22600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.460674189528883e-06,
+      "loss": 1.4197,
+      "step": 22650
+    },
+    {
+      "epoch": 0.27,
+      "eval_runtime": 12.8718,
+      "eval_samples_per_second": 388.446,
+      "eval_steps_per_second": 0.777,
+      "step": 22650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.466106141382148e-06,
+      "loss": 1.4272,
+      "step": 22700
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.8575,
+      "eval_samples_per_second": 388.879,
+      "eval_steps_per_second": 0.778,
+      "step": 22700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.4715380932354125e-06,
+      "loss": 1.425,
+      "step": 22750
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.5626,
+      "eval_samples_per_second": 398.006,
+      "eval_steps_per_second": 0.796,
+      "step": 22750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.4769700450886774e-06,
+      "loss": 1.4219,
+      "step": 22800
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.8831,
+      "eval_samples_per_second": 388.105,
+      "eval_steps_per_second": 0.776,
+      "step": 22800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.4824019969419415e-06,
+      "loss": 1.4235,
+      "step": 22850
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.9534,
+      "eval_samples_per_second": 386.0,
+      "eval_steps_per_second": 0.772,
+      "step": 22850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.4878339487952064e-06,
+      "loss": 1.4184,
+      "step": 22900
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.9577,
+      "eval_samples_per_second": 385.87,
+      "eval_steps_per_second": 0.772,
+      "step": 22900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.493265900648471e-06,
+      "loss": 1.4247,
+      "step": 22950
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 13.028,
+      "eval_samples_per_second": 383.789,
+      "eval_steps_per_second": 0.768,
+      "step": 22950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.498697852501736e-06,
+      "loss": 1.4163,
+      "step": 23000
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.8678,
+      "eval_samples_per_second": 388.567,
+      "eval_steps_per_second": 0.777,
+      "step": 23000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.504129804355e-06,
+      "loss": 1.421,
+      "step": 23050
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.669,
+      "eval_samples_per_second": 394.664,
+      "eval_steps_per_second": 0.789,
+      "step": 23050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.509561756208265e-06,
+      "loss": 1.4166,
+      "step": 23100
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.7452,
+      "eval_samples_per_second": 392.306,
+      "eval_steps_per_second": 0.785,
+      "step": 23100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.51499370806153e-06,
+      "loss": 1.4171,
+      "step": 23150
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.5193,
+      "eval_samples_per_second": 399.384,
+      "eval_steps_per_second": 0.799,
+      "step": 23150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5204256599147944e-06,
+      "loss": 1.4179,
+      "step": 23200
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.9404,
+      "eval_samples_per_second": 386.387,
+      "eval_steps_per_second": 0.773,
+      "step": 23200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.525857611768059e-06,
+      "loss": 1.4142,
+      "step": 23250
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.6759,
+      "eval_samples_per_second": 394.448,
+      "eval_steps_per_second": 0.789,
+      "step": 23250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5312895636213234e-06,
+      "loss": 1.4151,
+      "step": 23300
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.543,
+      "eval_samples_per_second": 398.63,
+      "eval_steps_per_second": 0.797,
+      "step": 23300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5367215154745883e-06,
+      "loss": 1.4138,
+      "step": 23350
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.5148,
+      "eval_samples_per_second": 399.528,
+      "eval_steps_per_second": 0.799,
+      "step": 23350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.542153467327853e-06,
+      "loss": 1.4102,
+      "step": 23400
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.9058,
+      "eval_samples_per_second": 387.421,
+      "eval_steps_per_second": 0.775,
+      "step": 23400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5475854191811174e-06,
+      "loss": 1.4103,
+      "step": 23450
+    },
+    {
+      "epoch": 0.28,
+      "eval_runtime": 12.952,
+      "eval_samples_per_second": 386.04,
+      "eval_steps_per_second": 0.772,
+      "step": 23450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.553017371034382e-06,
+      "loss": 1.4129,
+      "step": 23500
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.8729,
+      "eval_samples_per_second": 388.413,
+      "eval_steps_per_second": 0.777,
+      "step": 23500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.558449322887647e-06,
+      "loss": 1.4096,
+      "step": 23550
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.845,
+      "eval_samples_per_second": 389.256,
+      "eval_steps_per_second": 0.779,
+      "step": 23550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5638812747409113e-06,
+      "loss": 1.4042,
+      "step": 23600
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.8239,
+      "eval_samples_per_second": 389.896,
+      "eval_steps_per_second": 0.78,
+      "step": 23600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.569313226594176e-06,
+      "loss": 1.4024,
+      "step": 23650
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.7227,
+      "eval_samples_per_second": 392.998,
+      "eval_steps_per_second": 0.786,
+      "step": 23650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.574745178447441e-06,
+      "loss": 1.4042,
+      "step": 23700
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.8793,
+      "eval_samples_per_second": 388.221,
+      "eval_steps_per_second": 0.776,
+      "step": 23700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5801771303007053e-06,
+      "loss": 1.4072,
+      "step": 23750
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.9428,
+      "eval_samples_per_second": 386.314,
+      "eval_steps_per_second": 0.773,
+      "step": 23750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5856090821539702e-06,
+      "loss": 1.4046,
+      "step": 23800
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.6564,
+      "eval_samples_per_second": 395.056,
+      "eval_steps_per_second": 0.79,
+      "step": 23800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5910410340072343e-06,
+      "loss": 1.4069,
+      "step": 23850
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.6719,
+      "eval_samples_per_second": 394.573,
+      "eval_steps_per_second": 0.789,
+      "step": 23850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5964729858604993e-06,
+      "loss": 1.4049,
+      "step": 23900
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.7785,
+      "eval_samples_per_second": 391.282,
+      "eval_steps_per_second": 0.783,
+      "step": 23900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.601904937713764e-06,
+      "loss": 1.4031,
+      "step": 23950
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.773,
+      "eval_samples_per_second": 391.45,
+      "eval_steps_per_second": 0.783,
+      "step": 23950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6073368895670287e-06,
+      "loss": 1.4016,
+      "step": 24000
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.6098,
+      "eval_samples_per_second": 396.516,
+      "eval_steps_per_second": 0.793,
+      "step": 24000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6127688414202932e-06,
+      "loss": 1.4038,
+      "step": 24050
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.7582,
+      "eval_samples_per_second": 391.906,
+      "eval_steps_per_second": 0.784,
+      "step": 24050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6182007932735578e-06,
+      "loss": 1.4062,
+      "step": 24100
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.8556,
+      "eval_samples_per_second": 388.936,
+      "eval_steps_per_second": 0.778,
+      "step": 24100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6236327451268227e-06,
+      "loss": 1.4019,
+      "step": 24150
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.579,
+      "eval_samples_per_second": 397.488,
+      "eval_steps_per_second": 0.795,
+      "step": 24150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.629064696980087e-06,
+      "loss": 1.3966,
+      "step": 24200
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.5422,
+      "eval_samples_per_second": 398.656,
+      "eval_steps_per_second": 0.797,
+      "step": 24200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6344966488333517e-06,
+      "loss": 1.4031,
+      "step": 24250
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.8628,
+      "eval_samples_per_second": 388.719,
+      "eval_steps_per_second": 0.777,
+      "step": 24250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.6399286006866162e-06,
+      "loss": 1.3936,
+      "step": 24300
+    },
+    {
+      "epoch": 0.29,
+      "eval_runtime": 12.8936,
+      "eval_samples_per_second": 387.789,
+      "eval_steps_per_second": 0.776,
+      "step": 24300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.645360552539881e-06,
+      "loss": 1.4015,
+      "step": 24350
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8749,
+      "eval_samples_per_second": 388.351,
+      "eval_steps_per_second": 0.777,
+      "step": 24350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6507925043931457e-06,
+      "loss": 1.3976,
+      "step": 24400
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.6549,
+      "eval_samples_per_second": 395.104,
+      "eval_steps_per_second": 0.79,
+      "step": 24400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.65622445624641e-06,
+      "loss": 1.393,
+      "step": 24450
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8767,
+      "eval_samples_per_second": 388.297,
+      "eval_steps_per_second": 0.777,
+      "step": 24450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.661656408099675e-06,
+      "loss": 1.3982,
+      "step": 24500
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.5434,
+      "eval_samples_per_second": 398.615,
+      "eval_steps_per_second": 0.797,
+      "step": 24500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6670883599529397e-06,
+      "loss": 1.3864,
+      "step": 24550
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.4971,
+      "eval_samples_per_second": 400.093,
+      "eval_steps_per_second": 0.8,
+      "step": 24550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.672520311806204e-06,
+      "loss": 1.394,
+      "step": 24600
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.9504,
+      "eval_samples_per_second": 386.088,
+      "eval_steps_per_second": 0.772,
+      "step": 24600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6779522636594687e-06,
+      "loss": 1.3891,
+      "step": 24650
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.7511,
+      "eval_samples_per_second": 392.122,
+      "eval_steps_per_second": 0.784,
+      "step": 24650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6833842155127336e-06,
+      "loss": 1.3897,
+      "step": 24700
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.9742,
+      "eval_samples_per_second": 385.379,
+      "eval_steps_per_second": 0.771,
+      "step": 24700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.688816167365998e-06,
+      "loss": 1.391,
+      "step": 24750
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8107,
+      "eval_samples_per_second": 390.3,
+      "eval_steps_per_second": 0.781,
+      "step": 24750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6942481192192627e-06,
+      "loss": 1.3968,
+      "step": 24800
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.7573,
+      "eval_samples_per_second": 391.933,
+      "eval_steps_per_second": 0.784,
+      "step": 24800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.699680071072527e-06,
+      "loss": 1.3924,
+      "step": 24850
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8706,
+      "eval_samples_per_second": 388.482,
+      "eval_steps_per_second": 0.777,
+      "step": 24850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.705112022925792e-06,
+      "loss": 1.3898,
+      "step": 24900
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8094,
+      "eval_samples_per_second": 390.338,
+      "eval_steps_per_second": 0.781,
+      "step": 24900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.710543974779057e-06,
+      "loss": 1.3888,
+      "step": 24950
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8192,
+      "eval_samples_per_second": 390.038,
+      "eval_steps_per_second": 0.78,
+      "step": 24950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.715975926632321e-06,
+      "loss": 1.3919,
+      "step": 25000
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8793,
+      "eval_samples_per_second": 388.22,
+      "eval_steps_per_second": 0.776,
+      "step": 25000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.721407878485586e-06,
+      "loss": 1.3802,
+      "step": 25050
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.9333,
+      "eval_samples_per_second": 386.599,
+      "eval_steps_per_second": 0.773,
+      "step": 25050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7268398303388506e-06,
+      "loss": 1.3871,
+      "step": 25100
+    },
+    {
+      "epoch": 0.3,
+      "eval_runtime": 12.8608,
+      "eval_samples_per_second": 388.778,
+      "eval_steps_per_second": 0.778,
+      "step": 25100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7322717821921155e-06,
+      "loss": 1.3844,
+      "step": 25150
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.7646,
+      "eval_samples_per_second": 391.708,
+      "eval_steps_per_second": 0.783,
+      "step": 25150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7377037340453796e-06,
+      "loss": 1.3832,
+      "step": 25200
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.8391,
+      "eval_samples_per_second": 389.436,
+      "eval_steps_per_second": 0.779,
+      "step": 25200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7431356858986446e-06,
+      "loss": 1.3831,
+      "step": 25250
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.8733,
+      "eval_samples_per_second": 388.402,
+      "eval_steps_per_second": 0.777,
+      "step": 25250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.748567637751909e-06,
+      "loss": 1.3774,
+      "step": 25300
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.9633,
+      "eval_samples_per_second": 385.704,
+      "eval_steps_per_second": 0.771,
+      "step": 25300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.753999589605174e-06,
+      "loss": 1.383,
+      "step": 25350
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.5735,
+      "eval_samples_per_second": 397.661,
+      "eval_steps_per_second": 0.795,
+      "step": 25350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7594315414584385e-06,
+      "loss": 1.3769,
+      "step": 25400
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.7487,
+      "eval_samples_per_second": 392.198,
+      "eval_steps_per_second": 0.784,
+      "step": 25400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.764863493311703e-06,
+      "loss": 1.3797,
+      "step": 25450
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.912,
+      "eval_samples_per_second": 387.237,
+      "eval_steps_per_second": 0.774,
+      "step": 25450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.770295445164968e-06,
+      "loss": 1.3836,
+      "step": 25500
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.8517,
+      "eval_samples_per_second": 389.054,
+      "eval_steps_per_second": 0.778,
+      "step": 25500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7757273970182325e-06,
+      "loss": 1.3848,
+      "step": 25550
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.8986,
+      "eval_samples_per_second": 387.638,
+      "eval_steps_per_second": 0.775,
+      "step": 25550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.781159348871497e-06,
+      "loss": 1.377,
+      "step": 25600
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.72,
+      "eval_samples_per_second": 393.081,
+      "eval_steps_per_second": 0.786,
+      "step": 25600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7865913007247615e-06,
+      "loss": 1.3807,
+      "step": 25650
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.526,
+      "eval_samples_per_second": 399.169,
+      "eval_steps_per_second": 0.798,
+      "step": 25650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.7920232525780265e-06,
+      "loss": 1.3782,
+      "step": 25700
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.9036,
+      "eval_samples_per_second": 387.49,
+      "eval_steps_per_second": 0.775,
+      "step": 25700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.797455204431291e-06,
+      "loss": 1.3727,
+      "step": 25750
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.8091,
+      "eval_samples_per_second": 390.347,
+      "eval_steps_per_second": 0.781,
+      "step": 25750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8028871562845555e-06,
+      "loss": 1.3732,
+      "step": 25800
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.7545,
+      "eval_samples_per_second": 392.019,
+      "eval_steps_per_second": 0.784,
+      "step": 25800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.8083191081378204e-06,
+      "loss": 1.3749,
+      "step": 25850
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.7446,
+      "eval_samples_per_second": 392.323,
+      "eval_steps_per_second": 0.785,
+      "step": 25850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.813751059991085e-06,
+      "loss": 1.3694,
+      "step": 25900
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.9188,
+      "eval_samples_per_second": 387.033,
+      "eval_steps_per_second": 0.774,
+      "step": 25900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.81918301184435e-06,
+      "loss": 1.3693,
+      "step": 25950
+    },
+    {
+      "epoch": 0.31,
+      "eval_runtime": 12.9713,
+      "eval_samples_per_second": 385.465,
+      "eval_steps_per_second": 0.771,
+      "step": 25950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.824614963697614e-06,
+      "loss": 1.3715,
+      "step": 26000
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.8892,
+      "eval_samples_per_second": 387.923,
+      "eval_steps_per_second": 0.776,
+      "step": 26000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.830046915550879e-06,
+      "loss": 1.3732,
+      "step": 26050
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9041,
+      "eval_samples_per_second": 387.472,
+      "eval_steps_per_second": 0.775,
+      "step": 26050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8354788674041434e-06,
+      "loss": 1.3662,
+      "step": 26100
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.8983,
+      "eval_samples_per_second": 387.648,
+      "eval_steps_per_second": 0.775,
+      "step": 26100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8409108192574084e-06,
+      "loss": 1.3741,
+      "step": 26150
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.8025,
+      "eval_samples_per_second": 390.549,
+      "eval_steps_per_second": 0.781,
+      "step": 26150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8463427711106725e-06,
+      "loss": 1.3703,
+      "step": 26200
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9171,
+      "eval_samples_per_second": 387.084,
+      "eval_steps_per_second": 0.774,
+      "step": 26200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8517747229639374e-06,
+      "loss": 1.3651,
+      "step": 26250
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9224,
+      "eval_samples_per_second": 386.926,
+      "eval_steps_per_second": 0.774,
+      "step": 26250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8572066748172024e-06,
+      "loss": 1.3706,
+      "step": 26300
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.5381,
+      "eval_samples_per_second": 398.785,
+      "eval_steps_per_second": 0.798,
+      "step": 26300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.862638626670467e-06,
+      "loss": 1.3647,
+      "step": 26350
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.8079,
+      "eval_samples_per_second": 390.386,
+      "eval_steps_per_second": 0.781,
+      "step": 26350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8680705785237314e-06,
+      "loss": 1.3708,
+      "step": 26400
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.7793,
+      "eval_samples_per_second": 391.258,
+      "eval_steps_per_second": 0.783,
+      "step": 26400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.873502530376996e-06,
+      "loss": 1.3682,
+      "step": 26450
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9354,
+      "eval_samples_per_second": 386.536,
+      "eval_steps_per_second": 0.773,
+      "step": 26450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.878934482230261e-06,
+      "loss": 1.3638,
+      "step": 26500
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9729,
+      "eval_samples_per_second": 385.42,
+      "eval_steps_per_second": 0.771,
+      "step": 26500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8843664340835254e-06,
+      "loss": 1.3605,
+      "step": 26550
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.915,
+      "eval_samples_per_second": 387.145,
+      "eval_steps_per_second": 0.774,
+      "step": 26550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.88979838593679e-06,
+      "loss": 1.3646,
+      "step": 26600
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.8849,
+      "eval_samples_per_second": 388.051,
+      "eval_steps_per_second": 0.776,
+      "step": 26600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.8952303377900544e-06,
+      "loss": 1.3638,
+      "step": 26650
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.8419,
+      "eval_samples_per_second": 389.35,
+      "eval_steps_per_second": 0.779,
+      "step": 26650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9006622896433193e-06,
+      "loss": 1.3611,
+      "step": 26700
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9645,
+      "eval_samples_per_second": 385.669,
+      "eval_steps_per_second": 0.771,
+      "step": 26700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.906094241496584e-06,
+      "loss": 1.3617,
+      "step": 26750
+    },
+    {
+      "epoch": 0.32,
+      "eval_runtime": 12.9136,
+      "eval_samples_per_second": 387.189,
+      "eval_steps_per_second": 0.774,
+      "step": 26750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9115261933498484e-06,
+      "loss": 1.3631,
+      "step": 26800
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.5366,
+      "eval_samples_per_second": 398.831,
+      "eval_steps_per_second": 0.798,
+      "step": 26800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9169581452031133e-06,
+      "loss": 1.3611,
+      "step": 26850
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.8824,
+      "eval_samples_per_second": 388.126,
+      "eval_steps_per_second": 0.776,
+      "step": 26850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.922390097056378e-06,
+      "loss": 1.3582,
+      "step": 26900
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.7011,
+      "eval_samples_per_second": 393.667,
+      "eval_steps_per_second": 0.787,
+      "step": 26900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9278220489096423e-06,
+      "loss": 1.3538,
+      "step": 26950
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.9978,
+      "eval_samples_per_second": 384.68,
+      "eval_steps_per_second": 0.769,
+      "step": 26950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.933254000762907e-06,
+      "loss": 1.3561,
+      "step": 27000
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.5654,
+      "eval_samples_per_second": 397.92,
+      "eval_steps_per_second": 0.796,
+      "step": 27000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9386859526161718e-06,
+      "loss": 1.3594,
+      "step": 27050
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.9157,
+      "eval_samples_per_second": 387.125,
+      "eval_steps_per_second": 0.774,
+      "step": 27050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9441179044694363e-06,
+      "loss": 1.3593,
+      "step": 27100
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.7935,
+      "eval_samples_per_second": 390.824,
+      "eval_steps_per_second": 0.782,
+      "step": 27100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.949549856322701e-06,
+      "loss": 1.3602,
+      "step": 27150
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.8351,
+      "eval_samples_per_second": 389.558,
+      "eval_steps_per_second": 0.779,
+      "step": 27150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9549818081759657e-06,
+      "loss": 1.3579,
+      "step": 27200
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.8533,
+      "eval_samples_per_second": 389.006,
+      "eval_steps_per_second": 0.778,
+      "step": 27200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9604137600292303e-06,
+      "loss": 1.3522,
+      "step": 27250
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.9522,
+      "eval_samples_per_second": 386.035,
+      "eval_steps_per_second": 0.772,
+      "step": 27250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.965845711882495e-06,
+      "loss": 1.3456,
+      "step": 27300
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.9077,
+      "eval_samples_per_second": 387.367,
+      "eval_steps_per_second": 0.775,
+      "step": 27300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9712776637357593e-06,
+      "loss": 1.3517,
+      "step": 27350
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.6337,
+      "eval_samples_per_second": 395.768,
+      "eval_steps_per_second": 0.792,
+      "step": 27350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9767096155890242e-06,
+      "loss": 1.3516,
+      "step": 27400
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.579,
+      "eval_samples_per_second": 397.489,
+      "eval_steps_per_second": 0.795,
+      "step": 27400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9821415674422887e-06,
+      "loss": 1.3498,
+      "step": 27450
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.9165,
+      "eval_samples_per_second": 387.101,
+      "eval_steps_per_second": 0.774,
+      "step": 27450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9875735192955537e-06,
+      "loss": 1.351,
+      "step": 27500
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.8459,
+      "eval_samples_per_second": 389.231,
+      "eval_steps_per_second": 0.778,
+      "step": 27500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9930054711488178e-06,
+      "loss": 1.3486,
+      "step": 27550
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.8999,
+      "eval_samples_per_second": 387.601,
+      "eval_steps_per_second": 0.775,
+      "step": 27550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9984374230020827e-06,
+      "loss": 1.3522,
+      "step": 27600
+    },
+    {
+      "epoch": 0.33,
+      "eval_runtime": 12.9123,
+      "eval_samples_per_second": 387.227,
+      "eval_steps_per_second": 0.774,
+      "step": 27600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0038693748553477e-06,
+      "loss": 1.3477,
+      "step": 27650
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8228,
+      "eval_samples_per_second": 389.932,
+      "eval_steps_per_second": 0.78,
+      "step": 27650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.009301326708612e-06,
+      "loss": 1.3501,
+      "step": 27700
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.9929,
+      "eval_samples_per_second": 384.825,
+      "eval_steps_per_second": 0.77,
+      "step": 27700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0147332785618767e-06,
+      "loss": 1.3425,
+      "step": 27750
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8341,
+      "eval_samples_per_second": 389.587,
+      "eval_steps_per_second": 0.779,
+      "step": 27750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.020165230415141e-06,
+      "loss": 1.3425,
+      "step": 27800
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.9857,
+      "eval_samples_per_second": 385.04,
+      "eval_steps_per_second": 0.77,
+      "step": 27800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.025597182268406e-06,
+      "loss": 1.3515,
+      "step": 27850
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.856,
+      "eval_samples_per_second": 388.924,
+      "eval_steps_per_second": 0.778,
+      "step": 27850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0310291341216707e-06,
+      "loss": 1.3438,
+      "step": 27900
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8902,
+      "eval_samples_per_second": 387.892,
+      "eval_steps_per_second": 0.776,
+      "step": 27900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.036461085974935e-06,
+      "loss": 1.3491,
+      "step": 27950
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.9793,
+      "eval_samples_per_second": 385.23,
+      "eval_steps_per_second": 0.77,
+      "step": 27950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0418930378282e-06,
+      "loss": 1.3389,
+      "step": 28000
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8716,
+      "eval_samples_per_second": 388.453,
+      "eval_steps_per_second": 0.777,
+      "step": 28000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0473249896814646e-06,
+      "loss": 1.3413,
+      "step": 28050
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.551,
+      "eval_samples_per_second": 398.375,
+      "eval_steps_per_second": 0.797,
+      "step": 28050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0527569415347296e-06,
+      "loss": 1.3401,
+      "step": 28100
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.5567,
+      "eval_samples_per_second": 398.195,
+      "eval_steps_per_second": 0.796,
+      "step": 28100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0581888933879937e-06,
+      "loss": 1.3405,
+      "step": 28150
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8489,
+      "eval_samples_per_second": 389.139,
+      "eval_steps_per_second": 0.778,
+      "step": 28150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.0636208452412586e-06,
+      "loss": 1.3386,
+      "step": 28200
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8733,
+      "eval_samples_per_second": 388.402,
+      "eval_steps_per_second": 0.777,
+      "step": 28200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.069052797094523e-06,
+      "loss": 1.3426,
+      "step": 28250
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.9357,
+      "eval_samples_per_second": 386.527,
+      "eval_steps_per_second": 0.773,
+      "step": 28250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.074484748947788e-06,
+      "loss": 1.338,
+      "step": 28300
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.9931,
+      "eval_samples_per_second": 384.819,
+      "eval_steps_per_second": 0.77,
+      "step": 28300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.079916700801052e-06,
+      "loss": 1.3402,
+      "step": 28350
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.7088,
+      "eval_samples_per_second": 393.429,
+      "eval_steps_per_second": 0.787,
+      "step": 28350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.085348652654317e-06,
+      "loss": 1.3352,
+      "step": 28400
+    },
+    {
+      "epoch": 0.34,
+      "eval_runtime": 12.8822,
+      "eval_samples_per_second": 388.133,
+      "eval_steps_per_second": 0.776,
+      "step": 28400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.090780604507582e-06,
+      "loss": 1.335,
+      "step": 28450
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.9209,
+      "eval_samples_per_second": 386.969,
+      "eval_steps_per_second": 0.774,
+      "step": 28450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.0962125563608465e-06,
+      "loss": 1.3363,
+      "step": 28500
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.557,
+      "eval_samples_per_second": 398.184,
+      "eval_steps_per_second": 0.796,
+      "step": 28500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.101644508214111e-06,
+      "loss": 1.3364,
+      "step": 28550
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.585,
+      "eval_samples_per_second": 397.299,
+      "eval_steps_per_second": 0.795,
+      "step": 28550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1070764600673756e-06,
+      "loss": 1.3355,
+      "step": 28600
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.7635,
+      "eval_samples_per_second": 391.744,
+      "eval_steps_per_second": 0.783,
+      "step": 28600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1125084119206405e-06,
+      "loss": 1.3308,
+      "step": 28650
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.8449,
+      "eval_samples_per_second": 389.261,
+      "eval_steps_per_second": 0.779,
+      "step": 28650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.117940363773905e-06,
+      "loss": 1.3335,
+      "step": 28700
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.9865,
+      "eval_samples_per_second": 385.016,
+      "eval_steps_per_second": 0.77,
+      "step": 28700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1233723156271695e-06,
+      "loss": 1.3258,
+      "step": 28750
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.5373,
+      "eval_samples_per_second": 398.809,
+      "eval_steps_per_second": 0.798,
+      "step": 28750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.128804267480434e-06,
+      "loss": 1.3318,
+      "step": 28800
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.8332,
+      "eval_samples_per_second": 389.615,
+      "eval_steps_per_second": 0.779,
+      "step": 28800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1342362193336986e-06,
+      "loss": 1.3331,
+      "step": 28850
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.7824,
+      "eval_samples_per_second": 391.163,
+      "eval_steps_per_second": 0.782,
+      "step": 28850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1396681711869635e-06,
+      "loss": 1.3292,
+      "step": 28900
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.9903,
+      "eval_samples_per_second": 384.901,
+      "eval_steps_per_second": 0.77,
+      "step": 28900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.145100123040228e-06,
+      "loss": 1.3254,
+      "step": 28950
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.8722,
+      "eval_samples_per_second": 388.433,
+      "eval_steps_per_second": 0.777,
+      "step": 28950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1505320748934925e-06,
+      "loss": 1.3247,
+      "step": 29000
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.5328,
+      "eval_samples_per_second": 398.954,
+      "eval_steps_per_second": 0.798,
+      "step": 29000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.155964026746758e-06,
+      "loss": 1.3294,
+      "step": 29050
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.9878,
+      "eval_samples_per_second": 384.976,
+      "eval_steps_per_second": 0.77,
+      "step": 29050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.161395978600022e-06,
+      "loss": 1.3258,
+      "step": 29100
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.8789,
+      "eval_samples_per_second": 388.232,
+      "eval_steps_per_second": 0.776,
+      "step": 29100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1668279304532865e-06,
+      "loss": 1.3252,
+      "step": 29150
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.964,
+      "eval_samples_per_second": 385.685,
+      "eval_steps_per_second": 0.771,
+      "step": 29150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.172259882306551e-06,
+      "loss": 1.3301,
+      "step": 29200
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.9936,
+      "eval_samples_per_second": 384.803,
+      "eval_steps_per_second": 0.77,
+      "step": 29200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.1776918341598164e-06,
+      "loss": 1.3214,
+      "step": 29250
+    },
+    {
+      "epoch": 0.35,
+      "eval_runtime": 12.5802,
+      "eval_samples_per_second": 397.45,
+      "eval_steps_per_second": 0.795,
+      "step": 29250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1831237860130805e-06,
+      "loss": 1.3238,
+      "step": 29300
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.5474,
+      "eval_samples_per_second": 398.488,
+      "eval_steps_per_second": 0.797,
+      "step": 29300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.188555737866345e-06,
+      "loss": 1.3265,
+      "step": 29350
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 13.2123,
+      "eval_samples_per_second": 378.435,
+      "eval_steps_per_second": 0.757,
+      "step": 29350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1939876897196103e-06,
+      "loss": 1.3238,
+      "step": 29400
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.8133,
+      "eval_samples_per_second": 390.221,
+      "eval_steps_per_second": 0.78,
+      "step": 29400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.199419641572875e-06,
+      "loss": 1.3203,
+      "step": 29450
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.8891,
+      "eval_samples_per_second": 387.924,
+      "eval_steps_per_second": 0.776,
+      "step": 29450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.204851593426139e-06,
+      "loss": 1.3218,
+      "step": 29500
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.9872,
+      "eval_samples_per_second": 384.995,
+      "eval_steps_per_second": 0.77,
+      "step": 29500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2102835452794035e-06,
+      "loss": 1.3183,
+      "step": 29550
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.8422,
+      "eval_samples_per_second": 389.341,
+      "eval_steps_per_second": 0.779,
+      "step": 29550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215715497132669e-06,
+      "loss": 1.3214,
+      "step": 29600
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.9734,
+      "eval_samples_per_second": 385.404,
+      "eval_steps_per_second": 0.771,
+      "step": 29600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2211474489859333e-06,
+      "loss": 1.322,
+      "step": 29650
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.83,
+      "eval_samples_per_second": 389.712,
+      "eval_steps_per_second": 0.779,
+      "step": 29650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.226579400839198e-06,
+      "loss": 1.3222,
+      "step": 29700
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.9275,
+      "eval_samples_per_second": 386.771,
+      "eval_steps_per_second": 0.774,
+      "step": 29700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.232011352692462e-06,
+      "loss": 1.3185,
+      "step": 29750
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.9737,
+      "eval_samples_per_second": 385.396,
+      "eval_steps_per_second": 0.771,
+      "step": 29750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2374433045457273e-06,
+      "loss": 1.3166,
+      "step": 29800
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.9881,
+      "eval_samples_per_second": 384.966,
+      "eval_steps_per_second": 0.77,
+      "step": 29800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.242875256398992e-06,
+      "loss": 1.3145,
+      "step": 29850
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.6044,
+      "eval_samples_per_second": 396.688,
+      "eval_steps_per_second": 0.793,
+      "step": 29850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2483072082522563e-06,
+      "loss": 1.3192,
+      "step": 29900
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.8999,
+      "eval_samples_per_second": 387.599,
+      "eval_steps_per_second": 0.775,
+      "step": 29900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2537391601055213e-06,
+      "loss": 1.3238,
+      "step": 29950
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.9951,
+      "eval_samples_per_second": 384.762,
+      "eval_steps_per_second": 0.77,
+      "step": 29950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.259171111958786e-06,
+      "loss": 1.3149,
+      "step": 30000
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.8767,
+      "eval_samples_per_second": 388.299,
+      "eval_steps_per_second": 0.777,
+      "step": 30000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2646030638120503e-06,
+      "loss": 1.3217,
+      "step": 30050
+    },
+    {
+      "epoch": 0.36,
+      "eval_runtime": 12.8392,
+      "eval_samples_per_second": 389.432,
+      "eval_steps_per_second": 0.779,
+      "step": 30050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.270035015665315e-06,
+      "loss": 1.3169,
+      "step": 30100
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.5318,
+      "eval_samples_per_second": 398.984,
+      "eval_steps_per_second": 0.798,
+      "step": 30100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2754669675185798e-06,
+      "loss": 1.316,
+      "step": 30150
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.9993,
+      "eval_samples_per_second": 384.635,
+      "eval_steps_per_second": 0.769,
+      "step": 30150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2808989193718443e-06,
+      "loss": 1.3112,
+      "step": 30200
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.556,
+      "eval_samples_per_second": 398.217,
+      "eval_steps_per_second": 0.796,
+      "step": 30200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.286330871225109e-06,
+      "loss": 1.3161,
+      "step": 30250
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.9946,
+      "eval_samples_per_second": 384.776,
+      "eval_steps_per_second": 0.77,
+      "step": 30250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2917628230783737e-06,
+      "loss": 1.3106,
+      "step": 30300
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.9018,
+      "eval_samples_per_second": 387.544,
+      "eval_steps_per_second": 0.775,
+      "step": 30300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2971947749316383e-06,
+      "loss": 1.314,
+      "step": 30350
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.545,
+      "eval_samples_per_second": 398.566,
+      "eval_steps_per_second": 0.797,
+      "step": 30350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3026267267849028e-06,
+      "loss": 1.3112,
+      "step": 30400
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.9578,
+      "eval_samples_per_second": 385.869,
+      "eval_steps_per_second": 0.772,
+      "step": 30400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3080586786381673e-06,
+      "loss": 1.3073,
+      "step": 30450
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.5561,
+      "eval_samples_per_second": 398.212,
+      "eval_steps_per_second": 0.796,
+      "step": 30450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3134906304914322e-06,
+      "loss": 1.3083,
+      "step": 30500
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.8127,
+      "eval_samples_per_second": 390.237,
+      "eval_steps_per_second": 0.78,
+      "step": 30500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3189225823446967e-06,
+      "loss": 1.3105,
+      "step": 30550
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.8628,
+      "eval_samples_per_second": 388.718,
+      "eval_steps_per_second": 0.777,
+      "step": 30550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3243545341979613e-06,
+      "loss": 1.3031,
+      "step": 30600
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 13.2779,
+      "eval_samples_per_second": 376.567,
+      "eval_steps_per_second": 0.753,
+      "step": 30600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3297864860512258e-06,
+      "loss": 1.3028,
+      "step": 30650
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.8274,
+      "eval_samples_per_second": 389.79,
+      "eval_steps_per_second": 0.78,
+      "step": 30650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3352184379044907e-06,
+      "loss": 1.3088,
+      "step": 30700
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.5861,
+      "eval_samples_per_second": 397.263,
+      "eval_steps_per_second": 0.795,
+      "step": 30700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3406503897577552e-06,
+      "loss": 1.3034,
+      "step": 30750
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.7445,
+      "eval_samples_per_second": 392.325,
+      "eval_steps_per_second": 0.785,
+      "step": 30750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3460823416110197e-06,
+      "loss": 1.3096,
+      "step": 30800
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.9591,
+      "eval_samples_per_second": 385.828,
+      "eval_steps_per_second": 0.772,
+      "step": 30800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.3515142934642847e-06,
+      "loss": 1.3029,
+      "step": 30850
+    },
+    {
+      "epoch": 0.37,
+      "eval_runtime": 12.9056,
+      "eval_samples_per_second": 387.429,
+      "eval_steps_per_second": 0.775,
+      "step": 30850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.356946245317549e-06,
+      "loss": 1.2975,
+      "step": 30900
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 13.1367,
+      "eval_samples_per_second": 380.613,
+      "eval_steps_per_second": 0.761,
+      "step": 30900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3623781971708137e-06,
+      "loss": 1.2994,
+      "step": 30950
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.7685,
+      "eval_samples_per_second": 391.59,
+      "eval_steps_per_second": 0.783,
+      "step": 30950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3678101490240782e-06,
+      "loss": 1.2996,
+      "step": 31000
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.8538,
+      "eval_samples_per_second": 388.989,
+      "eval_steps_per_second": 0.778,
+      "step": 31000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.373242100877343e-06,
+      "loss": 1.3012,
+      "step": 31050
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.6552,
+      "eval_samples_per_second": 395.095,
+      "eval_steps_per_second": 0.79,
+      "step": 31050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3786740527306077e-06,
+      "loss": 1.2987,
+      "step": 31100
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.5245,
+      "eval_samples_per_second": 399.217,
+      "eval_steps_per_second": 0.798,
+      "step": 31100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.384106004583872e-06,
+      "loss": 1.2992,
+      "step": 31150
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.9577,
+      "eval_samples_per_second": 385.87,
+      "eval_steps_per_second": 0.772,
+      "step": 31150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3895379564371376e-06,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.8896,
+      "eval_samples_per_second": 387.909,
+      "eval_steps_per_second": 0.776,
+      "step": 31200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3949699082904016e-06,
+      "loss": 1.3,
+      "step": 31250
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.8804,
+      "eval_samples_per_second": 388.188,
+      "eval_steps_per_second": 0.776,
+      "step": 31250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.400401860143666e-06,
+      "loss": 1.2978,
+      "step": 31300
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.5376,
+      "eval_samples_per_second": 398.8,
+      "eval_steps_per_second": 0.798,
+      "step": 31300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4058338119969307e-06,
+      "loss": 1.2939,
+      "step": 31350
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.864,
+      "eval_samples_per_second": 388.681,
+      "eval_steps_per_second": 0.777,
+      "step": 31350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.411265763850196e-06,
+      "loss": 1.3005,
+      "step": 31400
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.7753,
+      "eval_samples_per_second": 391.38,
+      "eval_steps_per_second": 0.783,
+      "step": 31400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.41669771570346e-06,
+      "loss": 1.2903,
+      "step": 31450
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.7773,
+      "eval_samples_per_second": 391.318,
+      "eval_steps_per_second": 0.783,
+      "step": 31450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4221296675567246e-06,
+      "loss": 1.2932,
+      "step": 31500
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.8853,
+      "eval_samples_per_second": 388.04,
+      "eval_steps_per_second": 0.776,
+      "step": 31500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.427561619409989e-06,
+      "loss": 1.2945,
+      "step": 31550
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.9889,
+      "eval_samples_per_second": 384.944,
+      "eval_steps_per_second": 0.77,
+      "step": 31550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4329935712632545e-06,
+      "loss": 1.2933,
+      "step": 31600
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 13.003,
+      "eval_samples_per_second": 384.527,
+      "eval_steps_per_second": 0.769,
+      "step": 31600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.438425523116519e-06,
+      "loss": 1.2928,
+      "step": 31650
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.9097,
+      "eval_samples_per_second": 387.305,
+      "eval_steps_per_second": 0.775,
+      "step": 31650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.443857474969783e-06,
+      "loss": 1.2884,
+      "step": 31700
+    },
+    {
+      "epoch": 0.38,
+      "eval_runtime": 12.8981,
+      "eval_samples_per_second": 387.653,
+      "eval_steps_per_second": 0.775,
+      "step": 31700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4492894268230485e-06,
+      "loss": 1.2862,
+      "step": 31750
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8461,
+      "eval_samples_per_second": 389.223,
+      "eval_steps_per_second": 0.778,
+      "step": 31750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.454721378676313e-06,
+      "loss": 1.2942,
+      "step": 31800
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8403,
+      "eval_samples_per_second": 389.398,
+      "eval_steps_per_second": 0.779,
+      "step": 31800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4601533305295775e-06,
+      "loss": 1.2905,
+      "step": 31850
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 13.0054,
+      "eval_samples_per_second": 384.457,
+      "eval_steps_per_second": 0.769,
+      "step": 31850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4655852823828416e-06,
+      "loss": 1.2925,
+      "step": 31900
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8474,
+      "eval_samples_per_second": 389.183,
+      "eval_steps_per_second": 0.778,
+      "step": 31900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.471017234236107e-06,
+      "loss": 1.2902,
+      "step": 31950
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.967,
+      "eval_samples_per_second": 385.594,
+      "eval_steps_per_second": 0.771,
+      "step": 31950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4764491860893715e-06,
+      "loss": 1.2876,
+      "step": 32000
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8997,
+      "eval_samples_per_second": 387.606,
+      "eval_steps_per_second": 0.775,
+      "step": 32000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.481881137942636e-06,
+      "loss": 1.2942,
+      "step": 32050
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.5702,
+      "eval_samples_per_second": 397.765,
+      "eval_steps_per_second": 0.796,
+      "step": 32050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.487313089795901e-06,
+      "loss": 1.2914,
+      "step": 32100
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.9297,
+      "eval_samples_per_second": 386.708,
+      "eval_steps_per_second": 0.773,
+      "step": 32100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4927450416491655e-06,
+      "loss": 1.2835,
+      "step": 32150
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.5787,
+      "eval_samples_per_second": 397.496,
+      "eval_steps_per_second": 0.795,
+      "step": 32150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.49817699350243e-06,
+      "loss": 1.2829,
+      "step": 32200
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.7391,
+      "eval_samples_per_second": 392.494,
+      "eval_steps_per_second": 0.785,
+      "step": 32200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5036089453556945e-06,
+      "loss": 1.2775,
+      "step": 32250
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.956,
+      "eval_samples_per_second": 385.921,
+      "eval_steps_per_second": 0.772,
+      "step": 32250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5090408972089594e-06,
+      "loss": 1.2845,
+      "step": 32300
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 13.0798,
+      "eval_samples_per_second": 382.27,
+      "eval_steps_per_second": 0.765,
+      "step": 32300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.514472849062224e-06,
+      "loss": 1.2857,
+      "step": 32350
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.988,
+      "eval_samples_per_second": 384.97,
+      "eval_steps_per_second": 0.77,
+      "step": 32350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5199048009154885e-06,
+      "loss": 1.2831,
+      "step": 32400
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8519,
+      "eval_samples_per_second": 389.048,
+      "eval_steps_per_second": 0.778,
+      "step": 32400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5253367527687534e-06,
+      "loss": 1.2793,
+      "step": 32450
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8226,
+      "eval_samples_per_second": 389.937,
+      "eval_steps_per_second": 0.78,
+      "step": 32450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.530768704622018e-06,
+      "loss": 1.2824,
+      "step": 32500
+    },
+    {
+      "epoch": 0.39,
+      "eval_runtime": 12.8417,
+      "eval_samples_per_second": 389.358,
+      "eval_steps_per_second": 0.779,
+      "step": 32500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5362006564752824e-06,
+      "loss": 1.2776,
+      "step": 32550
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.8845,
+      "eval_samples_per_second": 388.064,
+      "eval_steps_per_second": 0.776,
+      "step": 32550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.541632608328547e-06,
+      "loss": 1.2833,
+      "step": 32600
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.5749,
+      "eval_samples_per_second": 397.617,
+      "eval_steps_per_second": 0.795,
+      "step": 32600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.547064560181812e-06,
+      "loss": 1.2786,
+      "step": 32650
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.8299,
+      "eval_samples_per_second": 389.716,
+      "eval_steps_per_second": 0.779,
+      "step": 32650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5524965120350764e-06,
+      "loss": 1.2784,
+      "step": 32700
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.7636,
+      "eval_samples_per_second": 391.74,
+      "eval_steps_per_second": 0.783,
+      "step": 32700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.557928463888341e-06,
+      "loss": 1.281,
+      "step": 32750
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.5733,
+      "eval_samples_per_second": 397.668,
+      "eval_steps_per_second": 0.795,
+      "step": 32750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5633604157416054e-06,
+      "loss": 1.2788,
+      "step": 32800
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.5825,
+      "eval_samples_per_second": 397.379,
+      "eval_steps_per_second": 0.795,
+      "step": 32800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5687923675948704e-06,
+      "loss": 1.2763,
+      "step": 32850
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.7856,
+      "eval_samples_per_second": 391.065,
+      "eval_steps_per_second": 0.782,
+      "step": 32850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.574224319448135e-06,
+      "loss": 1.2751,
+      "step": 32900
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.9278,
+      "eval_samples_per_second": 386.763,
+      "eval_steps_per_second": 0.774,
+      "step": 32900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5796562713013994e-06,
+      "loss": 1.2749,
+      "step": 32950
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.5664,
+      "eval_samples_per_second": 397.886,
+      "eval_steps_per_second": 0.796,
+      "step": 32950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5850882231546643e-06,
+      "loss": 1.2751,
+      "step": 33000
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.5711,
+      "eval_samples_per_second": 397.736,
+      "eval_steps_per_second": 0.795,
+      "step": 33000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.590520175007929e-06,
+      "loss": 1.278,
+      "step": 33050
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.9547,
+      "eval_samples_per_second": 385.962,
+      "eval_steps_per_second": 0.772,
+      "step": 33050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.5959521268611934e-06,
+      "loss": 1.2793,
+      "step": 33100
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.9222,
+      "eval_samples_per_second": 386.93,
+      "eval_steps_per_second": 0.774,
+      "step": 33100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.601384078714458e-06,
+      "loss": 1.2765,
+      "step": 33150
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.9318,
+      "eval_samples_per_second": 386.643,
+      "eval_steps_per_second": 0.773,
+      "step": 33150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.606816030567723e-06,
+      "loss": 1.2718,
+      "step": 33200
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.8372,
+      "eval_samples_per_second": 389.493,
+      "eval_steps_per_second": 0.779,
+      "step": 33200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6122479824209873e-06,
+      "loss": 1.2768,
+      "step": 33250
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.7393,
+      "eval_samples_per_second": 392.488,
+      "eval_steps_per_second": 0.785,
+      "step": 33250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.617679934274252e-06,
+      "loss": 1.2762,
+      "step": 33300
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.9247,
+      "eval_samples_per_second": 386.855,
+      "eval_steps_per_second": 0.774,
+      "step": 33300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.6231118861275172e-06,
+      "loss": 1.276,
+      "step": 33350
+    },
+    {
+      "epoch": 0.4,
+      "eval_runtime": 12.7913,
+      "eval_samples_per_second": 390.89,
+      "eval_steps_per_second": 0.782,
+      "step": 33350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6285438379807813e-06,
+      "loss": 1.2777,
+      "step": 33400
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 13.0126,
+      "eval_samples_per_second": 384.244,
+      "eval_steps_per_second": 0.768,
+      "step": 33400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.633975789834046e-06,
+      "loss": 1.2661,
+      "step": 33450
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.8779,
+      "eval_samples_per_second": 388.263,
+      "eval_steps_per_second": 0.777,
+      "step": 33450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6394077416873103e-06,
+      "loss": 1.2692,
+      "step": 33500
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.9052,
+      "eval_samples_per_second": 387.441,
+      "eval_steps_per_second": 0.775,
+      "step": 33500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6448396935405757e-06,
+      "loss": 1.273,
+      "step": 33550
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 13.0111,
+      "eval_samples_per_second": 384.286,
+      "eval_steps_per_second": 0.769,
+      "step": 33550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.65027164539384e-06,
+      "loss": 1.2694,
+      "step": 33600
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.5484,
+      "eval_samples_per_second": 398.456,
+      "eval_steps_per_second": 0.797,
+      "step": 33600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6557035972471043e-06,
+      "loss": 1.2699,
+      "step": 33650
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.8383,
+      "eval_samples_per_second": 389.459,
+      "eval_steps_per_second": 0.779,
+      "step": 33650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.661135549100369e-06,
+      "loss": 1.2694,
+      "step": 33700
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 13.0309,
+      "eval_samples_per_second": 383.705,
+      "eval_steps_per_second": 0.767,
+      "step": 33700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.666567500953634e-06,
+      "loss": 1.2684,
+      "step": 33750
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.9578,
+      "eval_samples_per_second": 385.869,
+      "eval_steps_per_second": 0.772,
+      "step": 33750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6719994528068987e-06,
+      "loss": 1.267,
+      "step": 33800
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.6157,
+      "eval_samples_per_second": 396.332,
+      "eval_steps_per_second": 0.793,
+      "step": 33800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.677431404660163e-06,
+      "loss": 1.2688,
+      "step": 33850
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.93,
+      "eval_samples_per_second": 386.697,
+      "eval_steps_per_second": 0.773,
+      "step": 33850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.682863356513428e-06,
+      "loss": 1.2703,
+      "step": 33900
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.8437,
+      "eval_samples_per_second": 389.297,
+      "eval_steps_per_second": 0.779,
+      "step": 33900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6882953083666927e-06,
+      "loss": 1.2685,
+      "step": 33950
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 13.007,
+      "eval_samples_per_second": 384.408,
+      "eval_steps_per_second": 0.769,
+      "step": 33950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.693727260219957e-06,
+      "loss": 1.262,
+      "step": 34000
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.9701,
+      "eval_samples_per_second": 385.502,
+      "eval_steps_per_second": 0.771,
+      "step": 34000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.6991592120732213e-06,
+      "loss": 1.2662,
+      "step": 34050
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 13.0187,
+      "eval_samples_per_second": 384.063,
+      "eval_steps_per_second": 0.768,
+      "step": 34050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.7045911639264866e-06,
+      "loss": 1.2682,
+      "step": 34100
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.867,
+      "eval_samples_per_second": 388.592,
+      "eval_steps_per_second": 0.777,
+      "step": 34100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.710023115779751e-06,
+      "loss": 1.26,
+      "step": 34150
+    },
+    {
+      "epoch": 0.41,
+      "eval_runtime": 12.6096,
+      "eval_samples_per_second": 396.523,
+      "eval_steps_per_second": 0.793,
+      "step": 34150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7154550676330157e-06,
+      "loss": 1.2614,
+      "step": 34200
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.6009,
+      "eval_samples_per_second": 396.797,
+      "eval_steps_per_second": 0.794,
+      "step": 34200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7208870194862806e-06,
+      "loss": 1.2606,
+      "step": 34250
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.8046,
+      "eval_samples_per_second": 390.485,
+      "eval_steps_per_second": 0.781,
+      "step": 34250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.726318971339545e-06,
+      "loss": 1.2566,
+      "step": 34300
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.9309,
+      "eval_samples_per_second": 386.67,
+      "eval_steps_per_second": 0.773,
+      "step": 34300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7317509231928096e-06,
+      "loss": 1.2642,
+      "step": 34350
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.7931,
+      "eval_samples_per_second": 390.835,
+      "eval_steps_per_second": 0.782,
+      "step": 34350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.737182875046074e-06,
+      "loss": 1.2627,
+      "step": 34400
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.5655,
+      "eval_samples_per_second": 397.916,
+      "eval_steps_per_second": 0.796,
+      "step": 34400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.742614826899339e-06,
+      "loss": 1.2623,
+      "step": 34450
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.5748,
+      "eval_samples_per_second": 397.622,
+      "eval_steps_per_second": 0.795,
+      "step": 34450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7480467787526036e-06,
+      "loss": 1.2591,
+      "step": 34500
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.9562,
+      "eval_samples_per_second": 385.915,
+      "eval_steps_per_second": 0.772,
+      "step": 34500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.753478730605868e-06,
+      "loss": 1.2603,
+      "step": 34550
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.8363,
+      "eval_samples_per_second": 389.52,
+      "eval_steps_per_second": 0.779,
+      "step": 34550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7589106824591326e-06,
+      "loss": 1.2591,
+      "step": 34600
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.5786,
+      "eval_samples_per_second": 397.5,
+      "eval_steps_per_second": 0.795,
+      "step": 34600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7643426343123976e-06,
+      "loss": 1.2605,
+      "step": 34650
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.9947,
+      "eval_samples_per_second": 384.773,
+      "eval_steps_per_second": 0.77,
+      "step": 34650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.769774586165662e-06,
+      "loss": 1.2631,
+      "step": 34700
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.6275,
+      "eval_samples_per_second": 395.962,
+      "eval_steps_per_second": 0.792,
+      "step": 34700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7752065380189266e-06,
+      "loss": 1.2533,
+      "step": 34750
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.7987,
+      "eval_samples_per_second": 390.666,
+      "eval_steps_per_second": 0.781,
+      "step": 34750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7806384898721915e-06,
+      "loss": 1.254,
+      "step": 34800
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.9731,
+      "eval_samples_per_second": 385.413,
+      "eval_steps_per_second": 0.771,
+      "step": 34800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.786070441725456e-06,
+      "loss": 1.2508,
+      "step": 34850
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.7317,
+      "eval_samples_per_second": 392.721,
+      "eval_steps_per_second": 0.785,
+      "step": 34850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.7915023935787206e-06,
+      "loss": 1.2578,
+      "step": 34900
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.8201,
+      "eval_samples_per_second": 390.012,
+      "eval_steps_per_second": 0.78,
+      "step": 34900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.796934345431985e-06,
+      "loss": 1.2576,
+      "step": 34950
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 13.0235,
+      "eval_samples_per_second": 383.922,
+      "eval_steps_per_second": 0.768,
+      "step": 34950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.80236629728525e-06,
+      "loss": 1.2532,
+      "step": 35000
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 12.9186,
+      "eval_samples_per_second": 387.037,
+      "eval_steps_per_second": 0.774,
+      "step": 35000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8077982491385145e-06,
+      "loss": 1.256,
+      "step": 35050
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.5865,
+      "eval_samples_per_second": 397.25,
+      "eval_steps_per_second": 0.794,
+      "step": 35050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.813230200991779e-06,
+      "loss": 1.2462,
+      "step": 35100
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.8865,
+      "eval_samples_per_second": 388.002,
+      "eval_steps_per_second": 0.776,
+      "step": 35100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8186621528450444e-06,
+      "loss": 1.2484,
+      "step": 35150
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.9192,
+      "eval_samples_per_second": 387.02,
+      "eval_steps_per_second": 0.774,
+      "step": 35150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8240941046983085e-06,
+      "loss": 1.2594,
+      "step": 35200
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 13.075,
+      "eval_samples_per_second": 382.411,
+      "eval_steps_per_second": 0.765,
+      "step": 35200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8295260565515735e-06,
+      "loss": 1.2546,
+      "step": 35250
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.9377,
+      "eval_samples_per_second": 386.469,
+      "eval_steps_per_second": 0.773,
+      "step": 35250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8349580084048375e-06,
+      "loss": 1.2509,
+      "step": 35300
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.6674,
+      "eval_samples_per_second": 394.715,
+      "eval_steps_per_second": 0.789,
+      "step": 35300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8403899602581025e-06,
+      "loss": 1.2563,
+      "step": 35350
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.9184,
+      "eval_samples_per_second": 387.044,
+      "eval_steps_per_second": 0.774,
+      "step": 35350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8458219121113674e-06,
+      "loss": 1.248,
+      "step": 35400
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.895,
+      "eval_samples_per_second": 387.748,
+      "eval_steps_per_second": 0.775,
+      "step": 35400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8512538639646315e-06,
+      "loss": 1.2543,
+      "step": 35450
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.578,
+      "eval_samples_per_second": 397.52,
+      "eval_steps_per_second": 0.795,
+      "step": 35450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8566858158178965e-06,
+      "loss": 1.2466,
+      "step": 35500
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.5817,
+      "eval_samples_per_second": 397.404,
+      "eval_steps_per_second": 0.795,
+      "step": 35500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.862117767671161e-06,
+      "loss": 1.2471,
+      "step": 35550
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.8878,
+      "eval_samples_per_second": 387.963,
+      "eval_steps_per_second": 0.776,
+      "step": 35550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8675497195244255e-06,
+      "loss": 1.2424,
+      "step": 35600
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.8667,
+      "eval_samples_per_second": 388.601,
+      "eval_steps_per_second": 0.777,
+      "step": 35600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8729816713776904e-06,
+      "loss": 1.248,
+      "step": 35650
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.9214,
+      "eval_samples_per_second": 386.956,
+      "eval_steps_per_second": 0.774,
+      "step": 35650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.878413623230955e-06,
+      "loss": 1.2445,
+      "step": 35700
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.8827,
+      "eval_samples_per_second": 388.117,
+      "eval_steps_per_second": 0.776,
+      "step": 35700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.8838455750842195e-06,
+      "loss": 1.2467,
+      "step": 35750
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 12.9419,
+      "eval_samples_per_second": 386.342,
+      "eval_steps_per_second": 0.773,
+      "step": 35750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.889277526937484e-06,
+      "loss": 1.2492,
+      "step": 35800
+    },
+    {
+      "epoch": 0.43,
+      "eval_runtime": 13.0525,
+      "eval_samples_per_second": 383.069,
+      "eval_steps_per_second": 0.766,
+      "step": 35800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.8947094787907485e-06,
+      "loss": 1.2438,
+      "step": 35850
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.7971,
+      "eval_samples_per_second": 390.714,
+      "eval_steps_per_second": 0.781,
+      "step": 35850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9001414306440134e-06,
+      "loss": 1.2442,
+      "step": 35900
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8535,
+      "eval_samples_per_second": 388.998,
+      "eval_steps_per_second": 0.778,
+      "step": 35900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.905573382497278e-06,
+      "loss": 1.2394,
+      "step": 35950
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8918,
+      "eval_samples_per_second": 387.842,
+      "eval_steps_per_second": 0.776,
+      "step": 35950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9110053343505425e-06,
+      "loss": 1.2425,
+      "step": 36000
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8984,
+      "eval_samples_per_second": 387.645,
+      "eval_steps_per_second": 0.775,
+      "step": 36000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.916437286203807e-06,
+      "loss": 1.2373,
+      "step": 36050
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 13.067,
+      "eval_samples_per_second": 382.643,
+      "eval_steps_per_second": 0.765,
+      "step": 36050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.921869238057072e-06,
+      "loss": 1.2485,
+      "step": 36100
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.5794,
+      "eval_samples_per_second": 397.475,
+      "eval_steps_per_second": 0.795,
+      "step": 36100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9273011899103364e-06,
+      "loss": 1.2386,
+      "step": 36150
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8528,
+      "eval_samples_per_second": 389.021,
+      "eval_steps_per_second": 0.778,
+      "step": 36150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.932733141763601e-06,
+      "loss": 1.2392,
+      "step": 36200
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8681,
+      "eval_samples_per_second": 388.557,
+      "eval_steps_per_second": 0.777,
+      "step": 36200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.938165093616866e-06,
+      "loss": 1.2391,
+      "step": 36250
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.9067,
+      "eval_samples_per_second": 387.397,
+      "eval_steps_per_second": 0.775,
+      "step": 36250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.94359704547013e-06,
+      "loss": 1.2375,
+      "step": 36300
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.7755,
+      "eval_samples_per_second": 391.375,
+      "eval_steps_per_second": 0.783,
+      "step": 36300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.949028997323395e-06,
+      "loss": 1.2476,
+      "step": 36350
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8658,
+      "eval_samples_per_second": 388.626,
+      "eval_steps_per_second": 0.777,
+      "step": 36350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9544609491766594e-06,
+      "loss": 1.2411,
+      "step": 36400
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8445,
+      "eval_samples_per_second": 389.27,
+      "eval_steps_per_second": 0.779,
+      "step": 36400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.959892901029924e-06,
+      "loss": 1.2355,
+      "step": 36450
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.9565,
+      "eval_samples_per_second": 385.905,
+      "eval_steps_per_second": 0.772,
+      "step": 36450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.965324852883189e-06,
+      "loss": 1.2333,
+      "step": 36500
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.8267,
+      "eval_samples_per_second": 389.81,
+      "eval_steps_per_second": 0.78,
+      "step": 36500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.970756804736453e-06,
+      "loss": 1.2405,
+      "step": 36550
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 13.0495,
+      "eval_samples_per_second": 383.156,
+      "eval_steps_per_second": 0.766,
+      "step": 36550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.976188756589719e-06,
+      "loss": 1.2356,
+      "step": 36600
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 12.6931,
+      "eval_samples_per_second": 393.916,
+      "eval_steps_per_second": 0.788,
+      "step": 36600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.981620708442983e-06,
+      "loss": 1.2348,
+      "step": 36650
+    },
+    {
+      "epoch": 0.44,
+      "eval_runtime": 13.0064,
+      "eval_samples_per_second": 384.425,
+      "eval_steps_per_second": 0.769,
+      "step": 36650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.987052660296247e-06,
+      "loss": 1.235,
+      "step": 36700
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.9974,
+      "eval_samples_per_second": 384.691,
+      "eval_steps_per_second": 0.769,
+      "step": 36700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.992484612149512e-06,
+      "loss": 1.2349,
+      "step": 36750
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 13.0834,
+      "eval_samples_per_second": 382.163,
+      "eval_steps_per_second": 0.764,
+      "step": 36750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997916564002777e-06,
+      "loss": 1.2381,
+      "step": 36800
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.601,
+      "eval_samples_per_second": 396.795,
+      "eval_steps_per_second": 0.794,
+      "step": 36800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.003348515856041e-06,
+      "loss": 1.2382,
+      "step": 36850
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.8719,
+      "eval_samples_per_second": 388.444,
+      "eval_steps_per_second": 0.777,
+      "step": 36850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.008780467709306e-06,
+      "loss": 1.2329,
+      "step": 36900
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.7544,
+      "eval_samples_per_second": 392.021,
+      "eval_steps_per_second": 0.784,
+      "step": 36900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.014212419562571e-06,
+      "loss": 1.2359,
+      "step": 36950
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 13.1481,
+      "eval_samples_per_second": 380.283,
+      "eval_steps_per_second": 0.761,
+      "step": 36950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.019644371415836e-06,
+      "loss": 1.2263,
+      "step": 37000
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.7859,
+      "eval_samples_per_second": 391.056,
+      "eval_steps_per_second": 0.782,
+      "step": 37000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0250763232691e-06,
+      "loss": 1.2363,
+      "step": 37050
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.9314,
+      "eval_samples_per_second": 386.656,
+      "eval_steps_per_second": 0.773,
+      "step": 37050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.030508275122364e-06,
+      "loss": 1.2357,
+      "step": 37100
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.89,
+      "eval_samples_per_second": 387.899,
+      "eval_steps_per_second": 0.776,
+      "step": 37100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.03594022697563e-06,
+      "loss": 1.2308,
+      "step": 37150
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.5726,
+      "eval_samples_per_second": 397.689,
+      "eval_steps_per_second": 0.795,
+      "step": 37150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.041372178828894e-06,
+      "loss": 1.2357,
+      "step": 37200
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.762,
+      "eval_samples_per_second": 391.789,
+      "eval_steps_per_second": 0.784,
+      "step": 37200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.046804130682158e-06,
+      "loss": 1.2314,
+      "step": 37250
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.9305,
+      "eval_samples_per_second": 386.682,
+      "eval_steps_per_second": 0.773,
+      "step": 37250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.052236082535423e-06,
+      "loss": 1.2255,
+      "step": 37300
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.9759,
+      "eval_samples_per_second": 385.331,
+      "eval_steps_per_second": 0.771,
+      "step": 37300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.057668034388688e-06,
+      "loss": 1.2231,
+      "step": 37350
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.5706,
+      "eval_samples_per_second": 397.753,
+      "eval_steps_per_second": 0.796,
+      "step": 37350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.063099986241953e-06,
+      "loss": 1.2268,
+      "step": 37400
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.7616,
+      "eval_samples_per_second": 391.801,
+      "eval_steps_per_second": 0.784,
+      "step": 37400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.068531938095217e-06,
+      "loss": 1.23,
+      "step": 37450
+    },
+    {
+      "epoch": 0.45,
+      "eval_runtime": 12.9774,
+      "eval_samples_per_second": 385.286,
+      "eval_steps_per_second": 0.771,
+      "step": 37450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.073963889948482e-06,
+      "loss": 1.2288,
+      "step": 37500
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.898,
+      "eval_samples_per_second": 387.656,
+      "eval_steps_per_second": 0.775,
+      "step": 37500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.079395841801747e-06,
+      "loss": 1.2281,
+      "step": 37550
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.9166,
+      "eval_samples_per_second": 387.098,
+      "eval_steps_per_second": 0.774,
+      "step": 37550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.084827793655011e-06,
+      "loss": 1.2243,
+      "step": 37600
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.9312,
+      "eval_samples_per_second": 386.662,
+      "eval_steps_per_second": 0.773,
+      "step": 37600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.090259745508276e-06,
+      "loss": 1.2258,
+      "step": 37650
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.924,
+      "eval_samples_per_second": 386.876,
+      "eval_steps_per_second": 0.774,
+      "step": 37650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.095691697361541e-06,
+      "loss": 1.2261,
+      "step": 37700
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.8455,
+      "eval_samples_per_second": 389.241,
+      "eval_steps_per_second": 0.778,
+      "step": 37700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.101123649214805e-06,
+      "loss": 1.2204,
+      "step": 37750
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.7312,
+      "eval_samples_per_second": 392.735,
+      "eval_steps_per_second": 0.785,
+      "step": 37750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.10655560106807e-06,
+      "loss": 1.2306,
+      "step": 37800
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 13.0094,
+      "eval_samples_per_second": 384.337,
+      "eval_steps_per_second": 0.769,
+      "step": 37800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.111987552921335e-06,
+      "loss": 1.2237,
+      "step": 37850
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.8796,
+      "eval_samples_per_second": 388.211,
+      "eval_steps_per_second": 0.776,
+      "step": 37850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.117419504774599e-06,
+      "loss": 1.222,
+      "step": 37900
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.6362,
+      "eval_samples_per_second": 395.687,
+      "eval_steps_per_second": 0.791,
+      "step": 37900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.122851456627864e-06,
+      "loss": 1.222,
+      "step": 37950
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 13.0629,
+      "eval_samples_per_second": 382.763,
+      "eval_steps_per_second": 0.766,
+      "step": 37950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.128283408481128e-06,
+      "loss": 1.2202,
+      "step": 38000
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.792,
+      "eval_samples_per_second": 390.868,
+      "eval_steps_per_second": 0.782,
+      "step": 38000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.133715360334393e-06,
+      "loss": 1.2204,
+      "step": 38050
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.8698,
+      "eval_samples_per_second": 388.506,
+      "eval_steps_per_second": 0.777,
+      "step": 38050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.139147312187658e-06,
+      "loss": 1.2215,
+      "step": 38100
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.8559,
+      "eval_samples_per_second": 388.925,
+      "eval_steps_per_second": 0.778,
+      "step": 38100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.144579264040922e-06,
+      "loss": 1.2202,
+      "step": 38150
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.8152,
+      "eval_samples_per_second": 390.163,
+      "eval_steps_per_second": 0.78,
+      "step": 38150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.150011215894187e-06,
+      "loss": 1.2191,
+      "step": 38200
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.9491,
+      "eval_samples_per_second": 386.127,
+      "eval_steps_per_second": 0.772,
+      "step": 38200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.155443167747452e-06,
+      "loss": 1.2159,
+      "step": 38250
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.8915,
+      "eval_samples_per_second": 387.853,
+      "eval_steps_per_second": 0.776,
+      "step": 38250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.160875119600716e-06,
+      "loss": 1.2187,
+      "step": 38300
+    },
+    {
+      "epoch": 0.46,
+      "eval_runtime": 12.7079,
+      "eval_samples_per_second": 393.456,
+      "eval_steps_per_second": 0.787,
+      "step": 38300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.166307071453981e-06,
+      "loss": 1.2201,
+      "step": 38350
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.5741,
+      "eval_samples_per_second": 397.643,
+      "eval_steps_per_second": 0.795,
+      "step": 38350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.171739023307246e-06,
+      "loss": 1.2206,
+      "step": 38400
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 13.062,
+      "eval_samples_per_second": 382.79,
+      "eval_steps_per_second": 0.766,
+      "step": 38400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.17717097516051e-06,
+      "loss": 1.2205,
+      "step": 38450
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.5827,
+      "eval_samples_per_second": 397.371,
+      "eval_steps_per_second": 0.795,
+      "step": 38450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.182602927013775e-06,
+      "loss": 1.2226,
+      "step": 38500
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.8156,
+      "eval_samples_per_second": 390.15,
+      "eval_steps_per_second": 0.78,
+      "step": 38500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.188034878867039e-06,
+      "loss": 1.2192,
+      "step": 38550
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.9661,
+      "eval_samples_per_second": 385.622,
+      "eval_steps_per_second": 0.771,
+      "step": 38550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.193466830720304e-06,
+      "loss": 1.2167,
+      "step": 38600
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.789,
+      "eval_samples_per_second": 390.962,
+      "eval_steps_per_second": 0.782,
+      "step": 38600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.198898782573569e-06,
+      "loss": 1.215,
+      "step": 38650
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.7676,
+      "eval_samples_per_second": 391.616,
+      "eval_steps_per_second": 0.783,
+      "step": 38650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.204330734426833e-06,
+      "loss": 1.2238,
+      "step": 38700
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.5918,
+      "eval_samples_per_second": 397.082,
+      "eval_steps_per_second": 0.794,
+      "step": 38700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.209762686280099e-06,
+      "loss": 1.217,
+      "step": 38750
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.7567,
+      "eval_samples_per_second": 391.951,
+      "eval_steps_per_second": 0.784,
+      "step": 38750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.215194638133363e-06,
+      "loss": 1.2078,
+      "step": 38800
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 13.0447,
+      "eval_samples_per_second": 383.297,
+      "eval_steps_per_second": 0.767,
+      "step": 38800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.220626589986627e-06,
+      "loss": 1.218,
+      "step": 38850
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.918,
+      "eval_samples_per_second": 387.057,
+      "eval_steps_per_second": 0.774,
+      "step": 38850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.226058541839892e-06,
+      "loss": 1.2151,
+      "step": 38900
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.866,
+      "eval_samples_per_second": 388.621,
+      "eval_steps_per_second": 0.777,
+      "step": 38900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.231490493693157e-06,
+      "loss": 1.2117,
+      "step": 38950
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.9514,
+      "eval_samples_per_second": 386.058,
+      "eval_steps_per_second": 0.772,
+      "step": 38950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.236922445546421e-06,
+      "loss": 1.2126,
+      "step": 39000
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.9754,
+      "eval_samples_per_second": 385.345,
+      "eval_steps_per_second": 0.771,
+      "step": 39000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.242354397399686e-06,
+      "loss": 1.2082,
+      "step": 39050
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 13.0139,
+      "eval_samples_per_second": 384.204,
+      "eval_steps_per_second": 0.768,
+      "step": 39050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.24778634925295e-06,
+      "loss": 1.2092,
+      "step": 39100
+    },
+    {
+      "epoch": 0.47,
+      "eval_runtime": 12.64,
+      "eval_samples_per_second": 395.57,
+      "eval_steps_per_second": 0.791,
+      "step": 39100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.253218301106216e-06,
+      "loss": 1.2099,
+      "step": 39150
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.8918,
+      "eval_samples_per_second": 387.844,
+      "eval_steps_per_second": 0.776,
+      "step": 39150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.25865025295948e-06,
+      "loss": 1.2092,
+      "step": 39200
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 13.0438,
+      "eval_samples_per_second": 383.325,
+      "eval_steps_per_second": 0.767,
+      "step": 39200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.264082204812744e-06,
+      "loss": 1.21,
+      "step": 39250
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.6645,
+      "eval_samples_per_second": 394.804,
+      "eval_steps_per_second": 0.79,
+      "step": 39250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.26951415666601e-06,
+      "loss": 1.2068,
+      "step": 39300
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.9153,
+      "eval_samples_per_second": 387.137,
+      "eval_steps_per_second": 0.774,
+      "step": 39300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.274946108519274e-06,
+      "loss": 1.2103,
+      "step": 39350
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.5519,
+      "eval_samples_per_second": 398.346,
+      "eval_steps_per_second": 0.797,
+      "step": 39350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.280378060372538e-06,
+      "loss": 1.2061,
+      "step": 39400
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 13.2352,
+      "eval_samples_per_second": 377.781,
+      "eval_steps_per_second": 0.756,
+      "step": 39400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.285810012225803e-06,
+      "loss": 1.211,
+      "step": 39450
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.8349,
+      "eval_samples_per_second": 389.562,
+      "eval_steps_per_second": 0.779,
+      "step": 39450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.291241964079068e-06,
+      "loss": 1.2126,
+      "step": 39500
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.834,
+      "eval_samples_per_second": 389.59,
+      "eval_steps_per_second": 0.779,
+      "step": 39500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.296673915932333e-06,
+      "loss": 1.2098,
+      "step": 39550
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.6194,
+      "eval_samples_per_second": 396.214,
+      "eval_steps_per_second": 0.792,
+      "step": 39550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.302105867785597e-06,
+      "loss": 1.2105,
+      "step": 39600
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.6532,
+      "eval_samples_per_second": 395.157,
+      "eval_steps_per_second": 0.79,
+      "step": 39600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.307537819638862e-06,
+      "loss": 1.2115,
+      "step": 39650
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.731,
+      "eval_samples_per_second": 392.741,
+      "eval_steps_per_second": 0.785,
+      "step": 39650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.312969771492127e-06,
+      "loss": 1.2034,
+      "step": 39700
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 13.1298,
+      "eval_samples_per_second": 380.813,
+      "eval_steps_per_second": 0.762,
+      "step": 39700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.318401723345391e-06,
+      "loss": 1.2037,
+      "step": 39750
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.6826,
+      "eval_samples_per_second": 394.24,
+      "eval_steps_per_second": 0.788,
+      "step": 39750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.323833675198656e-06,
+      "loss": 1.2063,
+      "step": 39800
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.914,
+      "eval_samples_per_second": 387.176,
+      "eval_steps_per_second": 0.774,
+      "step": 39800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.329265627051921e-06,
+      "loss": 1.2088,
+      "step": 39850
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.6807,
+      "eval_samples_per_second": 394.301,
+      "eval_steps_per_second": 0.789,
+      "step": 39850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.334697578905185e-06,
+      "loss": 1.204,
+      "step": 39900
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.9408,
+      "eval_samples_per_second": 386.374,
+      "eval_steps_per_second": 0.773,
+      "step": 39900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.34012953075845e-06,
+      "loss": 1.2056,
+      "step": 39950
+    },
+    {
+      "epoch": 0.48,
+      "eval_runtime": 12.9654,
+      "eval_samples_per_second": 385.642,
+      "eval_steps_per_second": 0.771,
+      "step": 39950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.345561482611715e-06,
+      "loss": 1.2039,
+      "step": 40000
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.9092,
+      "eval_samples_per_second": 387.322,
+      "eval_steps_per_second": 0.775,
+      "step": 40000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.350993434464979e-06,
+      "loss": 1.2014,
+      "step": 40050
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 13.0102,
+      "eval_samples_per_second": 384.313,
+      "eval_steps_per_second": 0.769,
+      "step": 40050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.356425386318244e-06,
+      "loss": 1.2002,
+      "step": 40100
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8319,
+      "eval_samples_per_second": 389.655,
+      "eval_steps_per_second": 0.779,
+      "step": 40100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.361857338171508e-06,
+      "loss": 1.2033,
+      "step": 40150
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.9662,
+      "eval_samples_per_second": 385.617,
+      "eval_steps_per_second": 0.771,
+      "step": 40150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.367289290024773e-06,
+      "loss": 1.2014,
+      "step": 40200
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.6207,
+      "eval_samples_per_second": 396.175,
+      "eval_steps_per_second": 0.792,
+      "step": 40200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.372721241878038e-06,
+      "loss": 1.1998,
+      "step": 40250
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8403,
+      "eval_samples_per_second": 389.399,
+      "eval_steps_per_second": 0.779,
+      "step": 40250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.378153193731302e-06,
+      "loss": 1.2008,
+      "step": 40300
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8947,
+      "eval_samples_per_second": 387.755,
+      "eval_steps_per_second": 0.776,
+      "step": 40300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.383585145584567e-06,
+      "loss": 1.2016,
+      "step": 40350
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8447,
+      "eval_samples_per_second": 389.266,
+      "eval_steps_per_second": 0.779,
+      "step": 40350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.389017097437832e-06,
+      "loss": 1.1984,
+      "step": 40400
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.9611,
+      "eval_samples_per_second": 385.77,
+      "eval_steps_per_second": 0.772,
+      "step": 40400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.394449049291096e-06,
+      "loss": 1.2073,
+      "step": 40450
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.7753,
+      "eval_samples_per_second": 391.38,
+      "eval_steps_per_second": 0.783,
+      "step": 40450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.399881001144361e-06,
+      "loss": 1.2059,
+      "step": 40500
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.7423,
+      "eval_samples_per_second": 392.393,
+      "eval_steps_per_second": 0.785,
+      "step": 40500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.405312952997626e-06,
+      "loss": 1.2015,
+      "step": 40550
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8786,
+      "eval_samples_per_second": 388.241,
+      "eval_steps_per_second": 0.776,
+      "step": 40550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.41074490485089e-06,
+      "loss": 1.1971,
+      "step": 40600
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8399,
+      "eval_samples_per_second": 389.411,
+      "eval_steps_per_second": 0.779,
+      "step": 40600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.416176856704155e-06,
+      "loss": 1.1967,
+      "step": 40650
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.5739,
+      "eval_samples_per_second": 397.648,
+      "eval_steps_per_second": 0.795,
+      "step": 40650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.421608808557419e-06,
+      "loss": 1.1955,
+      "step": 40700
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 13.0063,
+      "eval_samples_per_second": 384.43,
+      "eval_steps_per_second": 0.769,
+      "step": 40700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.427040760410684e-06,
+      "loss": 1.1892,
+      "step": 40750
+    },
+    {
+      "epoch": 0.49,
+      "eval_runtime": 12.8881,
+      "eval_samples_per_second": 387.955,
+      "eval_steps_per_second": 0.776,
+      "step": 40750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.432472712263949e-06,
+      "loss": 1.1954,
+      "step": 40800
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.6107,
+      "eval_samples_per_second": 396.49,
+      "eval_steps_per_second": 0.793,
+      "step": 40800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.437904664117213e-06,
+      "loss": 1.1956,
+      "step": 40850
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.7914,
+      "eval_samples_per_second": 390.889,
+      "eval_steps_per_second": 0.782,
+      "step": 40850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4433366159704785e-06,
+      "loss": 1.1962,
+      "step": 40900
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.5634,
+      "eval_samples_per_second": 397.981,
+      "eval_steps_per_second": 0.796,
+      "step": 40900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.448768567823743e-06,
+      "loss": 1.1968,
+      "step": 40950
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.8123,
+      "eval_samples_per_second": 390.249,
+      "eval_steps_per_second": 0.78,
+      "step": 40950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.454200519677007e-06,
+      "loss": 1.1976,
+      "step": 41000
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.7912,
+      "eval_samples_per_second": 390.892,
+      "eval_steps_per_second": 0.782,
+      "step": 41000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.459632471530272e-06,
+      "loss": 1.1989,
+      "step": 41050
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.5765,
+      "eval_samples_per_second": 397.567,
+      "eval_steps_per_second": 0.795,
+      "step": 41050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4650644233835366e-06,
+      "loss": 1.1956,
+      "step": 41100
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 13.022,
+      "eval_samples_per_second": 383.965,
+      "eval_steps_per_second": 0.768,
+      "step": 41100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.470496375236801e-06,
+      "loss": 1.1945,
+      "step": 41150
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.6095,
+      "eval_samples_per_second": 396.528,
+      "eval_steps_per_second": 0.793,
+      "step": 41150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.475928327090066e-06,
+      "loss": 1.1899,
+      "step": 41200
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.9348,
+      "eval_samples_per_second": 386.554,
+      "eval_steps_per_second": 0.773,
+      "step": 41200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.48136027894333e-06,
+      "loss": 1.1906,
+      "step": 41250
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.716,
+      "eval_samples_per_second": 393.205,
+      "eval_steps_per_second": 0.786,
+      "step": 41250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4867922307965955e-06,
+      "loss": 1.1863,
+      "step": 41300
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.9609,
+      "eval_samples_per_second": 385.774,
+      "eval_steps_per_second": 0.772,
+      "step": 41300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4922241826498596e-06,
+      "loss": 1.1904,
+      "step": 41350
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.9386,
+      "eval_samples_per_second": 386.44,
+      "eval_steps_per_second": 0.773,
+      "step": 41350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.497656134503124e-06,
+      "loss": 1.1903,
+      "step": 41400
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.8541,
+      "eval_samples_per_second": 388.98,
+      "eval_steps_per_second": 0.778,
+      "step": 41400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5030880863563894e-06,
+      "loss": 1.1907,
+      "step": 41450
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.877,
+      "eval_samples_per_second": 388.289,
+      "eval_steps_per_second": 0.777,
+      "step": 41450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5085200382096535e-06,
+      "loss": 1.1867,
+      "step": 41500
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.616,
+      "eval_samples_per_second": 396.322,
+      "eval_steps_per_second": 0.793,
+      "step": 41500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5139519900629185e-06,
+      "loss": 1.1872,
+      "step": 41550
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.7098,
+      "eval_samples_per_second": 393.398,
+      "eval_steps_per_second": 0.787,
+      "step": 41550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5193839419161826e-06,
+      "loss": 1.1924,
+      "step": 41600
+    },
+    {
+      "epoch": 0.5,
+      "eval_runtime": 12.9254,
+      "eval_samples_per_second": 386.834,
+      "eval_steps_per_second": 0.774,
+      "step": 41600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5248158937694475e-06,
+      "loss": 1.1854,
+      "step": 41650
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.5876,
+      "eval_samples_per_second": 397.217,
+      "eval_steps_per_second": 0.794,
+      "step": 41650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5302478456227124e-06,
+      "loss": 1.1887,
+      "step": 41700
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.9656,
+      "eval_samples_per_second": 385.636,
+      "eval_steps_per_second": 0.771,
+      "step": 41700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5356797974759765e-06,
+      "loss": 1.1892,
+      "step": 41750
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.8335,
+      "eval_samples_per_second": 389.605,
+      "eval_steps_per_second": 0.779,
+      "step": 41750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5411117493292415e-06,
+      "loss": 1.186,
+      "step": 41800
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.6087,
+      "eval_samples_per_second": 396.553,
+      "eval_steps_per_second": 0.793,
+      "step": 41800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.546543701182506e-06,
+      "loss": 1.1848,
+      "step": 41850
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.8019,
+      "eval_samples_per_second": 390.566,
+      "eval_steps_per_second": 0.781,
+      "step": 41850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5519756530357705e-06,
+      "loss": 1.1873,
+      "step": 41900
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.7601,
+      "eval_samples_per_second": 391.846,
+      "eval_steps_per_second": 0.784,
+      "step": 41900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5574076048890354e-06,
+      "loss": 1.1828,
+      "step": 41950
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.9987,
+      "eval_samples_per_second": 384.655,
+      "eval_steps_per_second": 0.769,
+      "step": 41950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5628395567423e-06,
+      "loss": 1.1857,
+      "step": 42000
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.9021,
+      "eval_samples_per_second": 387.532,
+      "eval_steps_per_second": 0.775,
+      "step": 42000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5682715085955645e-06,
+      "loss": 1.1827,
+      "step": 42050
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 13.03,
+      "eval_samples_per_second": 383.731,
+      "eval_steps_per_second": 0.767,
+      "step": 42050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.573703460448829e-06,
+      "loss": 1.182,
+      "step": 42100
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.837,
+      "eval_samples_per_second": 389.5,
+      "eval_steps_per_second": 0.779,
+      "step": 42100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5791354123020935e-06,
+      "loss": 1.1853,
+      "step": 42150
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.7772,
+      "eval_samples_per_second": 391.323,
+      "eval_steps_per_second": 0.783,
+      "step": 42150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5845673641553584e-06,
+      "loss": 1.1807,
+      "step": 42200
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 13.0006,
+      "eval_samples_per_second": 384.597,
+      "eval_steps_per_second": 0.769,
+      "step": 42200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.589999316008623e-06,
+      "loss": 1.1855,
+      "step": 42250
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.898,
+      "eval_samples_per_second": 387.656,
+      "eval_steps_per_second": 0.775,
+      "step": 42250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5954312678618875e-06,
+      "loss": 1.1825,
+      "step": 42300
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.5729,
+      "eval_samples_per_second": 397.681,
+      "eval_steps_per_second": 0.795,
+      "step": 42300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.600863219715152e-06,
+      "loss": 1.1816,
+      "step": 42350
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.7126,
+      "eval_samples_per_second": 393.311,
+      "eval_steps_per_second": 0.787,
+      "step": 42350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.606295171568417e-06,
+      "loss": 1.1865,
+      "step": 42400
+    },
+    {
+      "epoch": 0.51,
+      "eval_runtime": 12.7977,
+      "eval_samples_per_second": 390.696,
+      "eval_steps_per_second": 0.781,
+      "step": 42400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6117271234216814e-06,
+      "loss": 1.1855,
+      "step": 42450
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.8803,
+      "eval_samples_per_second": 388.189,
+      "eval_steps_per_second": 0.776,
+      "step": 42450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.617159075274946e-06,
+      "loss": 1.1783,
+      "step": 42500
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.5909,
+      "eval_samples_per_second": 397.112,
+      "eval_steps_per_second": 0.794,
+      "step": 42500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.622591027128211e-06,
+      "loss": 1.1812,
+      "step": 42550
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.7727,
+      "eval_samples_per_second": 391.46,
+      "eval_steps_per_second": 0.783,
+      "step": 42550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.628022978981475e-06,
+      "loss": 1.1833,
+      "step": 42600
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.9531,
+      "eval_samples_per_second": 386.008,
+      "eval_steps_per_second": 0.772,
+      "step": 42600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.63345493083474e-06,
+      "loss": 1.1814,
+      "step": 42650
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.8334,
+      "eval_samples_per_second": 389.609,
+      "eval_steps_per_second": 0.779,
+      "step": 42650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.638886882688005e-06,
+      "loss": 1.1808,
+      "step": 42700
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.9504,
+      "eval_samples_per_second": 386.088,
+      "eval_steps_per_second": 0.772,
+      "step": 42700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.644318834541269e-06,
+      "loss": 1.1841,
+      "step": 42750
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.8688,
+      "eval_samples_per_second": 388.537,
+      "eval_steps_per_second": 0.777,
+      "step": 42750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.649750786394534e-06,
+      "loss": 1.1791,
+      "step": 42800
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.8733,
+      "eval_samples_per_second": 388.401,
+      "eval_steps_per_second": 0.777,
+      "step": 42800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.655182738247798e-06,
+      "loss": 1.1738,
+      "step": 42850
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.8776,
+      "eval_samples_per_second": 388.271,
+      "eval_steps_per_second": 0.777,
+      "step": 42850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.660614690101063e-06,
+      "loss": 1.1725,
+      "step": 42900
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.9506,
+      "eval_samples_per_second": 386.081,
+      "eval_steps_per_second": 0.772,
+      "step": 42900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.666046641954328e-06,
+      "loss": 1.1754,
+      "step": 42950
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.8954,
+      "eval_samples_per_second": 387.736,
+      "eval_steps_per_second": 0.775,
+      "step": 42950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.671478593807592e-06,
+      "loss": 1.176,
+      "step": 43000
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.5476,
+      "eval_samples_per_second": 398.482,
+      "eval_steps_per_second": 0.797,
+      "step": 43000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676910545660857e-06,
+      "loss": 1.1738,
+      "step": 43050
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.9226,
+      "eval_samples_per_second": 386.918,
+      "eval_steps_per_second": 0.774,
+      "step": 43050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.682342497514122e-06,
+      "loss": 1.1713,
+      "step": 43100
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.9559,
+      "eval_samples_per_second": 385.924,
+      "eval_steps_per_second": 0.772,
+      "step": 43100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.687774449367386e-06,
+      "loss": 1.1793,
+      "step": 43150
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 13.1045,
+      "eval_samples_per_second": 381.548,
+      "eval_steps_per_second": 0.763,
+      "step": 43150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.693206401220651e-06,
+      "loss": 1.1743,
+      "step": 43200
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.923,
+      "eval_samples_per_second": 386.906,
+      "eval_steps_per_second": 0.774,
+      "step": 43200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.698638353073916e-06,
+      "loss": 1.1738,
+      "step": 43250
+    },
+    {
+      "epoch": 0.52,
+      "eval_runtime": 12.924,
+      "eval_samples_per_second": 386.877,
+      "eval_steps_per_second": 0.774,
+      "step": 43250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70407030492718e-06,
+      "loss": 1.171,
+      "step": 43300
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.9911,
+      "eval_samples_per_second": 384.878,
+      "eval_steps_per_second": 0.77,
+      "step": 43300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709502256780445e-06,
+      "loss": 1.1694,
+      "step": 43350
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 13.0248,
+      "eval_samples_per_second": 383.884,
+      "eval_steps_per_second": 0.768,
+      "step": 43350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714934208633709e-06,
+      "loss": 1.1741,
+      "step": 43400
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.7601,
+      "eval_samples_per_second": 391.847,
+      "eval_steps_per_second": 0.784,
+      "step": 43400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720366160486975e-06,
+      "loss": 1.1695,
+      "step": 43450
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.8385,
+      "eval_samples_per_second": 389.453,
+      "eval_steps_per_second": 0.779,
+      "step": 43450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.725798112340239e-06,
+      "loss": 1.1761,
+      "step": 43500
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 13.0845,
+      "eval_samples_per_second": 382.132,
+      "eval_steps_per_second": 0.764,
+      "step": 43500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731230064193503e-06,
+      "loss": 1.174,
+      "step": 43550
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.6145,
+      "eval_samples_per_second": 396.37,
+      "eval_steps_per_second": 0.793,
+      "step": 43550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736662016046769e-06,
+      "loss": 1.169,
+      "step": 43600
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.91,
+      "eval_samples_per_second": 387.296,
+      "eval_steps_per_second": 0.775,
+      "step": 43600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742093967900033e-06,
+      "loss": 1.1683,
+      "step": 43650
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.8271,
+      "eval_samples_per_second": 389.8,
+      "eval_steps_per_second": 0.78,
+      "step": 43650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747525919753298e-06,
+      "loss": 1.1703,
+      "step": 43700
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 13.0151,
+      "eval_samples_per_second": 384.169,
+      "eval_steps_per_second": 0.768,
+      "step": 43700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.752957871606562e-06,
+      "loss": 1.171,
+      "step": 43750
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.6919,
+      "eval_samples_per_second": 393.951,
+      "eval_steps_per_second": 0.788,
+      "step": 43750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.758389823459827e-06,
+      "loss": 1.1684,
+      "step": 43800
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.9123,
+      "eval_samples_per_second": 387.227,
+      "eval_steps_per_second": 0.774,
+      "step": 43800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.763821775313092e-06,
+      "loss": 1.1738,
+      "step": 43850
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.9574,
+      "eval_samples_per_second": 385.879,
+      "eval_steps_per_second": 0.772,
+      "step": 43850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.769253727166356e-06,
+      "loss": 1.1682,
+      "step": 43900
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.6175,
+      "eval_samples_per_second": 396.274,
+      "eval_steps_per_second": 0.793,
+      "step": 43900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.77468567901962e-06,
+      "loss": 1.1607,
+      "step": 43950
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.9893,
+      "eval_samples_per_second": 384.931,
+      "eval_steps_per_second": 0.77,
+      "step": 43950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780117630872886e-06,
+      "loss": 1.1673,
+      "step": 44000
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.6154,
+      "eval_samples_per_second": 396.342,
+      "eval_steps_per_second": 0.793,
+      "step": 44000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78554958272615e-06,
+      "loss": 1.1685,
+      "step": 44050
+    },
+    {
+      "epoch": 0.53,
+      "eval_runtime": 12.8338,
+      "eval_samples_per_second": 389.596,
+      "eval_steps_per_second": 0.779,
+      "step": 44050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.790981534579415e-06,
+      "loss": 1.1696,
+      "step": 44100
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.9837,
+      "eval_samples_per_second": 385.098,
+      "eval_steps_per_second": 0.77,
+      "step": 44100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79641348643268e-06,
+      "loss": 1.1669,
+      "step": 44150
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.5994,
+      "eval_samples_per_second": 396.844,
+      "eval_steps_per_second": 0.794,
+      "step": 44150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801845438285944e-06,
+      "loss": 1.1684,
+      "step": 44200
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.9672,
+      "eval_samples_per_second": 385.589,
+      "eval_steps_per_second": 0.771,
+      "step": 44200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.807277390139209e-06,
+      "loss": 1.1658,
+      "step": 44250
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.9272,
+      "eval_samples_per_second": 386.781,
+      "eval_steps_per_second": 0.774,
+      "step": 44250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.812709341992473e-06,
+      "loss": 1.1649,
+      "step": 44300
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.8595,
+      "eval_samples_per_second": 388.819,
+      "eval_steps_per_second": 0.778,
+      "step": 44300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818141293845738e-06,
+      "loss": 1.1678,
+      "step": 44350
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.6307,
+      "eval_samples_per_second": 395.862,
+      "eval_steps_per_second": 0.792,
+      "step": 44350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.823573245699003e-06,
+      "loss": 1.1676,
+      "step": 44400
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.6146,
+      "eval_samples_per_second": 396.366,
+      "eval_steps_per_second": 0.793,
+      "step": 44400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.829005197552267e-06,
+      "loss": 1.1635,
+      "step": 44450
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.8255,
+      "eval_samples_per_second": 389.847,
+      "eval_steps_per_second": 0.78,
+      "step": 44450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.834437149405532e-06,
+      "loss": 1.1625,
+      "step": 44500
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 13.0663,
+      "eval_samples_per_second": 382.665,
+      "eval_steps_per_second": 0.765,
+      "step": 44500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.839869101258797e-06,
+      "loss": 1.1631,
+      "step": 44550
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.9434,
+      "eval_samples_per_second": 386.296,
+      "eval_steps_per_second": 0.773,
+      "step": 44550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.845301053112061e-06,
+      "loss": 1.1613,
+      "step": 44600
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.738,
+      "eval_samples_per_second": 392.527,
+      "eval_steps_per_second": 0.785,
+      "step": 44600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.850733004965326e-06,
+      "loss": 1.1619,
+      "step": 44650
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.8562,
+      "eval_samples_per_second": 388.917,
+      "eval_steps_per_second": 0.778,
+      "step": 44650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.856164956818591e-06,
+      "loss": 1.1559,
+      "step": 44700
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.8603,
+      "eval_samples_per_second": 388.794,
+      "eval_steps_per_second": 0.778,
+      "step": 44700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.861596908671855e-06,
+      "loss": 1.1649,
+      "step": 44750
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 13.032,
+      "eval_samples_per_second": 383.669,
+      "eval_steps_per_second": 0.767,
+      "step": 44750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.86702886052512e-06,
+      "loss": 1.1619,
+      "step": 44800
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.827,
+      "eval_samples_per_second": 389.802,
+      "eval_steps_per_second": 0.78,
+      "step": 44800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.872460812378384e-06,
+      "loss": 1.16,
+      "step": 44850
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.9317,
+      "eval_samples_per_second": 386.648,
+      "eval_steps_per_second": 0.773,
+      "step": 44850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.877892764231649e-06,
+      "loss": 1.1584,
+      "step": 44900
+    },
+    {
+      "epoch": 0.54,
+      "eval_runtime": 12.6425,
+      "eval_samples_per_second": 395.49,
+      "eval_steps_per_second": 0.791,
+      "step": 44900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.883324716084914e-06,
+      "loss": 1.1596,
+      "step": 44950
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.6039,
+      "eval_samples_per_second": 396.701,
+      "eval_steps_per_second": 0.793,
+      "step": 44950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.888756667938178e-06,
+      "loss": 1.1577,
+      "step": 45000
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.7881,
+      "eval_samples_per_second": 390.99,
+      "eval_steps_per_second": 0.782,
+      "step": 45000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.894188619791443e-06,
+      "loss": 1.16,
+      "step": 45050
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.9527,
+      "eval_samples_per_second": 386.019,
+      "eval_steps_per_second": 0.772,
+      "step": 45050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.899620571644708e-06,
+      "loss": 1.1565,
+      "step": 45100
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.5918,
+      "eval_samples_per_second": 397.084,
+      "eval_steps_per_second": 0.794,
+      "step": 45100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.905052523497972e-06,
+      "loss": 1.1585,
+      "step": 45150
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.6022,
+      "eval_samples_per_second": 396.755,
+      "eval_steps_per_second": 0.794,
+      "step": 45150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.910484475351237e-06,
+      "loss": 1.1581,
+      "step": 45200
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.6217,
+      "eval_samples_per_second": 396.142,
+      "eval_steps_per_second": 0.792,
+      "step": 45200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915916427204502e-06,
+      "loss": 1.155,
+      "step": 45250
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.6422,
+      "eval_samples_per_second": 395.501,
+      "eval_steps_per_second": 0.791,
+      "step": 45250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.921348379057766e-06,
+      "loss": 1.1581,
+      "step": 45300
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.9609,
+      "eval_samples_per_second": 385.775,
+      "eval_steps_per_second": 0.772,
+      "step": 45300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.926780330911031e-06,
+      "loss": 1.1562,
+      "step": 45350
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.9281,
+      "eval_samples_per_second": 386.755,
+      "eval_steps_per_second": 0.774,
+      "step": 45350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.932212282764296e-06,
+      "loss": 1.1583,
+      "step": 45400
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.6169,
+      "eval_samples_per_second": 396.294,
+      "eval_steps_per_second": 0.793,
+      "step": 45400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.93764423461756e-06,
+      "loss": 1.1552,
+      "step": 45450
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.602,
+      "eval_samples_per_second": 396.762,
+      "eval_steps_per_second": 0.794,
+      "step": 45450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.943076186470825e-06,
+      "loss": 1.1609,
+      "step": 45500
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 13.0064,
+      "eval_samples_per_second": 384.425,
+      "eval_steps_per_second": 0.769,
+      "step": 45500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.948508138324089e-06,
+      "loss": 1.1572,
+      "step": 45550
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.9469,
+      "eval_samples_per_second": 386.192,
+      "eval_steps_per_second": 0.772,
+      "step": 45550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.953940090177355e-06,
+      "loss": 1.1564,
+      "step": 45600
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.9014,
+      "eval_samples_per_second": 387.556,
+      "eval_steps_per_second": 0.775,
+      "step": 45600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.959372042030619e-06,
+      "loss": 1.154,
+      "step": 45650
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.9785,
+      "eval_samples_per_second": 385.254,
+      "eval_steps_per_second": 0.771,
+      "step": 45650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.964803993883883e-06,
+      "loss": 1.1563,
+      "step": 45700
+    },
+    {
+      "epoch": 0.55,
+      "eval_runtime": 12.954,
+      "eval_samples_per_second": 385.982,
+      "eval_steps_per_second": 0.772,
+      "step": 45700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.970235945737148e-06,
+      "loss": 1.1539,
+      "step": 45750
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.8404,
+      "eval_samples_per_second": 389.395,
+      "eval_steps_per_second": 0.779,
+      "step": 45750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.975667897590413e-06,
+      "loss": 1.1492,
+      "step": 45800
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.8873,
+      "eval_samples_per_second": 387.98,
+      "eval_steps_per_second": 0.776,
+      "step": 45800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.981099849443678e-06,
+      "loss": 1.1487,
+      "step": 45850
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.705,
+      "eval_samples_per_second": 393.547,
+      "eval_steps_per_second": 0.787,
+      "step": 45850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.986531801296942e-06,
+      "loss": 1.149,
+      "step": 45900
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.9068,
+      "eval_samples_per_second": 387.394,
+      "eval_steps_per_second": 0.775,
+      "step": 45900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.991963753150207e-06,
+      "loss": 1.152,
+      "step": 45950
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.924,
+      "eval_samples_per_second": 386.879,
+      "eval_steps_per_second": 0.774,
+      "step": 45950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.997395705003472e-06,
+      "loss": 1.1542,
+      "step": 46000
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 13.0138,
+      "eval_samples_per_second": 384.208,
+      "eval_steps_per_second": 0.768,
+      "step": 46000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.002827656856736e-06,
+      "loss": 1.1497,
+      "step": 46050
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.8961,
+      "eval_samples_per_second": 387.714,
+      "eval_steps_per_second": 0.775,
+      "step": 46050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.00825960871e-06,
+      "loss": 1.1508,
+      "step": 46100
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.6392,
+      "eval_samples_per_second": 395.595,
+      "eval_steps_per_second": 0.791,
+      "step": 46100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.013691560563266e-06,
+      "loss": 1.1531,
+      "step": 46150
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.7429,
+      "eval_samples_per_second": 392.374,
+      "eval_steps_per_second": 0.785,
+      "step": 46150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.01912351241653e-06,
+      "loss": 1.1441,
+      "step": 46200
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.756,
+      "eval_samples_per_second": 391.973,
+      "eval_steps_per_second": 0.784,
+      "step": 46200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.024555464269795e-06,
+      "loss": 1.1494,
+      "step": 46250
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 13.0977,
+      "eval_samples_per_second": 381.747,
+      "eval_steps_per_second": 0.763,
+      "step": 46250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.02998741612306e-06,
+      "loss": 1.1445,
+      "step": 46300
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 13.0173,
+      "eval_samples_per_second": 384.103,
+      "eval_steps_per_second": 0.768,
+      "step": 46300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.035419367976324e-06,
+      "loss": 1.1472,
+      "step": 46350
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.9872,
+      "eval_samples_per_second": 384.994,
+      "eval_steps_per_second": 0.77,
+      "step": 46350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.040851319829589e-06,
+      "loss": 1.1524,
+      "step": 46400
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 13.0413,
+      "eval_samples_per_second": 383.397,
+      "eval_steps_per_second": 0.767,
+      "step": 46400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.046283271682853e-06,
+      "loss": 1.1428,
+      "step": 46450
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.9208,
+      "eval_samples_per_second": 386.972,
+      "eval_steps_per_second": 0.774,
+      "step": 46450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.051715223536118e-06,
+      "loss": 1.1493,
+      "step": 46500
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.9433,
+      "eval_samples_per_second": 386.3,
+      "eval_steps_per_second": 0.773,
+      "step": 46500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.057147175389383e-06,
+      "loss": 1.148,
+      "step": 46550
+    },
+    {
+      "epoch": 0.56,
+      "eval_runtime": 12.8841,
+      "eval_samples_per_second": 388.074,
+      "eval_steps_per_second": 0.776,
+      "step": 46550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.062579127242647e-06,
+      "loss": 1.1477,
+      "step": 46600
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.9047,
+      "eval_samples_per_second": 387.455,
+      "eval_steps_per_second": 0.775,
+      "step": 46600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.068011079095912e-06,
+      "loss": 1.1488,
+      "step": 46650
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.8638,
+      "eval_samples_per_second": 388.687,
+      "eval_steps_per_second": 0.777,
+      "step": 46650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.073443030949177e-06,
+      "loss": 1.1462,
+      "step": 46700
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.8249,
+      "eval_samples_per_second": 389.865,
+      "eval_steps_per_second": 0.78,
+      "step": 46700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.078874982802441e-06,
+      "loss": 1.1469,
+      "step": 46750
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.8392,
+      "eval_samples_per_second": 389.433,
+      "eval_steps_per_second": 0.779,
+      "step": 46750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.084306934655706e-06,
+      "loss": 1.1469,
+      "step": 46800
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.5988,
+      "eval_samples_per_second": 396.865,
+      "eval_steps_per_second": 0.794,
+      "step": 46800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.089738886508971e-06,
+      "loss": 1.1447,
+      "step": 46850
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.9979,
+      "eval_samples_per_second": 384.679,
+      "eval_steps_per_second": 0.769,
+      "step": 46850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.095170838362235e-06,
+      "loss": 1.1414,
+      "step": 46900
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.8335,
+      "eval_samples_per_second": 389.605,
+      "eval_steps_per_second": 0.779,
+      "step": 46900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.1006027902155e-06,
+      "loss": 1.1437,
+      "step": 46950
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.9782,
+      "eval_samples_per_second": 385.26,
+      "eval_steps_per_second": 0.771,
+      "step": 46950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.106034742068764e-06,
+      "loss": 1.1445,
+      "step": 47000
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.6038,
+      "eval_samples_per_second": 396.704,
+      "eval_steps_per_second": 0.793,
+      "step": 47000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.111466693922029e-06,
+      "loss": 1.144,
+      "step": 47050
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.6508,
+      "eval_samples_per_second": 395.232,
+      "eval_steps_per_second": 0.79,
+      "step": 47050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.116898645775294e-06,
+      "loss": 1.1432,
+      "step": 47100
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.9471,
+      "eval_samples_per_second": 386.186,
+      "eval_steps_per_second": 0.772,
+      "step": 47100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.122330597628558e-06,
+      "loss": 1.1427,
+      "step": 47150
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.9161,
+      "eval_samples_per_second": 387.114,
+      "eval_steps_per_second": 0.774,
+      "step": 47150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.127762549481823e-06,
+      "loss": 1.1456,
+      "step": 47200
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 13.0773,
+      "eval_samples_per_second": 382.343,
+      "eval_steps_per_second": 0.765,
+      "step": 47200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.133194501335088e-06,
+      "loss": 1.1437,
+      "step": 47250
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.5819,
+      "eval_samples_per_second": 397.395,
+      "eval_steps_per_second": 0.795,
+      "step": 47250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.138626453188352e-06,
+      "loss": 1.1447,
+      "step": 47300
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 12.827,
+      "eval_samples_per_second": 389.803,
+      "eval_steps_per_second": 0.78,
+      "step": 47300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.144058405041617e-06,
+      "loss": 1.1419,
+      "step": 47350
+    },
+    {
+      "epoch": 0.57,
+      "eval_runtime": 13.0079,
+      "eval_samples_per_second": 384.382,
+      "eval_steps_per_second": 0.769,
+      "step": 47350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.149490356894882e-06,
+      "loss": 1.1414,
+      "step": 47400
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.8597,
+      "eval_samples_per_second": 388.811,
+      "eval_steps_per_second": 0.778,
+      "step": 47400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.154922308748146e-06,
+      "loss": 1.1438,
+      "step": 47450
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.9668,
+      "eval_samples_per_second": 385.6,
+      "eval_steps_per_second": 0.771,
+      "step": 47450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.160354260601411e-06,
+      "loss": 1.1356,
+      "step": 47500
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 13.0062,
+      "eval_samples_per_second": 384.433,
+      "eval_steps_per_second": 0.769,
+      "step": 47500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.165786212454675e-06,
+      "loss": 1.1384,
+      "step": 47550
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.8828,
+      "eval_samples_per_second": 388.114,
+      "eval_steps_per_second": 0.776,
+      "step": 47550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1712181643079405e-06,
+      "loss": 1.1445,
+      "step": 47600
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.674,
+      "eval_samples_per_second": 394.508,
+      "eval_steps_per_second": 0.789,
+      "step": 47600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.176650116161205e-06,
+      "loss": 1.1459,
+      "step": 47650
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 13.0221,
+      "eval_samples_per_second": 383.964,
+      "eval_steps_per_second": 0.768,
+      "step": 47650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.182082068014469e-06,
+      "loss": 1.1375,
+      "step": 47700
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.6073,
+      "eval_samples_per_second": 396.596,
+      "eval_steps_per_second": 0.793,
+      "step": 47700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1875140198677345e-06,
+      "loss": 1.1336,
+      "step": 47750
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.6838,
+      "eval_samples_per_second": 394.205,
+      "eval_steps_per_second": 0.788,
+      "step": 47750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.1929459717209985e-06,
+      "loss": 1.1394,
+      "step": 47800
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.8286,
+      "eval_samples_per_second": 389.753,
+      "eval_steps_per_second": 0.78,
+      "step": 47800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.198377923574263e-06,
+      "loss": 1.1399,
+      "step": 47850
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.9275,
+      "eval_samples_per_second": 386.772,
+      "eval_steps_per_second": 0.774,
+      "step": 47850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.203809875427528e-06,
+      "loss": 1.1386,
+      "step": 47900
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.9365,
+      "eval_samples_per_second": 386.505,
+      "eval_steps_per_second": 0.773,
+      "step": 47900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2092418272807925e-06,
+      "loss": 1.1384,
+      "step": 47950
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.8091,
+      "eval_samples_per_second": 390.348,
+      "eval_steps_per_second": 0.781,
+      "step": 47950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2146737791340575e-06,
+      "loss": 1.1376,
+      "step": 48000
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.923,
+      "eval_samples_per_second": 386.906,
+      "eval_steps_per_second": 0.774,
+      "step": 48000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2201057309873215e-06,
+      "loss": 1.1318,
+      "step": 48050
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.8594,
+      "eval_samples_per_second": 388.822,
+      "eval_steps_per_second": 0.778,
+      "step": 48050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2255376828405865e-06,
+      "loss": 1.1325,
+      "step": 48100
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 13.1235,
+      "eval_samples_per_second": 380.997,
+      "eval_steps_per_second": 0.762,
+      "step": 48100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2309696346938514e-06,
+      "loss": 1.1342,
+      "step": 48150
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.9437,
+      "eval_samples_per_second": 386.289,
+      "eval_steps_per_second": 0.773,
+      "step": 48150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.2364015865471155e-06,
+      "loss": 1.1394,
+      "step": 48200
+    },
+    {
+      "epoch": 0.58,
+      "eval_runtime": 12.971,
+      "eval_samples_per_second": 385.477,
+      "eval_steps_per_second": 0.771,
+      "step": 48200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.24183353840038e-06,
+      "loss": 1.1294,
+      "step": 48250
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 13.0342,
+      "eval_samples_per_second": 383.605,
+      "eval_steps_per_second": 0.767,
+      "step": 48250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.247265490253645e-06,
+      "loss": 1.1286,
+      "step": 48300
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.6938,
+      "eval_samples_per_second": 393.893,
+      "eval_steps_per_second": 0.788,
+      "step": 48300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2526974421069095e-06,
+      "loss": 1.1379,
+      "step": 48350
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 13.0068,
+      "eval_samples_per_second": 384.414,
+      "eval_steps_per_second": 0.769,
+      "step": 48350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.258129393960174e-06,
+      "loss": 1.1351,
+      "step": 48400
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.6456,
+      "eval_samples_per_second": 395.393,
+      "eval_steps_per_second": 0.791,
+      "step": 48400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.263561345813439e-06,
+      "loss": 1.1302,
+      "step": 48450
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.8138,
+      "eval_samples_per_second": 390.203,
+      "eval_steps_per_second": 0.78,
+      "step": 48450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2689932976667035e-06,
+      "loss": 1.1346,
+      "step": 48500
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.8954,
+      "eval_samples_per_second": 387.736,
+      "eval_steps_per_second": 0.775,
+      "step": 48500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.274425249519968e-06,
+      "loss": 1.1259,
+      "step": 48550
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.602,
+      "eval_samples_per_second": 396.763,
+      "eval_steps_per_second": 0.794,
+      "step": 48550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2798572013732325e-06,
+      "loss": 1.1324,
+      "step": 48600
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.9987,
+      "eval_samples_per_second": 384.652,
+      "eval_steps_per_second": 0.769,
+      "step": 48600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.285289153226497e-06,
+      "loss": 1.1237,
+      "step": 48650
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 13.1304,
+      "eval_samples_per_second": 380.796,
+      "eval_steps_per_second": 0.762,
+      "step": 48650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.290721105079762e-06,
+      "loss": 1.1288,
+      "step": 48700
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.6939,
+      "eval_samples_per_second": 393.891,
+      "eval_steps_per_second": 0.788,
+      "step": 48700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.2961530569330265e-06,
+      "loss": 1.1281,
+      "step": 48750
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.9344,
+      "eval_samples_per_second": 386.567,
+      "eval_steps_per_second": 0.773,
+      "step": 48750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.301585008786291e-06,
+      "loss": 1.1361,
+      "step": 48800
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.6274,
+      "eval_samples_per_second": 395.964,
+      "eval_steps_per_second": 0.792,
+      "step": 48800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.307016960639556e-06,
+      "loss": 1.13,
+      "step": 48850
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.9561,
+      "eval_samples_per_second": 385.919,
+      "eval_steps_per_second": 0.772,
+      "step": 48850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.31244891249282e-06,
+      "loss": 1.1302,
+      "step": 48900
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 13.0363,
+      "eval_samples_per_second": 383.546,
+      "eval_steps_per_second": 0.767,
+      "step": 48900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.317880864346085e-06,
+      "loss": 1.1313,
+      "step": 48950
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.9524,
+      "eval_samples_per_second": 386.029,
+      "eval_steps_per_second": 0.772,
+      "step": 48950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.32331281619935e-06,
+      "loss": 1.1284,
+      "step": 49000
+    },
+    {
+      "epoch": 0.59,
+      "eval_runtime": 12.9414,
+      "eval_samples_per_second": 386.358,
+      "eval_steps_per_second": 0.773,
+      "step": 49000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.328744768052614e-06,
+      "loss": 1.1285,
+      "step": 49050
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.8154,
+      "eval_samples_per_second": 390.155,
+      "eval_steps_per_second": 0.78,
+      "step": 49050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.334176719905879e-06,
+      "loss": 1.1217,
+      "step": 49100
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.6693,
+      "eval_samples_per_second": 394.656,
+      "eval_steps_per_second": 0.789,
+      "step": 49100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.339608671759143e-06,
+      "loss": 1.1337,
+      "step": 49150
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 13.0045,
+      "eval_samples_per_second": 384.481,
+      "eval_steps_per_second": 0.769,
+      "step": 49150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.345040623612408e-06,
+      "loss": 1.1314,
+      "step": 49200
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.9451,
+      "eval_samples_per_second": 386.246,
+      "eval_steps_per_second": 0.772,
+      "step": 49200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.350472575465673e-06,
+      "loss": 1.1277,
+      "step": 49250
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.666,
+      "eval_samples_per_second": 394.759,
+      "eval_steps_per_second": 0.79,
+      "step": 49250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.355904527318937e-06,
+      "loss": 1.124,
+      "step": 49300
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.9408,
+      "eval_samples_per_second": 386.374,
+      "eval_steps_per_second": 0.773,
+      "step": 49300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.361336479172202e-06,
+      "loss": 1.1258,
+      "step": 49350
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 13.1275,
+      "eval_samples_per_second": 380.88,
+      "eval_steps_per_second": 0.762,
+      "step": 49350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.366768431025467e-06,
+      "loss": 1.12,
+      "step": 49400
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 13.0502,
+      "eval_samples_per_second": 383.136,
+      "eval_steps_per_second": 0.766,
+      "step": 49400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.372200382878731e-06,
+      "loss": 1.1283,
+      "step": 49450
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.8727,
+      "eval_samples_per_second": 388.42,
+      "eval_steps_per_second": 0.777,
+      "step": 49450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.377632334731996e-06,
+      "loss": 1.1261,
+      "step": 49500
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 13.0624,
+      "eval_samples_per_second": 382.778,
+      "eval_steps_per_second": 0.766,
+      "step": 49500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.383064286585261e-06,
+      "loss": 1.1202,
+      "step": 49550
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.8756,
+      "eval_samples_per_second": 388.332,
+      "eval_steps_per_second": 0.777,
+      "step": 49550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.388496238438525e-06,
+      "loss": 1.1223,
+      "step": 49600
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.8271,
+      "eval_samples_per_second": 389.8,
+      "eval_steps_per_second": 0.78,
+      "step": 49600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.39392819029179e-06,
+      "loss": 1.1249,
+      "step": 49650
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.9854,
+      "eval_samples_per_second": 385.049,
+      "eval_steps_per_second": 0.77,
+      "step": 49650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.399360142145054e-06,
+      "loss": 1.1254,
+      "step": 49700
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.999,
+      "eval_samples_per_second": 384.646,
+      "eval_steps_per_second": 0.769,
+      "step": 49700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.40479209399832e-06,
+      "loss": 1.1201,
+      "step": 49750
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.686,
+      "eval_samples_per_second": 394.136,
+      "eval_steps_per_second": 0.788,
+      "step": 49750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.410224045851584e-06,
+      "loss": 1.1222,
+      "step": 49800
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 12.6429,
+      "eval_samples_per_second": 395.478,
+      "eval_steps_per_second": 0.791,
+      "step": 49800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.415655997704848e-06,
+      "loss": 1.1213,
+      "step": 49850
+    },
+    {
+      "epoch": 0.6,
+      "eval_runtime": 13.0,
+      "eval_samples_per_second": 384.616,
+      "eval_steps_per_second": 0.769,
+      "step": 49850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.421087949558114e-06,
+      "loss": 1.1201,
+      "step": 49900
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9321,
+      "eval_samples_per_second": 386.635,
+      "eval_steps_per_second": 0.773,
+      "step": 49900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.426519901411378e-06,
+      "loss": 1.1231,
+      "step": 49950
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9503,
+      "eval_samples_per_second": 386.092,
+      "eval_steps_per_second": 0.772,
+      "step": 49950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.431951853264642e-06,
+      "loss": 1.1088,
+      "step": 50000
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9356,
+      "eval_samples_per_second": 386.529,
+      "eval_steps_per_second": 0.773,
+      "step": 50000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.437383805117907e-06,
+      "loss": 1.1205,
+      "step": 50050
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 13.0079,
+      "eval_samples_per_second": 384.382,
+      "eval_steps_per_second": 0.769,
+      "step": 50050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.442815756971172e-06,
+      "loss": 1.1238,
+      "step": 50100
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 13.0528,
+      "eval_samples_per_second": 383.059,
+      "eval_steps_per_second": 0.766,
+      "step": 50100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.448247708824437e-06,
+      "loss": 1.1281,
+      "step": 50150
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9074,
+      "eval_samples_per_second": 387.375,
+      "eval_steps_per_second": 0.775,
+      "step": 50150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.453679660677701e-06,
+      "loss": 1.1213,
+      "step": 50200
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9089,
+      "eval_samples_per_second": 387.331,
+      "eval_steps_per_second": 0.775,
+      "step": 50200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.459111612530966e-06,
+      "loss": 1.1139,
+      "step": 50250
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9471,
+      "eval_samples_per_second": 386.187,
+      "eval_steps_per_second": 0.772,
+      "step": 50250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.464543564384231e-06,
+      "loss": 1.1224,
+      "step": 50300
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 13.0045,
+      "eval_samples_per_second": 384.482,
+      "eval_steps_per_second": 0.769,
+      "step": 50300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.469975516237495e-06,
+      "loss": 1.1199,
+      "step": 50350
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9597,
+      "eval_samples_per_second": 385.81,
+      "eval_steps_per_second": 0.772,
+      "step": 50350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.475407468090759e-06,
+      "loss": 1.1204,
+      "step": 50400
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9285,
+      "eval_samples_per_second": 386.743,
+      "eval_steps_per_second": 0.773,
+      "step": 50400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.480839419944025e-06,
+      "loss": 1.1221,
+      "step": 50450
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.6426,
+      "eval_samples_per_second": 395.489,
+      "eval_steps_per_second": 0.791,
+      "step": 50450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.486271371797289e-06,
+      "loss": 1.1206,
+      "step": 50500
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.7177,
+      "eval_samples_per_second": 393.152,
+      "eval_steps_per_second": 0.786,
+      "step": 50500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.491703323650554e-06,
+      "loss": 1.1167,
+      "step": 50550
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9994,
+      "eval_samples_per_second": 384.632,
+      "eval_steps_per_second": 0.769,
+      "step": 50550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.497135275503818e-06,
+      "loss": 1.12,
+      "step": 50600
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.8743,
+      "eval_samples_per_second": 388.372,
+      "eval_steps_per_second": 0.777,
+      "step": 50600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.502567227357083e-06,
+      "loss": 1.1177,
+      "step": 50650
+    },
+    {
+      "epoch": 0.61,
+      "eval_runtime": 12.9593,
+      "eval_samples_per_second": 385.824,
+      "eval_steps_per_second": 0.772,
+      "step": 50650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.507999179210348e-06,
+      "loss": 1.114,
+      "step": 50700
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.6534,
+      "eval_samples_per_second": 395.15,
+      "eval_steps_per_second": 0.79,
+      "step": 50700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.513431131063612e-06,
+      "loss": 1.1206,
+      "step": 50750
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.9323,
+      "eval_samples_per_second": 386.629,
+      "eval_steps_per_second": 0.773,
+      "step": 50750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.518863082916877e-06,
+      "loss": 1.115,
+      "step": 50800
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.9688,
+      "eval_samples_per_second": 385.542,
+      "eval_steps_per_second": 0.771,
+      "step": 50800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.524295034770142e-06,
+      "loss": 1.1154,
+      "step": 50850
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.8163,
+      "eval_samples_per_second": 390.127,
+      "eval_steps_per_second": 0.78,
+      "step": 50850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.529726986623406e-06,
+      "loss": 1.112,
+      "step": 50900
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.8298,
+      "eval_samples_per_second": 389.717,
+      "eval_steps_per_second": 0.779,
+      "step": 50900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.535158938476671e-06,
+      "loss": 1.1105,
+      "step": 50950
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 13.1794,
+      "eval_samples_per_second": 379.381,
+      "eval_steps_per_second": 0.759,
+      "step": 50950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.540590890329936e-06,
+      "loss": 1.1143,
+      "step": 51000
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 13.2893,
+      "eval_samples_per_second": 376.244,
+      "eval_steps_per_second": 0.752,
+      "step": 51000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.5460228421832e-06,
+      "loss": 1.1104,
+      "step": 51050
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.8096,
+      "eval_samples_per_second": 390.332,
+      "eval_steps_per_second": 0.781,
+      "step": 51050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.551454794036465e-06,
+      "loss": 1.1068,
+      "step": 51100
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.8655,
+      "eval_samples_per_second": 388.636,
+      "eval_steps_per_second": 0.777,
+      "step": 51100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.55688674588973e-06,
+      "loss": 1.1095,
+      "step": 51150
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.862,
+      "eval_samples_per_second": 388.742,
+      "eval_steps_per_second": 0.777,
+      "step": 51150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.562318697742994e-06,
+      "loss": 1.1129,
+      "step": 51200
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.631,
+      "eval_samples_per_second": 395.851,
+      "eval_steps_per_second": 0.792,
+      "step": 51200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.567750649596259e-06,
+      "loss": 1.1076,
+      "step": 51250
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.8855,
+      "eval_samples_per_second": 388.032,
+      "eval_steps_per_second": 0.776,
+      "step": 51250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.573182601449523e-06,
+      "loss": 1.1127,
+      "step": 51300
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.9737,
+      "eval_samples_per_second": 385.394,
+      "eval_steps_per_second": 0.771,
+      "step": 51300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.578614553302788e-06,
+      "loss": 1.1107,
+      "step": 51350
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.9272,
+      "eval_samples_per_second": 386.782,
+      "eval_steps_per_second": 0.774,
+      "step": 51350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.584046505156053e-06,
+      "loss": 1.1111,
+      "step": 51400
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.6478,
+      "eval_samples_per_second": 395.325,
+      "eval_steps_per_second": 0.791,
+      "step": 51400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.589478457009317e-06,
+      "loss": 1.1135,
+      "step": 51450
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.6327,
+      "eval_samples_per_second": 395.798,
+      "eval_steps_per_second": 0.792,
+      "step": 51450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.594910408862582e-06,
+      "loss": 1.1139,
+      "step": 51500
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 12.7261,
+      "eval_samples_per_second": 392.892,
+      "eval_steps_per_second": 0.786,
+      "step": 51500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.600342360715847e-06,
+      "loss": 1.1092,
+      "step": 51550
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.9734,
+      "eval_samples_per_second": 385.403,
+      "eval_steps_per_second": 0.771,
+      "step": 51550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.605774312569111e-06,
+      "loss": 1.1023,
+      "step": 51600
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.9914,
+      "eval_samples_per_second": 384.871,
+      "eval_steps_per_second": 0.77,
+      "step": 51600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.611206264422376e-06,
+      "loss": 1.1145,
+      "step": 51650
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.8598,
+      "eval_samples_per_second": 388.807,
+      "eval_steps_per_second": 0.778,
+      "step": 51650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.616638216275641e-06,
+      "loss": 1.1106,
+      "step": 51700
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.936,
+      "eval_samples_per_second": 386.518,
+      "eval_steps_per_second": 0.773,
+      "step": 51700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.622070168128905e-06,
+      "loss": 1.1131,
+      "step": 51750
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.6394,
+      "eval_samples_per_second": 395.588,
+      "eval_steps_per_second": 0.791,
+      "step": 51750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.62750211998217e-06,
+      "loss": 1.1089,
+      "step": 51800
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.746,
+      "eval_samples_per_second": 392.28,
+      "eval_steps_per_second": 0.785,
+      "step": 51800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.632934071835434e-06,
+      "loss": 1.1114,
+      "step": 51850
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.9363,
+      "eval_samples_per_second": 386.509,
+      "eval_steps_per_second": 0.773,
+      "step": 51850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6383660236887e-06,
+      "loss": 1.1072,
+      "step": 51900
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.6228,
+      "eval_samples_per_second": 396.107,
+      "eval_steps_per_second": 0.792,
+      "step": 51900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.643797975541964e-06,
+      "loss": 1.1073,
+      "step": 51950
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.8231,
+      "eval_samples_per_second": 389.923,
+      "eval_steps_per_second": 0.78,
+      "step": 51950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.649229927395228e-06,
+      "loss": 1.1053,
+      "step": 52000
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.9357,
+      "eval_samples_per_second": 386.528,
+      "eval_steps_per_second": 0.773,
+      "step": 52000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.654661879248494e-06,
+      "loss": 1.1064,
+      "step": 52050
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.6737,
+      "eval_samples_per_second": 394.517,
+      "eval_steps_per_second": 0.789,
+      "step": 52050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.660093831101758e-06,
+      "loss": 1.1062,
+      "step": 52100
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.9187,
+      "eval_samples_per_second": 387.035,
+      "eval_steps_per_second": 0.774,
+      "step": 52100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.665525782955022e-06,
+      "loss": 1.1107,
+      "step": 52150
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 13.0237,
+      "eval_samples_per_second": 383.915,
+      "eval_steps_per_second": 0.768,
+      "step": 52150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.670957734808287e-06,
+      "loss": 1.1051,
+      "step": 52200
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 12.7422,
+      "eval_samples_per_second": 392.398,
+      "eval_steps_per_second": 0.785,
+      "step": 52200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.676389686661552e-06,
+      "loss": 1.1023,
+      "step": 52250
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 13.0339,
+      "eval_samples_per_second": 383.615,
+      "eval_steps_per_second": 0.767,
+      "step": 52250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.681821638514817e-06,
+      "loss": 1.1028,
+      "step": 52300
+    },
+    {
+      "epoch": 0.63,
+      "eval_runtime": 13.1075,
+      "eval_samples_per_second": 381.461,
+      "eval_steps_per_second": 0.763,
+      "step": 52300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.687253590368081e-06,
+      "loss": 1.102,
+      "step": 52350
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.8899,
+      "eval_samples_per_second": 387.902,
+      "eval_steps_per_second": 0.776,
+      "step": 52350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.692685542221345e-06,
+      "loss": 1.1023,
+      "step": 52400
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.964,
+      "eval_samples_per_second": 385.684,
+      "eval_steps_per_second": 0.771,
+      "step": 52400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.698117494074611e-06,
+      "loss": 1.1076,
+      "step": 52450
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.6162,
+      "eval_samples_per_second": 396.316,
+      "eval_steps_per_second": 0.793,
+      "step": 52450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.703549445927875e-06,
+      "loss": 1.1118,
+      "step": 52500
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 13.0014,
+      "eval_samples_per_second": 384.574,
+      "eval_steps_per_second": 0.769,
+      "step": 52500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.708981397781139e-06,
+      "loss": 1.1027,
+      "step": 52550
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.9821,
+      "eval_samples_per_second": 385.145,
+      "eval_steps_per_second": 0.77,
+      "step": 52550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.714413349634405e-06,
+      "loss": 1.1054,
+      "step": 52600
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.6718,
+      "eval_samples_per_second": 394.576,
+      "eval_steps_per_second": 0.789,
+      "step": 52600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.719845301487669e-06,
+      "loss": 1.1031,
+      "step": 52650
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.9609,
+      "eval_samples_per_second": 385.776,
+      "eval_steps_per_second": 0.772,
+      "step": 52650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.725277253340934e-06,
+      "loss": 1.1008,
+      "step": 52700
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.9515,
+      "eval_samples_per_second": 386.056,
+      "eval_steps_per_second": 0.772,
+      "step": 52700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.730709205194198e-06,
+      "loss": 1.1059,
+      "step": 52750
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 13.1927,
+      "eval_samples_per_second": 378.997,
+      "eval_steps_per_second": 0.758,
+      "step": 52750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.736141157047463e-06,
+      "loss": 1.1012,
+      "step": 52800
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.8235,
+      "eval_samples_per_second": 389.908,
+      "eval_steps_per_second": 0.78,
+      "step": 52800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.741573108900728e-06,
+      "loss": 1.1007,
+      "step": 52850
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.631,
+      "eval_samples_per_second": 395.853,
+      "eval_steps_per_second": 0.792,
+      "step": 52850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.747005060753992e-06,
+      "loss": 1.1073,
+      "step": 52900
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.9488,
+      "eval_samples_per_second": 386.135,
+      "eval_steps_per_second": 0.772,
+      "step": 52900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.752437012607257e-06,
+      "loss": 1.1037,
+      "step": 52950
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 13.2632,
+      "eval_samples_per_second": 376.984,
+      "eval_steps_per_second": 0.754,
+      "step": 52950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.757868964460522e-06,
+      "loss": 1.1011,
+      "step": 53000
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 13.0488,
+      "eval_samples_per_second": 383.176,
+      "eval_steps_per_second": 0.766,
+      "step": 53000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.763300916313786e-06,
+      "loss": 1.098,
+      "step": 53050
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 12.6602,
+      "eval_samples_per_second": 394.939,
+      "eval_steps_per_second": 0.79,
+      "step": 53050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.768732868167051e-06,
+      "loss": 1.1,
+      "step": 53100
+    },
+    {
+      "epoch": 0.64,
+      "eval_runtime": 13.1836,
+      "eval_samples_per_second": 379.26,
+      "eval_steps_per_second": 0.759,
+      "step": 53100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774164820020316e-06,
+      "loss": 1.1,
+      "step": 53150
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.9874,
+      "eval_samples_per_second": 384.989,
+      "eval_steps_per_second": 0.77,
+      "step": 53150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77959677187358e-06,
+      "loss": 1.0976,
+      "step": 53200
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.6758,
+      "eval_samples_per_second": 394.452,
+      "eval_steps_per_second": 0.789,
+      "step": 53200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785028723726845e-06,
+      "loss": 1.0996,
+      "step": 53250
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.9274,
+      "eval_samples_per_second": 386.776,
+      "eval_steps_per_second": 0.774,
+      "step": 53250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790460675580109e-06,
+      "loss": 1.1033,
+      "step": 53300
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.9723,
+      "eval_samples_per_second": 385.437,
+      "eval_steps_per_second": 0.771,
+      "step": 53300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795892627433374e-06,
+      "loss": 1.0961,
+      "step": 53350
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 13.2038,
+      "eval_samples_per_second": 378.679,
+      "eval_steps_per_second": 0.757,
+      "step": 53350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801324579286639e-06,
+      "loss": 1.1032,
+      "step": 53400
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 13.0176,
+      "eval_samples_per_second": 384.096,
+      "eval_steps_per_second": 0.768,
+      "step": 53400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806756531139903e-06,
+      "loss": 1.1014,
+      "step": 53450
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 13.0429,
+      "eval_samples_per_second": 383.351,
+      "eval_steps_per_second": 0.767,
+      "step": 53450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812188482993168e-06,
+      "loss": 1.0998,
+      "step": 53500
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.6803,
+      "eval_samples_per_second": 394.314,
+      "eval_steps_per_second": 0.789,
+      "step": 53500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817620434846433e-06,
+      "loss": 1.0955,
+      "step": 53550
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 13.1708,
+      "eval_samples_per_second": 379.629,
+      "eval_steps_per_second": 0.759,
+      "step": 53550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823052386699697e-06,
+      "loss": 1.0973,
+      "step": 53600
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.947,
+      "eval_samples_per_second": 386.189,
+      "eval_steps_per_second": 0.772,
+      "step": 53600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828484338552962e-06,
+      "loss": 1.0957,
+      "step": 53650
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.6964,
+      "eval_samples_per_second": 393.812,
+      "eval_steps_per_second": 0.788,
+      "step": 53650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833916290406227e-06,
+      "loss": 1.0949,
+      "step": 53700
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.8215,
+      "eval_samples_per_second": 389.969,
+      "eval_steps_per_second": 0.78,
+      "step": 53700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839348242259491e-06,
+      "loss": 1.0968,
+      "step": 53750
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.982,
+      "eval_samples_per_second": 385.148,
+      "eval_steps_per_second": 0.77,
+      "step": 53750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844780194112756e-06,
+      "loss": 1.0953,
+      "step": 53800
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.9413,
+      "eval_samples_per_second": 386.36,
+      "eval_steps_per_second": 0.773,
+      "step": 53800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8502121459660206e-06,
+      "loss": 1.093,
+      "step": 53850
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 13.252,
+      "eval_samples_per_second": 377.302,
+      "eval_steps_per_second": 0.755,
+      "step": 53850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855644097819285e-06,
+      "loss": 1.1025,
+      "step": 53900
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 12.9873,
+      "eval_samples_per_second": 384.992,
+      "eval_steps_per_second": 0.77,
+      "step": 53900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86107604967255e-06,
+      "loss": 1.0959,
+      "step": 53950
+    },
+    {
+      "epoch": 0.65,
+      "eval_runtime": 13.0236,
+      "eval_samples_per_second": 383.919,
+      "eval_steps_per_second": 0.768,
+      "step": 53950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.866508001525814e-06,
+      "loss": 1.0939,
+      "step": 54000
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.8617,
+      "eval_samples_per_second": 388.751,
+      "eval_steps_per_second": 0.778,
+      "step": 54000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8719399533790795e-06,
+      "loss": 1.0904,
+      "step": 54050
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.984,
+      "eval_samples_per_second": 385.09,
+      "eval_steps_per_second": 0.77,
+      "step": 54050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8773719052323436e-06,
+      "loss": 1.0943,
+      "step": 54100
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.8313,
+      "eval_samples_per_second": 389.672,
+      "eval_steps_per_second": 0.779,
+      "step": 54100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.882803857085608e-06,
+      "loss": 1.0918,
+      "step": 54150
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 13.056,
+      "eval_samples_per_second": 382.965,
+      "eval_steps_per_second": 0.766,
+      "step": 54150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.888235808938873e-06,
+      "loss": 1.0941,
+      "step": 54200
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 13.068,
+      "eval_samples_per_second": 382.614,
+      "eval_steps_per_second": 0.765,
+      "step": 54200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.8936677607921375e-06,
+      "loss": 1.0921,
+      "step": 54250
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.8967,
+      "eval_samples_per_second": 387.697,
+      "eval_steps_per_second": 0.775,
+      "step": 54250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.899099712645402e-06,
+      "loss": 1.0959,
+      "step": 54300
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.8543,
+      "eval_samples_per_second": 388.976,
+      "eval_steps_per_second": 0.778,
+      "step": 54300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9045316644986666e-06,
+      "loss": 1.0927,
+      "step": 54350
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.6504,
+      "eval_samples_per_second": 395.244,
+      "eval_steps_per_second": 0.79,
+      "step": 54350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9099636163519315e-06,
+      "loss": 1.0949,
+      "step": 54400
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.6549,
+      "eval_samples_per_second": 395.102,
+      "eval_steps_per_second": 0.79,
+      "step": 54400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9153955682051964e-06,
+      "loss": 1.0911,
+      "step": 54450
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.6582,
+      "eval_samples_per_second": 395.001,
+      "eval_steps_per_second": 0.79,
+      "step": 54450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9208275200584605e-06,
+      "loss": 1.0868,
+      "step": 54500
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.9413,
+      "eval_samples_per_second": 386.361,
+      "eval_steps_per_second": 0.773,
+      "step": 54500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.926259471911725e-06,
+      "loss": 1.0949,
+      "step": 54550
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.99,
+      "eval_samples_per_second": 384.912,
+      "eval_steps_per_second": 0.77,
+      "step": 54550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.93169142376499e-06,
+      "loss": 1.0941,
+      "step": 54600
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.975,
+      "eval_samples_per_second": 385.355,
+      "eval_steps_per_second": 0.771,
+      "step": 54600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9371233756182545e-06,
+      "loss": 1.0921,
+      "step": 54650
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.9656,
+      "eval_samples_per_second": 385.636,
+      "eval_steps_per_second": 0.771,
+      "step": 54650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.942555327471519e-06,
+      "loss": 1.0949,
+      "step": 54700
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 13.0118,
+      "eval_samples_per_second": 384.268,
+      "eval_steps_per_second": 0.769,
+      "step": 54700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.947987279324784e-06,
+      "loss": 1.0922,
+      "step": 54750
+    },
+    {
+      "epoch": 0.66,
+      "eval_runtime": 12.8546,
+      "eval_samples_per_second": 388.965,
+      "eval_steps_per_second": 0.778,
+      "step": 54750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9534192311780485e-06,
+      "loss": 1.0879,
+      "step": 54800
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.9834,
+      "eval_samples_per_second": 385.106,
+      "eval_steps_per_second": 0.77,
+      "step": 54800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.958851183031313e-06,
+      "loss": 1.0881,
+      "step": 54850
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 13.0311,
+      "eval_samples_per_second": 383.698,
+      "eval_steps_per_second": 0.767,
+      "step": 54850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9642831348845775e-06,
+      "loss": 1.0861,
+      "step": 54900
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.6591,
+      "eval_samples_per_second": 394.973,
+      "eval_steps_per_second": 0.79,
+      "step": 54900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9697150867378424e-06,
+      "loss": 1.09,
+      "step": 54950
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 13.0663,
+      "eval_samples_per_second": 382.665,
+      "eval_steps_per_second": 0.765,
+      "step": 54950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.975147038591107e-06,
+      "loss": 1.0887,
+      "step": 55000
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.87,
+      "eval_samples_per_second": 388.501,
+      "eval_steps_per_second": 0.777,
+      "step": 55000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9805789904443715e-06,
+      "loss": 1.0865,
+      "step": 55050
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.799,
+      "eval_samples_per_second": 390.657,
+      "eval_steps_per_second": 0.781,
+      "step": 55050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9860109422976356e-06,
+      "loss": 1.091,
+      "step": 55100
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.985,
+      "eval_samples_per_second": 385.06,
+      "eval_steps_per_second": 0.77,
+      "step": 55100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.991442894150901e-06,
+      "loss": 1.0897,
+      "step": 55150
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.9901,
+      "eval_samples_per_second": 384.908,
+      "eval_steps_per_second": 0.77,
+      "step": 55150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9968748460041654e-06,
+      "loss": 1.0914,
+      "step": 55200
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 13.1208,
+      "eval_samples_per_second": 381.073,
+      "eval_steps_per_second": 0.762,
+      "step": 55200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.00230679785743e-06,
+      "loss": 1.0888,
+      "step": 55250
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.7906,
+      "eval_samples_per_second": 390.912,
+      "eval_steps_per_second": 0.782,
+      "step": 55250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.007738749710695e-06,
+      "loss": 1.0851,
+      "step": 55300
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.9286,
+      "eval_samples_per_second": 386.739,
+      "eval_steps_per_second": 0.773,
+      "step": 55300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.013170701563959e-06,
+      "loss": 1.087,
+      "step": 55350
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.9202,
+      "eval_samples_per_second": 386.992,
+      "eval_steps_per_second": 0.774,
+      "step": 55350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.018602653417224e-06,
+      "loss": 1.0838,
+      "step": 55400
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 13.0302,
+      "eval_samples_per_second": 383.725,
+      "eval_steps_per_second": 0.767,
+      "step": 55400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0240346052704884e-06,
+      "loss": 1.0843,
+      "step": 55450
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 13.0283,
+      "eval_samples_per_second": 383.78,
+      "eval_steps_per_second": 0.768,
+      "step": 55450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.029466557123753e-06,
+      "loss": 1.0871,
+      "step": 55500
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 13.0146,
+      "eval_samples_per_second": 384.184,
+      "eval_steps_per_second": 0.768,
+      "step": 55500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.034898508977018e-06,
+      "loss": 1.0818,
+      "step": 55550
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.6553,
+      "eval_samples_per_second": 395.091,
+      "eval_steps_per_second": 0.79,
+      "step": 55550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.040330460830282e-06,
+      "loss": 1.0835,
+      "step": 55600
+    },
+    {
+      "epoch": 0.67,
+      "eval_runtime": 12.6768,
+      "eval_samples_per_second": 394.421,
+      "eval_steps_per_second": 0.789,
+      "step": 55600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.045762412683547e-06,
+      "loss": 1.0861,
+      "step": 55650
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.6554,
+      "eval_samples_per_second": 395.087,
+      "eval_steps_per_second": 0.79,
+      "step": 55650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.051194364536812e-06,
+      "loss": 1.0804,
+      "step": 55700
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.9064,
+      "eval_samples_per_second": 387.406,
+      "eval_steps_per_second": 0.775,
+      "step": 55700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.056626316390076e-06,
+      "loss": 1.0819,
+      "step": 55750
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.9672,
+      "eval_samples_per_second": 385.588,
+      "eval_steps_per_second": 0.771,
+      "step": 55750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.062058268243341e-06,
+      "loss": 1.0801,
+      "step": 55800
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.7471,
+      "eval_samples_per_second": 392.245,
+      "eval_steps_per_second": 0.784,
+      "step": 55800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.067490220096606e-06,
+      "loss": 1.0849,
+      "step": 55850
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 13.0327,
+      "eval_samples_per_second": 383.649,
+      "eval_steps_per_second": 0.767,
+      "step": 55850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.07292217194987e-06,
+      "loss": 1.0849,
+      "step": 55900
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 13.0205,
+      "eval_samples_per_second": 384.01,
+      "eval_steps_per_second": 0.768,
+      "step": 55900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.078354123803135e-06,
+      "loss": 1.0798,
+      "step": 55950
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.9084,
+      "eval_samples_per_second": 387.345,
+      "eval_steps_per_second": 0.775,
+      "step": 55950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.0837860756564e-06,
+      "loss": 1.0805,
+      "step": 56000
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.7699,
+      "eval_samples_per_second": 391.545,
+      "eval_steps_per_second": 0.783,
+      "step": 56000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.089218027509664e-06,
+      "loss": 1.0771,
+      "step": 56050
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.878,
+      "eval_samples_per_second": 388.259,
+      "eval_steps_per_second": 0.777,
+      "step": 56050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.094649979362929e-06,
+      "loss": 1.08,
+      "step": 56100
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.7857,
+      "eval_samples_per_second": 391.063,
+      "eval_steps_per_second": 0.782,
+      "step": 56100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.100081931216193e-06,
+      "loss": 1.0826,
+      "step": 56150
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 13.0297,
+      "eval_samples_per_second": 383.739,
+      "eval_steps_per_second": 0.767,
+      "step": 56150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.105513883069459e-06,
+      "loss": 1.0802,
+      "step": 56200
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 13.0359,
+      "eval_samples_per_second": 383.557,
+      "eval_steps_per_second": 0.767,
+      "step": 56200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.110945834922723e-06,
+      "loss": 1.082,
+      "step": 56250
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.6862,
+      "eval_samples_per_second": 394.128,
+      "eval_steps_per_second": 0.788,
+      "step": 56250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.116377786775987e-06,
+      "loss": 1.0777,
+      "step": 56300
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.6849,
+      "eval_samples_per_second": 394.17,
+      "eval_steps_per_second": 0.788,
+      "step": 56300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.121809738629252e-06,
+      "loss": 1.0855,
+      "step": 56350
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 12.7339,
+      "eval_samples_per_second": 392.653,
+      "eval_steps_per_second": 0.785,
+      "step": 56350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.127241690482517e-06,
+      "loss": 1.0734,
+      "step": 56400
+    },
+    {
+      "epoch": 0.68,
+      "eval_runtime": 13.0214,
+      "eval_samples_per_second": 383.982,
+      "eval_steps_per_second": 0.768,
+      "step": 56400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.132673642335781e-06,
+      "loss": 1.0796,
+      "step": 56450
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.9591,
+      "eval_samples_per_second": 385.828,
+      "eval_steps_per_second": 0.772,
+      "step": 56450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.138105594189046e-06,
+      "loss": 1.079,
+      "step": 56500
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.8936,
+      "eval_samples_per_second": 387.79,
+      "eval_steps_per_second": 0.776,
+      "step": 56500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.143537546042311e-06,
+      "loss": 1.0837,
+      "step": 56550
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.6396,
+      "eval_samples_per_second": 395.582,
+      "eval_steps_per_second": 0.791,
+      "step": 56550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.148969497895576e-06,
+      "loss": 1.0719,
+      "step": 56600
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 13.0069,
+      "eval_samples_per_second": 384.413,
+      "eval_steps_per_second": 0.769,
+      "step": 56600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.15440144974884e-06,
+      "loss": 1.0764,
+      "step": 56650
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.8262,
+      "eval_samples_per_second": 389.828,
+      "eval_steps_per_second": 0.78,
+      "step": 56650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.159833401602104e-06,
+      "loss": 1.0716,
+      "step": 56700
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.8205,
+      "eval_samples_per_second": 390.001,
+      "eval_steps_per_second": 0.78,
+      "step": 56700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.16526535345537e-06,
+      "loss": 1.0811,
+      "step": 56750
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.9586,
+      "eval_samples_per_second": 385.845,
+      "eval_steps_per_second": 0.772,
+      "step": 56750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.170697305308634e-06,
+      "loss": 1.0768,
+      "step": 56800
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 13.0377,
+      "eval_samples_per_second": 383.502,
+      "eval_steps_per_second": 0.767,
+      "step": 56800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.176129257161898e-06,
+      "loss": 1.0733,
+      "step": 56850
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.9651,
+      "eval_samples_per_second": 385.65,
+      "eval_steps_per_second": 0.771,
+      "step": 56850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.181561209015164e-06,
+      "loss": 1.0713,
+      "step": 56900
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 13.1629,
+      "eval_samples_per_second": 379.855,
+      "eval_steps_per_second": 0.76,
+      "step": 56900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.186993160868428e-06,
+      "loss": 1.073,
+      "step": 56950
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.9576,
+      "eval_samples_per_second": 385.875,
+      "eval_steps_per_second": 0.772,
+      "step": 56950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.192425112721693e-06,
+      "loss": 1.0815,
+      "step": 57000
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 13.1777,
+      "eval_samples_per_second": 379.428,
+      "eval_steps_per_second": 0.759,
+      "step": 57000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.197857064574957e-06,
+      "loss": 1.0722,
+      "step": 57050
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.6822,
+      "eval_samples_per_second": 394.253,
+      "eval_steps_per_second": 0.789,
+      "step": 57050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.203289016428222e-06,
+      "loss": 1.0737,
+      "step": 57100
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.6893,
+      "eval_samples_per_second": 394.033,
+      "eval_steps_per_second": 0.788,
+      "step": 57100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.208720968281487e-06,
+      "loss": 1.0743,
+      "step": 57150
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.8088,
+      "eval_samples_per_second": 390.358,
+      "eval_steps_per_second": 0.781,
+      "step": 57150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.214152920134751e-06,
+      "loss": 1.0754,
+      "step": 57200
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 13.0956,
+      "eval_samples_per_second": 381.808,
+      "eval_steps_per_second": 0.764,
+      "step": 57200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.219584871988016e-06,
+      "loss": 1.0756,
+      "step": 57250
+    },
+    {
+      "epoch": 0.69,
+      "eval_runtime": 12.8705,
+      "eval_samples_per_second": 388.486,
+      "eval_steps_per_second": 0.777,
+      "step": 57250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.225016823841281e-06,
+      "loss": 1.0694,
+      "step": 57300
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 13.0329,
+      "eval_samples_per_second": 383.646,
+      "eval_steps_per_second": 0.767,
+      "step": 57300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.230448775694545e-06,
+      "loss": 1.0766,
+      "step": 57350
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.9769,
+      "eval_samples_per_second": 385.3,
+      "eval_steps_per_second": 0.771,
+      "step": 57350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.23588072754781e-06,
+      "loss": 1.0771,
+      "step": 57400
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.9293,
+      "eval_samples_per_second": 386.718,
+      "eval_steps_per_second": 0.773,
+      "step": 57400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.241312679401075e-06,
+      "loss": 1.0723,
+      "step": 57450
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 13.0978,
+      "eval_samples_per_second": 381.744,
+      "eval_steps_per_second": 0.763,
+      "step": 57450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.246744631254339e-06,
+      "loss": 1.0702,
+      "step": 57500
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.946,
+      "eval_samples_per_second": 386.219,
+      "eval_steps_per_second": 0.772,
+      "step": 57500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.252176583107604e-06,
+      "loss": 1.0767,
+      "step": 57550
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.6438,
+      "eval_samples_per_second": 395.452,
+      "eval_steps_per_second": 0.791,
+      "step": 57550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.257608534960868e-06,
+      "loss": 1.0728,
+      "step": 57600
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.6877,
+      "eval_samples_per_second": 394.082,
+      "eval_steps_per_second": 0.788,
+      "step": 57600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.263040486814133e-06,
+      "loss": 1.0733,
+      "step": 57650
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.998,
+      "eval_samples_per_second": 384.676,
+      "eval_steps_per_second": 0.769,
+      "step": 57650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.268472438667397e-06,
+      "loss": 1.0705,
+      "step": 57700
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.9563,
+      "eval_samples_per_second": 385.911,
+      "eval_steps_per_second": 0.772,
+      "step": 57700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.273904390520663e-06,
+      "loss": 1.0726,
+      "step": 57750
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.6589,
+      "eval_samples_per_second": 394.978,
+      "eval_steps_per_second": 0.79,
+      "step": 57750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.279336342373927e-06,
+      "loss": 1.0677,
+      "step": 57800
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.9212,
+      "eval_samples_per_second": 386.962,
+      "eval_steps_per_second": 0.774,
+      "step": 57800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.284768294227192e-06,
+      "loss": 1.0669,
+      "step": 57850
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.9711,
+      "eval_samples_per_second": 385.473,
+      "eval_steps_per_second": 0.771,
+      "step": 57850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.290200246080456e-06,
+      "loss": 1.0712,
+      "step": 57900
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.9445,
+      "eval_samples_per_second": 386.265,
+      "eval_steps_per_second": 0.773,
+      "step": 57900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.295632197933721e-06,
+      "loss": 1.0667,
+      "step": 57950
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.6627,
+      "eval_samples_per_second": 394.859,
+      "eval_steps_per_second": 0.79,
+      "step": 57950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.301064149786985e-06,
+      "loss": 1.0664,
+      "step": 58000
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.8922,
+      "eval_samples_per_second": 387.833,
+      "eval_steps_per_second": 0.776,
+      "step": 58000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.30649610164025e-06,
+      "loss": 1.0678,
+      "step": 58050
+    },
+    {
+      "epoch": 0.7,
+      "eval_runtime": 12.8687,
+      "eval_samples_per_second": 388.541,
+      "eval_steps_per_second": 0.777,
+      "step": 58050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.311928053493516e-06,
+      "loss": 1.0633,
+      "step": 58100
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.6665,
+      "eval_samples_per_second": 394.743,
+      "eval_steps_per_second": 0.789,
+      "step": 58100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.31736000534678e-06,
+      "loss": 1.0648,
+      "step": 58150
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.1975,
+      "eval_samples_per_second": 378.86,
+      "eval_steps_per_second": 0.758,
+      "step": 58150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.322791957200044e-06,
+      "loss": 1.064,
+      "step": 58200
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.7402,
+      "eval_samples_per_second": 392.458,
+      "eval_steps_per_second": 0.785,
+      "step": 58200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.328223909053309e-06,
+      "loss": 1.0698,
+      "step": 58250
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.0192,
+      "eval_samples_per_second": 384.048,
+      "eval_steps_per_second": 0.768,
+      "step": 58250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.333655860906573e-06,
+      "loss": 1.0721,
+      "step": 58300
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.6794,
+      "eval_samples_per_second": 394.341,
+      "eval_steps_per_second": 0.789,
+      "step": 58300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.339087812759838e-06,
+      "loss": 1.0694,
+      "step": 58350
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.1353,
+      "eval_samples_per_second": 380.655,
+      "eval_steps_per_second": 0.761,
+      "step": 58350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.344519764613102e-06,
+      "loss": 1.0713,
+      "step": 58400
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.0202,
+      "eval_samples_per_second": 384.019,
+      "eval_steps_per_second": 0.768,
+      "step": 58400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.349951716466368e-06,
+      "loss": 1.0621,
+      "step": 58450
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.9128,
+      "eval_samples_per_second": 387.212,
+      "eval_steps_per_second": 0.774,
+      "step": 58450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.355383668319633e-06,
+      "loss": 1.0702,
+      "step": 58500
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.9885,
+      "eval_samples_per_second": 384.957,
+      "eval_steps_per_second": 0.77,
+      "step": 58500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.360815620172897e-06,
+      "loss": 1.0616,
+      "step": 58550
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.0321,
+      "eval_samples_per_second": 383.668,
+      "eval_steps_per_second": 0.767,
+      "step": 58550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.366247572026161e-06,
+      "loss": 1.0675,
+      "step": 58600
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.9437,
+      "eval_samples_per_second": 386.288,
+      "eval_steps_per_second": 0.773,
+      "step": 58600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.371679523879426e-06,
+      "loss": 1.068,
+      "step": 58650
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.8506,
+      "eval_samples_per_second": 389.088,
+      "eval_steps_per_second": 0.778,
+      "step": 58650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.37711147573269e-06,
+      "loss": 1.0626,
+      "step": 58700
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.0504,
+      "eval_samples_per_second": 383.129,
+      "eval_steps_per_second": 0.766,
+      "step": 58700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.382543427585955e-06,
+      "loss": 1.0681,
+      "step": 58750
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.8908,
+      "eval_samples_per_second": 387.875,
+      "eval_steps_per_second": 0.776,
+      "step": 58750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.387975379439221e-06,
+      "loss": 1.06,
+      "step": 58800
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.6507,
+      "eval_samples_per_second": 395.233,
+      "eval_steps_per_second": 0.79,
+      "step": 58800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.393407331292485e-06,
+      "loss": 1.0594,
+      "step": 58850
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 13.0152,
+      "eval_samples_per_second": 384.167,
+      "eval_steps_per_second": 0.768,
+      "step": 58850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.39883928314575e-06,
+      "loss": 1.0647,
+      "step": 58900
+    },
+    {
+      "epoch": 0.71,
+      "eval_runtime": 12.9764,
+      "eval_samples_per_second": 385.315,
+      "eval_steps_per_second": 0.771,
+      "step": 58900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.404271234999014e-06,
+      "loss": 1.0613,
+      "step": 58950
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.0971,
+      "eval_samples_per_second": 381.765,
+      "eval_steps_per_second": 0.764,
+      "step": 58950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.409703186852278e-06,
+      "loss": 1.067,
+      "step": 59000
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.1076,
+      "eval_samples_per_second": 381.459,
+      "eval_steps_per_second": 0.763,
+      "step": 59000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.415135138705543e-06,
+      "loss": 1.0655,
+      "step": 59050
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.0378,
+      "eval_samples_per_second": 383.499,
+      "eval_steps_per_second": 0.767,
+      "step": 59050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.420567090558807e-06,
+      "loss": 1.0575,
+      "step": 59100
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.1689,
+      "eval_samples_per_second": 379.683,
+      "eval_steps_per_second": 0.759,
+      "step": 59100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.425999042412073e-06,
+      "loss": 1.0687,
+      "step": 59150
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.9456,
+      "eval_samples_per_second": 386.231,
+      "eval_steps_per_second": 0.772,
+      "step": 59150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.431430994265338e-06,
+      "loss": 1.0613,
+      "step": 59200
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.0167,
+      "eval_samples_per_second": 384.121,
+      "eval_steps_per_second": 0.768,
+      "step": 59200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.436862946118602e-06,
+      "loss": 1.0583,
+      "step": 59250
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.074,
+      "eval_samples_per_second": 382.437,
+      "eval_steps_per_second": 0.765,
+      "step": 59250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.442294897971867e-06,
+      "loss": 1.0604,
+      "step": 59300
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.846,
+      "eval_samples_per_second": 389.228,
+      "eval_steps_per_second": 0.778,
+      "step": 59300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.447726849825131e-06,
+      "loss": 1.065,
+      "step": 59350
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.9087,
+      "eval_samples_per_second": 387.337,
+      "eval_steps_per_second": 0.775,
+      "step": 59350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.453158801678396e-06,
+      "loss": 1.0618,
+      "step": 59400
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.0058,
+      "eval_samples_per_second": 384.443,
+      "eval_steps_per_second": 0.769,
+      "step": 59400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.45859075353166e-06,
+      "loss": 1.0608,
+      "step": 59450
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.1795,
+      "eval_samples_per_second": 379.377,
+      "eval_steps_per_second": 0.759,
+      "step": 59450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.464022705384924e-06,
+      "loss": 1.0619,
+      "step": 59500
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.7662,
+      "eval_samples_per_second": 391.66,
+      "eval_steps_per_second": 0.783,
+      "step": 59500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.46945465723819e-06,
+      "loss": 1.0625,
+      "step": 59550
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 13.011,
+      "eval_samples_per_second": 384.291,
+      "eval_steps_per_second": 0.769,
+      "step": 59550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.474886609091455e-06,
+      "loss": 1.0553,
+      "step": 59600
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.7655,
+      "eval_samples_per_second": 391.681,
+      "eval_steps_per_second": 0.783,
+      "step": 59600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.480318560944719e-06,
+      "loss": 1.0572,
+      "step": 59650
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.9176,
+      "eval_samples_per_second": 387.07,
+      "eval_steps_per_second": 0.774,
+      "step": 59650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.485750512797984e-06,
+      "loss": 1.0636,
+      "step": 59700
+    },
+    {
+      "epoch": 0.72,
+      "eval_runtime": 12.8997,
+      "eval_samples_per_second": 387.605,
+      "eval_steps_per_second": 0.775,
+      "step": 59700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.491182464651248e-06,
+      "loss": 1.0621,
+      "step": 59750
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.7534,
+      "eval_samples_per_second": 392.052,
+      "eval_steps_per_second": 0.784,
+      "step": 59750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.496614416504513e-06,
+      "loss": 1.0528,
+      "step": 59800
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.725,
+      "eval_samples_per_second": 392.927,
+      "eval_steps_per_second": 0.786,
+      "step": 59800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.502046368357777e-06,
+      "loss": 1.0522,
+      "step": 59850
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.7218,
+      "eval_samples_per_second": 393.028,
+      "eval_steps_per_second": 0.786,
+      "step": 59850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.5074783202110426e-06,
+      "loss": 1.0605,
+      "step": 59900
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.6931,
+      "eval_samples_per_second": 393.915,
+      "eval_steps_per_second": 0.788,
+      "step": 59900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.512910272064307e-06,
+      "loss": 1.0548,
+      "step": 59950
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.666,
+      "eval_samples_per_second": 394.758,
+      "eval_steps_per_second": 0.79,
+      "step": 59950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.518342223917572e-06,
+      "loss": 1.0516,
+      "step": 60000
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.7213,
+      "eval_samples_per_second": 393.043,
+      "eval_steps_per_second": 0.786,
+      "step": 60000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.523774175770836e-06,
+      "loss": 1.051,
+      "step": 60050
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 13.1245,
+      "eval_samples_per_second": 380.966,
+      "eval_steps_per_second": 0.762,
+      "step": 60050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.529206127624101e-06,
+      "loss": 1.0556,
+      "step": 60100
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.8789,
+      "eval_samples_per_second": 388.232,
+      "eval_steps_per_second": 0.776,
+      "step": 60100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.534638079477365e-06,
+      "loss": 1.0598,
+      "step": 60150
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 13.0215,
+      "eval_samples_per_second": 383.979,
+      "eval_steps_per_second": 0.768,
+      "step": 60150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.54007003133063e-06,
+      "loss": 1.058,
+      "step": 60200
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 13.014,
+      "eval_samples_per_second": 384.202,
+      "eval_steps_per_second": 0.768,
+      "step": 60200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.5455019831838955e-06,
+      "loss": 1.0551,
+      "step": 60250
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 13.0191,
+      "eval_samples_per_second": 384.052,
+      "eval_steps_per_second": 0.768,
+      "step": 60250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.5509339350371595e-06,
+      "loss": 1.0572,
+      "step": 60300
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.7234,
+      "eval_samples_per_second": 392.977,
+      "eval_steps_per_second": 0.786,
+      "step": 60300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.556365886890424e-06,
+      "loss": 1.054,
+      "step": 60350
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.9508,
+      "eval_samples_per_second": 386.077,
+      "eval_steps_per_second": 0.772,
+      "step": 60350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.5617978387436886e-06,
+      "loss": 1.0562,
+      "step": 60400
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.9097,
+      "eval_samples_per_second": 387.306,
+      "eval_steps_per_second": 0.775,
+      "step": 60400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.567229790596953e-06,
+      "loss": 1.0548,
+      "step": 60450
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 13.0603,
+      "eval_samples_per_second": 382.841,
+      "eval_steps_per_second": 0.766,
+      "step": 60450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.572661742450218e-06,
+      "loss": 1.0562,
+      "step": 60500
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.994,
+      "eval_samples_per_second": 384.792,
+      "eval_steps_per_second": 0.77,
+      "step": 60500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 6.578093694303482e-06,
+      "loss": 1.055,
+      "step": 60550
+    },
+    {
+      "epoch": 0.73,
+      "eval_runtime": 12.9519,
+      "eval_samples_per_second": 386.043,
+      "eval_steps_per_second": 0.772,
+      "step": 60550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5835256461567475e-06,
+      "loss": 1.0517,
+      "step": 60600
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.9421,
+      "eval_samples_per_second": 386.337,
+      "eval_steps_per_second": 0.773,
+      "step": 60600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.588957598010012e-06,
+      "loss": 1.0571,
+      "step": 60650
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.815,
+      "eval_samples_per_second": 390.167,
+      "eval_steps_per_second": 0.78,
+      "step": 60650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.5943895498632765e-06,
+      "loss": 1.044,
+      "step": 60700
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.0263,
+      "eval_samples_per_second": 383.839,
+      "eval_steps_per_second": 0.768,
+      "step": 60700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.599821501716541e-06,
+      "loss": 1.051,
+      "step": 60750
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.0652,
+      "eval_samples_per_second": 382.696,
+      "eval_steps_per_second": 0.765,
+      "step": 60750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6052534535698055e-06,
+      "loss": 1.059,
+      "step": 60800
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.1904,
+      "eval_samples_per_second": 379.065,
+      "eval_steps_per_second": 0.758,
+      "step": 60800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.61068540542307e-06,
+      "loss": 1.0484,
+      "step": 60850
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.0497,
+      "eval_samples_per_second": 383.15,
+      "eval_steps_per_second": 0.766,
+      "step": 60850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6161173572763346e-06,
+      "loss": 1.056,
+      "step": 60900
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.8845,
+      "eval_samples_per_second": 388.064,
+      "eval_steps_per_second": 0.776,
+      "step": 60900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6215493091296e-06,
+      "loss": 1.0539,
+      "step": 60950
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.7466,
+      "eval_samples_per_second": 392.26,
+      "eval_steps_per_second": 0.785,
+      "step": 60950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6269812609828645e-06,
+      "loss": 1.0473,
+      "step": 61000
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.0404,
+      "eval_samples_per_second": 383.424,
+      "eval_steps_per_second": 0.767,
+      "step": 61000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.632413212836129e-06,
+      "loss": 1.0526,
+      "step": 61050
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.7447,
+      "eval_samples_per_second": 392.32,
+      "eval_steps_per_second": 0.785,
+      "step": 61050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6378451646893935e-06,
+      "loss": 1.0559,
+      "step": 61100
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.0622,
+      "eval_samples_per_second": 382.783,
+      "eval_steps_per_second": 0.766,
+      "step": 61100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6432771165426576e-06,
+      "loss": 1.0476,
+      "step": 61150
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.989,
+      "eval_samples_per_second": 384.94,
+      "eval_steps_per_second": 0.77,
+      "step": 61150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6487090683959225e-06,
+      "loss": 1.0537,
+      "step": 61200
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.7544,
+      "eval_samples_per_second": 392.02,
+      "eval_steps_per_second": 0.784,
+      "step": 61200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.654141020249187e-06,
+      "loss": 1.0534,
+      "step": 61250
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.007,
+      "eval_samples_per_second": 384.407,
+      "eval_steps_per_second": 0.769,
+      "step": 61250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.6595729721024515e-06,
+      "loss": 1.046,
+      "step": 61300
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 13.311,
+      "eval_samples_per_second": 375.629,
+      "eval_steps_per_second": 0.751,
+      "step": 61300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 6.665004923955717e-06,
+      "loss": 1.05,
+      "step": 61350
+    },
+    {
+      "epoch": 0.74,
+      "eval_runtime": 12.9567,
+      "eval_samples_per_second": 385.902,
+      "eval_steps_per_second": 0.772,
+      "step": 61350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.670436875808981e-06,
+      "loss": 1.0524,
+      "step": 61400
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.9732,
+      "eval_samples_per_second": 385.409,
+      "eval_steps_per_second": 0.771,
+      "step": 61400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.675868827662246e-06,
+      "loss": 1.0499,
+      "step": 61450
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.7031,
+      "eval_samples_per_second": 393.604,
+      "eval_steps_per_second": 0.787,
+      "step": 61450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.6813007795155105e-06,
+      "loss": 1.0511,
+      "step": 61500
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.8034,
+      "eval_samples_per_second": 390.522,
+      "eval_steps_per_second": 0.781,
+      "step": 61500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.686732731368775e-06,
+      "loss": 1.0545,
+      "step": 61550
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.9683,
+      "eval_samples_per_second": 385.554,
+      "eval_steps_per_second": 0.771,
+      "step": 61550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.6921646832220395e-06,
+      "loss": 1.0453,
+      "step": 61600
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.9651,
+      "eval_samples_per_second": 385.651,
+      "eval_steps_per_second": 0.771,
+      "step": 61600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.6975966350753036e-06,
+      "loss": 1.0471,
+      "step": 61650
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0238,
+      "eval_samples_per_second": 383.913,
+      "eval_steps_per_second": 0.768,
+      "step": 61650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.703028586928569e-06,
+      "loss": 1.041,
+      "step": 61700
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0098,
+      "eval_samples_per_second": 384.326,
+      "eval_steps_per_second": 0.769,
+      "step": 61700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.708460538781834e-06,
+      "loss": 1.0465,
+      "step": 61750
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.9426,
+      "eval_samples_per_second": 386.32,
+      "eval_steps_per_second": 0.773,
+      "step": 61750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.713892490635098e-06,
+      "loss": 1.0473,
+      "step": 61800
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.9757,
+      "eval_samples_per_second": 385.336,
+      "eval_steps_per_second": 0.771,
+      "step": 61800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.719324442488363e-06,
+      "loss": 1.0508,
+      "step": 61850
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0546,
+      "eval_samples_per_second": 383.006,
+      "eval_steps_per_second": 0.766,
+      "step": 61850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.724756394341627e-06,
+      "loss": 1.0453,
+      "step": 61900
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0888,
+      "eval_samples_per_second": 382.006,
+      "eval_steps_per_second": 0.764,
+      "step": 61900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.730188346194892e-06,
+      "loss": 1.0444,
+      "step": 61950
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.8834,
+      "eval_samples_per_second": 388.097,
+      "eval_steps_per_second": 0.776,
+      "step": 61950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.7356202980481565e-06,
+      "loss": 1.0427,
+      "step": 62000
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0322,
+      "eval_samples_per_second": 383.666,
+      "eval_steps_per_second": 0.767,
+      "step": 62000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.741052249901422e-06,
+      "loss": 1.0515,
+      "step": 62050
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0298,
+      "eval_samples_per_second": 383.735,
+      "eval_steps_per_second": 0.767,
+      "step": 62050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.746484201754686e-06,
+      "loss": 1.0412,
+      "step": 62100
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 13.0133,
+      "eval_samples_per_second": 384.221,
+      "eval_steps_per_second": 0.768,
+      "step": 62100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.751916153607951e-06,
+      "loss": 1.0446,
+      "step": 62150
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.6797,
+      "eval_samples_per_second": 394.33,
+      "eval_steps_per_second": 0.789,
+      "step": 62150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 6.757348105461215e-06,
+      "loss": 1.0458,
+      "step": 62200
+    },
+    {
+      "epoch": 0.75,
+      "eval_runtime": 12.7005,
+      "eval_samples_per_second": 393.686,
+      "eval_steps_per_second": 0.787,
+      "step": 62200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.76278005731448e-06,
+      "loss": 1.0424,
+      "step": 62250
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9012,
+      "eval_samples_per_second": 387.562,
+      "eval_steps_per_second": 0.775,
+      "step": 62250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.768212009167744e-06,
+      "loss": 1.0462,
+      "step": 62300
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 13.0084,
+      "eval_samples_per_second": 384.368,
+      "eval_steps_per_second": 0.769,
+      "step": 62300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.773643961021009e-06,
+      "loss": 1.0459,
+      "step": 62350
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9311,
+      "eval_samples_per_second": 386.665,
+      "eval_steps_per_second": 0.773,
+      "step": 62350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.779075912874275e-06,
+      "loss": 1.0382,
+      "step": 62400
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.6725,
+      "eval_samples_per_second": 394.556,
+      "eval_steps_per_second": 0.789,
+      "step": 62400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.784507864727539e-06,
+      "loss": 1.0425,
+      "step": 62450
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 13.063,
+      "eval_samples_per_second": 382.762,
+      "eval_steps_per_second": 0.766,
+      "step": 62450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.789939816580803e-06,
+      "loss": 1.0439,
+      "step": 62500
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 13.1229,
+      "eval_samples_per_second": 381.015,
+      "eval_steps_per_second": 0.762,
+      "step": 62500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.795371768434068e-06,
+      "loss": 1.0401,
+      "step": 62550
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9998,
+      "eval_samples_per_second": 384.623,
+      "eval_steps_per_second": 0.769,
+      "step": 62550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.800803720287332e-06,
+      "loss": 1.0419,
+      "step": 62600
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9817,
+      "eval_samples_per_second": 385.159,
+      "eval_steps_per_second": 0.77,
+      "step": 62600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.806235672140597e-06,
+      "loss": 1.0427,
+      "step": 62650
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 13.0158,
+      "eval_samples_per_second": 384.149,
+      "eval_steps_per_second": 0.768,
+      "step": 62650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.811667623993861e-06,
+      "loss": 1.0426,
+      "step": 62700
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.7603,
+      "eval_samples_per_second": 391.84,
+      "eval_steps_per_second": 0.784,
+      "step": 62700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.817099575847127e-06,
+      "loss": 1.0449,
+      "step": 62750
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 13.0726,
+      "eval_samples_per_second": 382.479,
+      "eval_steps_per_second": 0.765,
+      "step": 62750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.822531527700392e-06,
+      "loss": 1.0411,
+      "step": 62800
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9217,
+      "eval_samples_per_second": 386.945,
+      "eval_steps_per_second": 0.774,
+      "step": 62800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.827963479553656e-06,
+      "loss": 1.0345,
+      "step": 62850
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9343,
+      "eval_samples_per_second": 386.568,
+      "eval_steps_per_second": 0.773,
+      "step": 62850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.83339543140692e-06,
+      "loss": 1.0449,
+      "step": 62900
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9013,
+      "eval_samples_per_second": 387.557,
+      "eval_steps_per_second": 0.775,
+      "step": 62900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.838827383260185e-06,
+      "loss": 1.0399,
+      "step": 62950
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 12.9263,
+      "eval_samples_per_second": 386.808,
+      "eval_steps_per_second": 0.774,
+      "step": 62950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.844259335113449e-06,
+      "loss": 1.0398,
+      "step": 63000
+    },
+    {
+      "epoch": 0.76,
+      "eval_runtime": 13.1021,
+      "eval_samples_per_second": 381.618,
+      "eval_steps_per_second": 0.763,
+      "step": 63000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.849691286966714e-06,
+      "loss": 1.0411,
+      "step": 63050
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.0502,
+      "eval_samples_per_second": 383.137,
+      "eval_steps_per_second": 0.766,
+      "step": 63050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.855123238819978e-06,
+      "loss": 1.0453,
+      "step": 63100
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.273,
+      "eval_samples_per_second": 376.706,
+      "eval_steps_per_second": 0.753,
+      "step": 63100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.860555190673244e-06,
+      "loss": 1.0395,
+      "step": 63150
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.0135,
+      "eval_samples_per_second": 384.217,
+      "eval_steps_per_second": 0.768,
+      "step": 63150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.865987142526509e-06,
+      "loss": 1.0393,
+      "step": 63200
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.0375,
+      "eval_samples_per_second": 383.51,
+      "eval_steps_per_second": 0.767,
+      "step": 63200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.871419094379773e-06,
+      "loss": 1.0376,
+      "step": 63250
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.1549,
+      "eval_samples_per_second": 380.086,
+      "eval_steps_per_second": 0.76,
+      "step": 63250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.876851046233038e-06,
+      "loss": 1.0412,
+      "step": 63300
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 12.9295,
+      "eval_samples_per_second": 386.712,
+      "eval_steps_per_second": 0.773,
+      "step": 63300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.882282998086302e-06,
+      "loss": 1.0377,
+      "step": 63350
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.065,
+      "eval_samples_per_second": 382.703,
+      "eval_steps_per_second": 0.765,
+      "step": 63350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.887714949939566e-06,
+      "loss": 1.0336,
+      "step": 63400
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.05,
+      "eval_samples_per_second": 383.142,
+      "eval_steps_per_second": 0.766,
+      "step": 63400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.893146901792831e-06,
+      "loss": 1.04,
+      "step": 63450
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.0032,
+      "eval_samples_per_second": 384.52,
+      "eval_steps_per_second": 0.769,
+      "step": 63450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.898578853646097e-06,
+      "loss": 1.0384,
+      "step": 63500
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 12.9166,
+      "eval_samples_per_second": 387.1,
+      "eval_steps_per_second": 0.774,
+      "step": 63500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.904010805499361e-06,
+      "loss": 1.0313,
+      "step": 63550
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.0941,
+      "eval_samples_per_second": 381.85,
+      "eval_steps_per_second": 0.764,
+      "step": 63550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.909442757352626e-06,
+      "loss": 1.0419,
+      "step": 63600
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.067,
+      "eval_samples_per_second": 382.643,
+      "eval_steps_per_second": 0.765,
+      "step": 63600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.91487470920589e-06,
+      "loss": 1.0356,
+      "step": 63650
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 12.9849,
+      "eval_samples_per_second": 385.062,
+      "eval_steps_per_second": 0.77,
+      "step": 63650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.920306661059155e-06,
+      "loss": 1.0364,
+      "step": 63700
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 12.8828,
+      "eval_samples_per_second": 388.113,
+      "eval_steps_per_second": 0.776,
+      "step": 63700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.925738612912419e-06,
+      "loss": 1.0408,
+      "step": 63750
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 13.0048,
+      "eval_samples_per_second": 384.474,
+      "eval_steps_per_second": 0.769,
+      "step": 63750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.931170564765683e-06,
+      "loss": 1.032,
+      "step": 63800
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 12.9849,
+      "eval_samples_per_second": 385.062,
+      "eval_steps_per_second": 0.77,
+      "step": 63800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.936602516618949e-06,
+      "loss": 1.0362,
+      "step": 63850
+    },
+    {
+      "epoch": 0.77,
+      "eval_runtime": 12.9159,
+      "eval_samples_per_second": 387.121,
+      "eval_steps_per_second": 0.774,
+      "step": 63850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.942034468472214e-06,
+      "loss": 1.033,
+      "step": 63900
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.9229,
+      "eval_samples_per_second": 386.911,
+      "eval_steps_per_second": 0.774,
+      "step": 63900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.947466420325478e-06,
+      "loss": 1.0346,
+      "step": 63950
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0063,
+      "eval_samples_per_second": 384.429,
+      "eval_steps_per_second": 0.769,
+      "step": 63950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.952898372178743e-06,
+      "loss": 1.031,
+      "step": 64000
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.9859,
+      "eval_samples_per_second": 385.034,
+      "eval_steps_per_second": 0.77,
+      "step": 64000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.958330324032007e-06,
+      "loss": 1.0344,
+      "step": 64050
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0013,
+      "eval_samples_per_second": 384.577,
+      "eval_steps_per_second": 0.769,
+      "step": 64050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.963762275885272e-06,
+      "loss": 1.0335,
+      "step": 64100
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.9676,
+      "eval_samples_per_second": 385.576,
+      "eval_steps_per_second": 0.771,
+      "step": 64100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.969194227738536e-06,
+      "loss": 1.0296,
+      "step": 64150
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0951,
+      "eval_samples_per_second": 381.821,
+      "eval_steps_per_second": 0.764,
+      "step": 64150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.974626179591802e-06,
+      "loss": 1.0293,
+      "step": 64200
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.9548,
+      "eval_samples_per_second": 385.957,
+      "eval_steps_per_second": 0.772,
+      "step": 64200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.980058131445066e-06,
+      "loss": 1.0322,
+      "step": 64250
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0667,
+      "eval_samples_per_second": 382.652,
+      "eval_steps_per_second": 0.765,
+      "step": 64250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.985490083298331e-06,
+      "loss": 1.0331,
+      "step": 64300
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.7817,
+      "eval_samples_per_second": 391.185,
+      "eval_steps_per_second": 0.782,
+      "step": 64300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.990922035151595e-06,
+      "loss": 1.0361,
+      "step": 64350
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0412,
+      "eval_samples_per_second": 383.4,
+      "eval_steps_per_second": 0.767,
+      "step": 64350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.99635398700486e-06,
+      "loss": 1.031,
+      "step": 64400
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0767,
+      "eval_samples_per_second": 382.36,
+      "eval_steps_per_second": 0.765,
+      "step": 64400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.001785938858124e-06,
+      "loss": 1.0346,
+      "step": 64450
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 13.0376,
+      "eval_samples_per_second": 383.507,
+      "eval_steps_per_second": 0.767,
+      "step": 64450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.007217890711389e-06,
+      "loss": 1.0353,
+      "step": 64500
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.8593,
+      "eval_samples_per_second": 388.823,
+      "eval_steps_per_second": 0.778,
+      "step": 64500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.012649842564655e-06,
+      "loss": 1.0258,
+      "step": 64550
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.7684,
+      "eval_samples_per_second": 391.592,
+      "eval_steps_per_second": 0.783,
+      "step": 64550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.018081794417919e-06,
+      "loss": 1.0313,
+      "step": 64600
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.7159,
+      "eval_samples_per_second": 393.21,
+      "eval_steps_per_second": 0.786,
+      "step": 64600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.023513746271183e-06,
+      "loss": 1.0323,
+      "step": 64650
+    },
+    {
+      "epoch": 0.78,
+      "eval_runtime": 12.7398,
+      "eval_samples_per_second": 392.47,
+      "eval_steps_per_second": 0.785,
+      "step": 64650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.028945698124448e-06,
+      "loss": 1.0343,
+      "step": 64700
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.958,
+      "eval_samples_per_second": 385.863,
+      "eval_steps_per_second": 0.772,
+      "step": 64700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.034377649977712e-06,
+      "loss": 1.0275,
+      "step": 64750
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.0818,
+      "eval_samples_per_second": 382.21,
+      "eval_steps_per_second": 0.764,
+      "step": 64750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.039809601830977e-06,
+      "loss": 1.0327,
+      "step": 64800
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.0514,
+      "eval_samples_per_second": 383.1,
+      "eval_steps_per_second": 0.766,
+      "step": 64800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.045241553684241e-06,
+      "loss": 1.0271,
+      "step": 64850
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.9003,
+      "eval_samples_per_second": 387.587,
+      "eval_steps_per_second": 0.775,
+      "step": 64850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.050673505537507e-06,
+      "loss": 1.0282,
+      "step": 64900
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.1502,
+      "eval_samples_per_second": 380.223,
+      "eval_steps_per_second": 0.76,
+      "step": 64900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.056105457390772e-06,
+      "loss": 1.0222,
+      "step": 64950
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.9415,
+      "eval_samples_per_second": 386.354,
+      "eval_steps_per_second": 0.773,
+      "step": 64950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.061537409244036e-06,
+      "loss": 1.0247,
+      "step": 65000
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.9533,
+      "eval_samples_per_second": 386.001,
+      "eval_steps_per_second": 0.772,
+      "step": 65000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.0669693610973e-06,
+      "loss": 1.0351,
+      "step": 65050
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.8303,
+      "eval_samples_per_second": 389.701,
+      "eval_steps_per_second": 0.779,
+      "step": 65050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.072401312950565e-06,
+      "loss": 1.0301,
+      "step": 65100
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.7881,
+      "eval_samples_per_second": 390.989,
+      "eval_steps_per_second": 0.782,
+      "step": 65100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.077833264803829e-06,
+      "loss": 1.0247,
+      "step": 65150
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.7277,
+      "eval_samples_per_second": 392.844,
+      "eval_steps_per_second": 0.786,
+      "step": 65150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.083265216657094e-06,
+      "loss": 1.0256,
+      "step": 65200
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.9297,
+      "eval_samples_per_second": 386.707,
+      "eval_steps_per_second": 0.773,
+      "step": 65200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.088697168510358e-06,
+      "loss": 1.0304,
+      "step": 65250
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.0285,
+      "eval_samples_per_second": 383.773,
+      "eval_steps_per_second": 0.768,
+      "step": 65250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.094129120363624e-06,
+      "loss": 1.0299,
+      "step": 65300
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.1522,
+      "eval_samples_per_second": 380.166,
+      "eval_steps_per_second": 0.76,
+      "step": 65300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.099561072216889e-06,
+      "loss": 1.028,
+      "step": 65350
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.0253,
+      "eval_samples_per_second": 383.868,
+      "eval_steps_per_second": 0.768,
+      "step": 65350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.104993024070153e-06,
+      "loss": 1.0245,
+      "step": 65400
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 13.0453,
+      "eval_samples_per_second": 383.28,
+      "eval_steps_per_second": 0.767,
+      "step": 65400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.110424975923418e-06,
+      "loss": 1.0332,
+      "step": 65450
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.9053,
+      "eval_samples_per_second": 387.438,
+      "eval_steps_per_second": 0.775,
+      "step": 65450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.115856927776682e-06,
+      "loss": 1.0267,
+      "step": 65500
+    },
+    {
+      "epoch": 0.79,
+      "eval_runtime": 12.8737,
+      "eval_samples_per_second": 388.388,
+      "eval_steps_per_second": 0.777,
+      "step": 65500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.121288879629946e-06,
+      "loss": 1.0249,
+      "step": 65550
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.06,
+      "eval_samples_per_second": 382.849,
+      "eval_steps_per_second": 0.766,
+      "step": 65550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.126720831483211e-06,
+      "loss": 1.0201,
+      "step": 65600
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.1096,
+      "eval_samples_per_second": 381.399,
+      "eval_steps_per_second": 0.763,
+      "step": 65600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.132152783336477e-06,
+      "loss": 1.0244,
+      "step": 65650
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.8899,
+      "eval_samples_per_second": 387.902,
+      "eval_steps_per_second": 0.776,
+      "step": 65650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.137584735189741e-06,
+      "loss": 1.0312,
+      "step": 65700
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.7492,
+      "eval_samples_per_second": 392.182,
+      "eval_steps_per_second": 0.784,
+      "step": 65700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.143016687043006e-06,
+      "loss": 1.0246,
+      "step": 65750
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.7143,
+      "eval_samples_per_second": 393.258,
+      "eval_steps_per_second": 0.787,
+      "step": 65750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.14844863889627e-06,
+      "loss": 1.0216,
+      "step": 65800
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.7696,
+      "eval_samples_per_second": 391.556,
+      "eval_steps_per_second": 0.783,
+      "step": 65800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.153880590749535e-06,
+      "loss": 1.0233,
+      "step": 65850
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.0569,
+      "eval_samples_per_second": 382.939,
+      "eval_steps_per_second": 0.766,
+      "step": 65850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.159312542602799e-06,
+      "loss": 1.0226,
+      "step": 65900
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.0591,
+      "eval_samples_per_second": 382.874,
+      "eval_steps_per_second": 0.766,
+      "step": 65900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.164744494456063e-06,
+      "loss": 1.0256,
+      "step": 65950
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.8481,
+      "eval_samples_per_second": 389.163,
+      "eval_steps_per_second": 0.778,
+      "step": 65950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.170176446309329e-06,
+      "loss": 1.0216,
+      "step": 66000
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.8885,
+      "eval_samples_per_second": 387.942,
+      "eval_steps_per_second": 0.776,
+      "step": 66000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.175608398162594e-06,
+      "loss": 1.0249,
+      "step": 66050
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.9179,
+      "eval_samples_per_second": 387.06,
+      "eval_steps_per_second": 0.774,
+      "step": 66050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.181040350015858e-06,
+      "loss": 1.0163,
+      "step": 66100
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 12.9466,
+      "eval_samples_per_second": 386.202,
+      "eval_steps_per_second": 0.772,
+      "step": 66100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.186472301869123e-06,
+      "loss": 1.0201,
+      "step": 66150
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.1272,
+      "eval_samples_per_second": 380.888,
+      "eval_steps_per_second": 0.762,
+      "step": 66150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.191904253722387e-06,
+      "loss": 1.0205,
+      "step": 66200
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.0387,
+      "eval_samples_per_second": 383.473,
+      "eval_steps_per_second": 0.767,
+      "step": 66200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.197336205575652e-06,
+      "loss": 1.0214,
+      "step": 66250
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.1189,
+      "eval_samples_per_second": 381.129,
+      "eval_steps_per_second": 0.762,
+      "step": 66250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.202768157428916e-06,
+      "loss": 1.0165,
+      "step": 66300
+    },
+    {
+      "epoch": 0.8,
+      "eval_runtime": 13.1113,
+      "eval_samples_per_second": 381.35,
+      "eval_steps_per_second": 0.763,
+      "step": 66300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.2082001092821816e-06,
+      "loss": 1.0235,
+      "step": 66350
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.7485,
+      "eval_samples_per_second": 392.204,
+      "eval_steps_per_second": 0.784,
+      "step": 66350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.213632061135446e-06,
+      "loss": 1.0233,
+      "step": 66400
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.7357,
+      "eval_samples_per_second": 392.596,
+      "eval_steps_per_second": 0.785,
+      "step": 66400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.219064012988711e-06,
+      "loss": 1.0244,
+      "step": 66450
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.7422,
+      "eval_samples_per_second": 392.396,
+      "eval_steps_per_second": 0.785,
+      "step": 66450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.224495964841975e-06,
+      "loss": 1.0192,
+      "step": 66500
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.9262,
+      "eval_samples_per_second": 386.81,
+      "eval_steps_per_second": 0.774,
+      "step": 66500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.22992791669524e-06,
+      "loss": 1.0207,
+      "step": 66550
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.9091,
+      "eval_samples_per_second": 387.325,
+      "eval_steps_per_second": 0.775,
+      "step": 66550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.235359868548504e-06,
+      "loss": 1.0158,
+      "step": 66600
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.9037,
+      "eval_samples_per_second": 387.485,
+      "eval_steps_per_second": 0.775,
+      "step": 66600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.240791820401769e-06,
+      "loss": 1.018,
+      "step": 66650
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 13.0131,
+      "eval_samples_per_second": 384.228,
+      "eval_steps_per_second": 0.768,
+      "step": 66650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.2462237722550344e-06,
+      "loss": 1.0185,
+      "step": 66700
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.9936,
+      "eval_samples_per_second": 384.805,
+      "eval_steps_per_second": 0.77,
+      "step": 66700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.2516557241082985e-06,
+      "loss": 1.0222,
+      "step": 66750
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.9605,
+      "eval_samples_per_second": 385.787,
+      "eval_steps_per_second": 0.772,
+      "step": 66750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.257087675961563e-06,
+      "loss": 1.0173,
+      "step": 66800
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 13.27,
+      "eval_samples_per_second": 376.791,
+      "eval_steps_per_second": 0.754,
+      "step": 66800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.2625196278148276e-06,
+      "loss": 1.0206,
+      "step": 66850
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 13.2324,
+      "eval_samples_per_second": 377.86,
+      "eval_steps_per_second": 0.756,
+      "step": 66850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.267951579668092e-06,
+      "loss": 1.0207,
+      "step": 66900
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 13.1767,
+      "eval_samples_per_second": 379.459,
+      "eval_steps_per_second": 0.759,
+      "step": 66900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.273383531521357e-06,
+      "loss": 1.0198,
+      "step": 66950
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.8775,
+      "eval_samples_per_second": 388.275,
+      "eval_steps_per_second": 0.777,
+      "step": 66950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.278815483374621e-06,
+      "loss": 1.0161,
+      "step": 67000
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.961,
+      "eval_samples_per_second": 385.773,
+      "eval_steps_per_second": 0.772,
+      "step": 67000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.284247435227886e-06,
+      "loss": 1.0164,
+      "step": 67050
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 12.7423,
+      "eval_samples_per_second": 392.394,
+      "eval_steps_per_second": 0.785,
+      "step": 67050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.289679387081151e-06,
+      "loss": 1.0148,
+      "step": 67100
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 13.1296,
+      "eval_samples_per_second": 380.818,
+      "eval_steps_per_second": 0.762,
+      "step": 67100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.2951113389344155e-06,
+      "loss": 1.0204,
+      "step": 67150
+    },
+    {
+      "epoch": 0.81,
+      "eval_runtime": 13.0579,
+      "eval_samples_per_second": 382.911,
+      "eval_steps_per_second": 0.766,
+      "step": 67150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.30054329078768e-06,
+      "loss": 1.0119,
+      "step": 67200
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 13.1006,
+      "eval_samples_per_second": 381.661,
+      "eval_steps_per_second": 0.763,
+      "step": 67200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3059752426409445e-06,
+      "loss": 1.0228,
+      "step": 67250
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 13.1681,
+      "eval_samples_per_second": 379.706,
+      "eval_steps_per_second": 0.759,
+      "step": 67250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.311407194494209e-06,
+      "loss": 1.0168,
+      "step": 67300
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 13.059,
+      "eval_samples_per_second": 382.878,
+      "eval_steps_per_second": 0.766,
+      "step": 67300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3168391463474736e-06,
+      "loss": 1.0181,
+      "step": 67350
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 13.0255,
+      "eval_samples_per_second": 383.864,
+      "eval_steps_per_second": 0.768,
+      "step": 67350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.322271098200738e-06,
+      "loss": 1.0108,
+      "step": 67400
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.7984,
+      "eval_samples_per_second": 390.674,
+      "eval_steps_per_second": 0.781,
+      "step": 67400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3277030500540034e-06,
+      "loss": 1.0132,
+      "step": 67450
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.9764,
+      "eval_samples_per_second": 385.316,
+      "eval_steps_per_second": 0.771,
+      "step": 67450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.333135001907268e-06,
+      "loss": 1.0175,
+      "step": 67500
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.7296,
+      "eval_samples_per_second": 392.787,
+      "eval_steps_per_second": 0.786,
+      "step": 67500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3385669537605325e-06,
+      "loss": 1.0192,
+      "step": 67550
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 13.0762,
+      "eval_samples_per_second": 382.374,
+      "eval_steps_per_second": 0.765,
+      "step": 67550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.343998905613797e-06,
+      "loss": 1.0187,
+      "step": 67600
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.9306,
+      "eval_samples_per_second": 386.68,
+      "eval_steps_per_second": 0.773,
+      "step": 67600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3494308574670615e-06,
+      "loss": 1.0205,
+      "step": 67650
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.8797,
+      "eval_samples_per_second": 388.207,
+      "eval_steps_per_second": 0.776,
+      "step": 67650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.354862809320326e-06,
+      "loss": 1.0081,
+      "step": 67700
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.8136,
+      "eval_samples_per_second": 390.21,
+      "eval_steps_per_second": 0.78,
+      "step": 67700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3602947611735905e-06,
+      "loss": 1.0204,
+      "step": 67750
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.9577,
+      "eval_samples_per_second": 385.87,
+      "eval_steps_per_second": 0.772,
+      "step": 67750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.365726713026856e-06,
+      "loss": 1.0172,
+      "step": 67800
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.9543,
+      "eval_samples_per_second": 385.972,
+      "eval_steps_per_second": 0.772,
+      "step": 67800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.37115866488012e-06,
+      "loss": 1.014,
+      "step": 67850
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.909,
+      "eval_samples_per_second": 387.326,
+      "eval_steps_per_second": 0.775,
+      "step": 67850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.376590616733385e-06,
+      "loss": 1.0174,
+      "step": 67900
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.9074,
+      "eval_samples_per_second": 387.376,
+      "eval_steps_per_second": 0.775,
+      "step": 67900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.3820225685866494e-06,
+      "loss": 1.0171,
+      "step": 67950
+    },
+    {
+      "epoch": 0.82,
+      "eval_runtime": 12.9118,
+      "eval_samples_per_second": 387.241,
+      "eval_steps_per_second": 0.774,
+      "step": 67950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.387454520439914e-06,
+      "loss": 1.017,
+      "step": 68000
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.1244,
+      "eval_samples_per_second": 380.969,
+      "eval_steps_per_second": 0.762,
+      "step": 68000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3928864722931785e-06,
+      "loss": 1.0123,
+      "step": 68050
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9502,
+      "eval_samples_per_second": 386.094,
+      "eval_steps_per_second": 0.772,
+      "step": 68050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3983184241464426e-06,
+      "loss": 1.0114,
+      "step": 68100
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.0083,
+      "eval_samples_per_second": 384.371,
+      "eval_steps_per_second": 0.769,
+      "step": 68100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.403750375999708e-06,
+      "loss": 1.0104,
+      "step": 68150
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9061,
+      "eval_samples_per_second": 387.415,
+      "eval_steps_per_second": 0.775,
+      "step": 68150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.409182327852973e-06,
+      "loss": 1.0157,
+      "step": 68200
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9138,
+      "eval_samples_per_second": 387.181,
+      "eval_steps_per_second": 0.774,
+      "step": 68200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.414614279706237e-06,
+      "loss": 1.0121,
+      "step": 68250
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.0493,
+      "eval_samples_per_second": 383.162,
+      "eval_steps_per_second": 0.766,
+      "step": 68250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.420046231559502e-06,
+      "loss": 1.015,
+      "step": 68300
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.8297,
+      "eval_samples_per_second": 389.72,
+      "eval_steps_per_second": 0.779,
+      "step": 68300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.425478183412766e-06,
+      "loss": 1.0165,
+      "step": 68350
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.0139,
+      "eval_samples_per_second": 384.204,
+      "eval_steps_per_second": 0.768,
+      "step": 68350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.430910135266031e-06,
+      "loss": 1.0148,
+      "step": 68400
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9409,
+      "eval_samples_per_second": 386.372,
+      "eval_steps_per_second": 0.773,
+      "step": 68400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4363420871192954e-06,
+      "loss": 1.0096,
+      "step": 68450
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.7856,
+      "eval_samples_per_second": 391.064,
+      "eval_steps_per_second": 0.782,
+      "step": 68450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.441774038972561e-06,
+      "loss": 1.0137,
+      "step": 68500
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9079,
+      "eval_samples_per_second": 387.359,
+      "eval_steps_per_second": 0.775,
+      "step": 68500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.447205990825825e-06,
+      "loss": 1.0062,
+      "step": 68550
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.0477,
+      "eval_samples_per_second": 383.208,
+      "eval_steps_per_second": 0.766,
+      "step": 68550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.45263794267909e-06,
+      "loss": 1.0152,
+      "step": 68600
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9679,
+      "eval_samples_per_second": 385.568,
+      "eval_steps_per_second": 0.771,
+      "step": 68600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.458069894532354e-06,
+      "loss": 1.0069,
+      "step": 68650
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9905,
+      "eval_samples_per_second": 384.895,
+      "eval_steps_per_second": 0.77,
+      "step": 68650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.463501846385619e-06,
+      "loss": 1.0138,
+      "step": 68700
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.037,
+      "eval_samples_per_second": 383.523,
+      "eval_steps_per_second": 0.767,
+      "step": 68700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.468933798238883e-06,
+      "loss": 1.0111,
+      "step": 68750
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 12.9966,
+      "eval_samples_per_second": 384.717,
+      "eval_steps_per_second": 0.769,
+      "step": 68750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.474365750092148e-06,
+      "loss": 1.0092,
+      "step": 68800
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 13.0454,
+      "eval_samples_per_second": 383.276,
+      "eval_steps_per_second": 0.767,
+      "step": 68800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.479797701945412e-06,
+      "loss": 1.0038,
+      "step": 68850
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.8899,
+      "eval_samples_per_second": 387.901,
+      "eval_steps_per_second": 0.776,
+      "step": 68850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.485229653798678e-06,
+      "loss": 1.0075,
+      "step": 68900
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 13.0982,
+      "eval_samples_per_second": 381.733,
+      "eval_steps_per_second": 0.763,
+      "step": 68900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.490661605651942e-06,
+      "loss": 1.0089,
+      "step": 68950
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 13.0578,
+      "eval_samples_per_second": 382.913,
+      "eval_steps_per_second": 0.766,
+      "step": 68950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.496093557505207e-06,
+      "loss": 1.0098,
+      "step": 69000
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.9717,
+      "eval_samples_per_second": 385.456,
+      "eval_steps_per_second": 0.771,
+      "step": 69000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.501525509358471e-06,
+      "loss": 1.008,
+      "step": 69050
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.9276,
+      "eval_samples_per_second": 386.771,
+      "eval_steps_per_second": 0.774,
+      "step": 69050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.506957461211736e-06,
+      "loss": 1.0076,
+      "step": 69100
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.8869,
+      "eval_samples_per_second": 387.991,
+      "eval_steps_per_second": 0.776,
+      "step": 69100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.512389413065e-06,
+      "loss": 1.0061,
+      "step": 69150
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.9833,
+      "eval_samples_per_second": 385.111,
+      "eval_steps_per_second": 0.77,
+      "step": 69150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.517821364918265e-06,
+      "loss": 1.0092,
+      "step": 69200
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.7075,
+      "eval_samples_per_second": 393.467,
+      "eval_steps_per_second": 0.787,
+      "step": 69200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.523253316771531e-06,
+      "loss": 1.0067,
+      "step": 69250
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.9219,
+      "eval_samples_per_second": 386.94,
+      "eval_steps_per_second": 0.774,
+      "step": 69250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.528685268624795e-06,
+      "loss": 1.01,
+      "step": 69300
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.9386,
+      "eval_samples_per_second": 386.44,
+      "eval_steps_per_second": 0.773,
+      "step": 69300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.53411722047806e-06,
+      "loss": 1.0026,
+      "step": 69350
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.7549,
+      "eval_samples_per_second": 392.006,
+      "eval_steps_per_second": 0.784,
+      "step": 69350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.539549172331324e-06,
+      "loss": 1.0086,
+      "step": 69400
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.8544,
+      "eval_samples_per_second": 388.972,
+      "eval_steps_per_second": 0.778,
+      "step": 69400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.544981124184588e-06,
+      "loss": 1.0093,
+      "step": 69450
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 13.0481,
+      "eval_samples_per_second": 383.199,
+      "eval_steps_per_second": 0.766,
+      "step": 69450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.550413076037853e-06,
+      "loss": 1.009,
+      "step": 69500
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 13.0449,
+      "eval_samples_per_second": 383.292,
+      "eval_steps_per_second": 0.767,
+      "step": 69500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.555845027891117e-06,
+      "loss": 1.0087,
+      "step": 69550
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 12.9488,
+      "eval_samples_per_second": 386.135,
+      "eval_steps_per_second": 0.772,
+      "step": 69550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.561276979744383e-06,
+      "loss": 1.0081,
+      "step": 69600
+    },
+    {
+      "epoch": 0.84,
+      "eval_runtime": 13.1082,
+      "eval_samples_per_second": 381.441,
+      "eval_steps_per_second": 0.763,
+      "step": 69600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.566708931597648e-06,
+      "loss": 1.0005,
+      "step": 69650
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.0722,
+      "eval_samples_per_second": 382.49,
+      "eval_steps_per_second": 0.765,
+      "step": 69650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.572140883450912e-06,
+      "loss": 1.0068,
+      "step": 69700
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.134,
+      "eval_samples_per_second": 380.692,
+      "eval_steps_per_second": 0.761,
+      "step": 69700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.577572835304177e-06,
+      "loss": 1.0077,
+      "step": 69750
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.7972,
+      "eval_samples_per_second": 390.711,
+      "eval_steps_per_second": 0.781,
+      "step": 69750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.583004787157441e-06,
+      "loss": 1.0046,
+      "step": 69800
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.0903,
+      "eval_samples_per_second": 381.962,
+      "eval_steps_per_second": 0.764,
+      "step": 69800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.588436739010705e-06,
+      "loss": 1.0082,
+      "step": 69850
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.9589,
+      "eval_samples_per_second": 385.836,
+      "eval_steps_per_second": 0.772,
+      "step": 69850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.59386869086397e-06,
+      "loss": 1.0053,
+      "step": 69900
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.0652,
+      "eval_samples_per_second": 382.697,
+      "eval_steps_per_second": 0.765,
+      "step": 69900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.599300642717236e-06,
+      "loss": 1.0043,
+      "step": 69950
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.0796,
+      "eval_samples_per_second": 382.273,
+      "eval_steps_per_second": 0.765,
+      "step": 69950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.6047325945705e-06,
+      "loss": 1.0071,
+      "step": 70000
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.7661,
+      "eval_samples_per_second": 391.662,
+      "eval_steps_per_second": 0.783,
+      "step": 70000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.610164546423765e-06,
+      "loss": 1.0052,
+      "step": 70050
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.7063,
+      "eval_samples_per_second": 393.506,
+      "eval_steps_per_second": 0.787,
+      "step": 70050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.615596498277029e-06,
+      "loss": 1.008,
+      "step": 70100
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.0847,
+      "eval_samples_per_second": 382.125,
+      "eval_steps_per_second": 0.764,
+      "step": 70100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.621028450130294e-06,
+      "loss": 1.0029,
+      "step": 70150
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.7232,
+      "eval_samples_per_second": 392.984,
+      "eval_steps_per_second": 0.786,
+      "step": 70150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.626460401983558e-06,
+      "loss": 1.0003,
+      "step": 70200
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.8029,
+      "eval_samples_per_second": 390.538,
+      "eval_steps_per_second": 0.781,
+      "step": 70200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.631892353836823e-06,
+      "loss": 1.0027,
+      "step": 70250
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.7516,
+      "eval_samples_per_second": 392.107,
+      "eval_steps_per_second": 0.784,
+      "step": 70250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.637324305690089e-06,
+      "loss": 0.9994,
+      "step": 70300
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.7071,
+      "eval_samples_per_second": 393.482,
+      "eval_steps_per_second": 0.787,
+      "step": 70300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.642756257543353e-06,
+      "loss": 1.0002,
+      "step": 70350
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 13.1548,
+      "eval_samples_per_second": 380.089,
+      "eval_steps_per_second": 0.76,
+      "step": 70350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.648188209396617e-06,
+      "loss": 1.0026,
+      "step": 70400
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.9774,
+      "eval_samples_per_second": 385.284,
+      "eval_steps_per_second": 0.771,
+      "step": 70400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.653620161249881e-06,
+      "loss": 1.0096,
+      "step": 70450
+    },
+    {
+      "epoch": 0.85,
+      "eval_runtime": 12.9221,
+      "eval_samples_per_second": 386.934,
+      "eval_steps_per_second": 0.774,
+      "step": 70450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.659052113103147e-06,
+      "loss": 1.0038,
+      "step": 70500
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0012,
+      "eval_samples_per_second": 384.579,
+      "eval_steps_per_second": 0.769,
+      "step": 70500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.664484064956411e-06,
+      "loss": 1.0088,
+      "step": 70550
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.7479,
+      "eval_samples_per_second": 392.221,
+      "eval_steps_per_second": 0.784,
+      "step": 70550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.669916016809675e-06,
+      "loss": 1.0002,
+      "step": 70600
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.9913,
+      "eval_samples_per_second": 384.872,
+      "eval_steps_per_second": 0.77,
+      "step": 70600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.67534796866294e-06,
+      "loss": 0.9941,
+      "step": 70650
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0224,
+      "eval_samples_per_second": 383.955,
+      "eval_steps_per_second": 0.768,
+      "step": 70650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.680779920516205e-06,
+      "loss": 1.0009,
+      "step": 70700
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.8755,
+      "eval_samples_per_second": 388.335,
+      "eval_steps_per_second": 0.777,
+      "step": 70700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.686211872369469e-06,
+      "loss": 1.0015,
+      "step": 70750
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.7022,
+      "eval_samples_per_second": 393.634,
+      "eval_steps_per_second": 0.787,
+      "step": 70750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.691643824222735e-06,
+      "loss": 1.0002,
+      "step": 70800
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.7926,
+      "eval_samples_per_second": 390.852,
+      "eval_steps_per_second": 0.782,
+      "step": 70800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.697075776075999e-06,
+      "loss": 1.004,
+      "step": 70850
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.7609,
+      "eval_samples_per_second": 391.823,
+      "eval_steps_per_second": 0.784,
+      "step": 70850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.702507727929263e-06,
+      "loss": 0.9994,
+      "step": 70900
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0382,
+      "eval_samples_per_second": 383.488,
+      "eval_steps_per_second": 0.767,
+      "step": 70900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.707939679782527e-06,
+      "loss": 0.9984,
+      "step": 70950
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.9107,
+      "eval_samples_per_second": 387.275,
+      "eval_steps_per_second": 0.775,
+      "step": 70950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.713371631635793e-06,
+      "loss": 1.0025,
+      "step": 71000
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.746,
+      "eval_samples_per_second": 392.279,
+      "eval_steps_per_second": 0.785,
+      "step": 71000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.718803583489059e-06,
+      "loss": 0.9979,
+      "step": 71050
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0005,
+      "eval_samples_per_second": 384.602,
+      "eval_steps_per_second": 0.769,
+      "step": 71050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.724235535342323e-06,
+      "loss": 0.9985,
+      "step": 71100
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0426,
+      "eval_samples_per_second": 383.358,
+      "eval_steps_per_second": 0.767,
+      "step": 71100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.729667487195587e-06,
+      "loss": 0.9994,
+      "step": 71150
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0552,
+      "eval_samples_per_second": 382.99,
+      "eval_steps_per_second": 0.766,
+      "step": 71150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.735099439048851e-06,
+      "loss": 0.9995,
+      "step": 71200
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 12.79,
+      "eval_samples_per_second": 390.931,
+      "eval_steps_per_second": 0.782,
+      "step": 71200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.740531390902115e-06,
+      "loss": 0.9972,
+      "step": 71250
+    },
+    {
+      "epoch": 0.86,
+      "eval_runtime": 13.0489,
+      "eval_samples_per_second": 383.175,
+      "eval_steps_per_second": 0.766,
+      "step": 71250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.745963342755381e-06,
+      "loss": 0.9964,
+      "step": 71300
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.1539,
+      "eval_samples_per_second": 380.117,
+      "eval_steps_per_second": 0.76,
+      "step": 71300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.751395294608645e-06,
+      "loss": 0.9987,
+      "step": 71350
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.8118,
+      "eval_samples_per_second": 390.264,
+      "eval_steps_per_second": 0.781,
+      "step": 71350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.75682724646191e-06,
+      "loss": 0.996,
+      "step": 71400
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.7953,
+      "eval_samples_per_second": 390.768,
+      "eval_steps_per_second": 0.782,
+      "step": 71400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.762259198315175e-06,
+      "loss": 0.9971,
+      "step": 71450
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.1356,
+      "eval_samples_per_second": 380.646,
+      "eval_steps_per_second": 0.761,
+      "step": 71450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.767691150168439e-06,
+      "loss": 0.997,
+      "step": 71500
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.9826,
+      "eval_samples_per_second": 385.131,
+      "eval_steps_per_second": 0.77,
+      "step": 71500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.773123102021703e-06,
+      "loss": 0.9968,
+      "step": 71550
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.9394,
+      "eval_samples_per_second": 386.417,
+      "eval_steps_per_second": 0.773,
+      "step": 71550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.778555053874969e-06,
+      "loss": 0.997,
+      "step": 71600
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.9057,
+      "eval_samples_per_second": 387.426,
+      "eval_steps_per_second": 0.775,
+      "step": 71600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.783987005728233e-06,
+      "loss": 1.0024,
+      "step": 71650
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.9045,
+      "eval_samples_per_second": 387.462,
+      "eval_steps_per_second": 0.775,
+      "step": 71650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.789418957581497e-06,
+      "loss": 0.9953,
+      "step": 71700
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.0694,
+      "eval_samples_per_second": 382.572,
+      "eval_steps_per_second": 0.765,
+      "step": 71700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.794850909434763e-06,
+      "loss": 0.9962,
+      "step": 71750
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.0068,
+      "eval_samples_per_second": 384.414,
+      "eval_steps_per_second": 0.769,
+      "step": 71750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.800282861288027e-06,
+      "loss": 1.0001,
+      "step": 71800
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.0546,
+      "eval_samples_per_second": 383.008,
+      "eval_steps_per_second": 0.766,
+      "step": 71800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.805714813141293e-06,
+      "loss": 0.9937,
+      "step": 71850
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.1213,
+      "eval_samples_per_second": 381.059,
+      "eval_steps_per_second": 0.762,
+      "step": 71850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.811146764994557e-06,
+      "loss": 0.9963,
+      "step": 71900
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.8813,
+      "eval_samples_per_second": 388.159,
+      "eval_steps_per_second": 0.776,
+      "step": 71900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.81657871684782e-06,
+      "loss": 0.9926,
+      "step": 71950
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.1316,
+      "eval_samples_per_second": 380.761,
+      "eval_steps_per_second": 0.762,
+      "step": 71950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.822010668701085e-06,
+      "loss": 0.9967,
+      "step": 72000
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 12.7156,
+      "eval_samples_per_second": 393.219,
+      "eval_steps_per_second": 0.786,
+      "step": 72000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.827442620554349e-06,
+      "loss": 0.9982,
+      "step": 72050
+    },
+    {
+      "epoch": 0.87,
+      "eval_runtime": 13.1146,
+      "eval_samples_per_second": 381.253,
+      "eval_steps_per_second": 0.763,
+      "step": 72050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.832874572407615e-06,
+      "loss": 0.9954,
+      "step": 72100
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.8259,
+      "eval_samples_per_second": 389.837,
+      "eval_steps_per_second": 0.78,
+      "step": 72100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83830652426088e-06,
+      "loss": 0.9957,
+      "step": 72150
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.9323,
+      "eval_samples_per_second": 386.627,
+      "eval_steps_per_second": 0.773,
+      "step": 72150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.843738476114145e-06,
+      "loss": 0.9977,
+      "step": 72200
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.7253,
+      "eval_samples_per_second": 392.917,
+      "eval_steps_per_second": 0.786,
+      "step": 72200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.849170427967409e-06,
+      "loss": 1.0011,
+      "step": 72250
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.7258,
+      "eval_samples_per_second": 392.903,
+      "eval_steps_per_second": 0.786,
+      "step": 72250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.854602379820673e-06,
+      "loss": 0.9985,
+      "step": 72300
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 13.1182,
+      "eval_samples_per_second": 381.15,
+      "eval_steps_per_second": 0.762,
+      "step": 72300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.860034331673937e-06,
+      "loss": 0.9891,
+      "step": 72350
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.8127,
+      "eval_samples_per_second": 390.237,
+      "eval_steps_per_second": 0.78,
+      "step": 72350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.865466283527203e-06,
+      "loss": 0.9947,
+      "step": 72400
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 13.1093,
+      "eval_samples_per_second": 381.408,
+      "eval_steps_per_second": 0.763,
+      "step": 72400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.870898235380469e-06,
+      "loss": 0.9985,
+      "step": 72450
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.7816,
+      "eval_samples_per_second": 391.188,
+      "eval_steps_per_second": 0.782,
+      "step": 72450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876330187233733e-06,
+      "loss": 0.994,
+      "step": 72500
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.9415,
+      "eval_samples_per_second": 386.354,
+      "eval_steps_per_second": 0.773,
+      "step": 72500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.881762139086997e-06,
+      "loss": 0.9918,
+      "step": 72550
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.9872,
+      "eval_samples_per_second": 384.994,
+      "eval_steps_per_second": 0.77,
+      "step": 72550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.88719409094026e-06,
+      "loss": 0.9933,
+      "step": 72600
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.8832,
+      "eval_samples_per_second": 388.101,
+      "eval_steps_per_second": 0.776,
+      "step": 72600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.892626042793527e-06,
+      "loss": 0.9902,
+      "step": 72650
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 13.0526,
+      "eval_samples_per_second": 383.066,
+      "eval_steps_per_second": 0.766,
+      "step": 72650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.89805799464679e-06,
+      "loss": 0.9915,
+      "step": 72700
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.9699,
+      "eval_samples_per_second": 385.507,
+      "eval_steps_per_second": 0.771,
+      "step": 72700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903489946500055e-06,
+      "loss": 0.9862,
+      "step": 72750
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 13.0084,
+      "eval_samples_per_second": 384.368,
+      "eval_steps_per_second": 0.769,
+      "step": 72750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.908921898353319e-06,
+      "loss": 0.9926,
+      "step": 72800
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.9559,
+      "eval_samples_per_second": 385.923,
+      "eval_steps_per_second": 0.772,
+      "step": 72800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.914353850206585e-06,
+      "loss": 0.9907,
+      "step": 72850
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 12.8173,
+      "eval_samples_per_second": 390.098,
+      "eval_steps_per_second": 0.78,
+      "step": 72850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.919785802059849e-06,
+      "loss": 0.9931,
+      "step": 72900
+    },
+    {
+      "epoch": 0.88,
+      "eval_runtime": 13.0037,
+      "eval_samples_per_second": 384.506,
+      "eval_steps_per_second": 0.769,
+      "step": 72900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.925217753913115e-06,
+      "loss": 0.9947,
+      "step": 72950
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 13.1478,
+      "eval_samples_per_second": 380.29,
+      "eval_steps_per_second": 0.761,
+      "step": 72950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.930649705766379e-06,
+      "loss": 0.9946,
+      "step": 73000
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.7879,
+      "eval_samples_per_second": 390.995,
+      "eval_steps_per_second": 0.782,
+      "step": 73000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.936081657619643e-06,
+      "loss": 0.9919,
+      "step": 73050
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.8685,
+      "eval_samples_per_second": 388.545,
+      "eval_steps_per_second": 0.777,
+      "step": 73050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.941513609472907e-06,
+      "loss": 0.9908,
+      "step": 73100
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.8789,
+      "eval_samples_per_second": 388.23,
+      "eval_steps_per_second": 0.776,
+      "step": 73100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.946945561326173e-06,
+      "loss": 0.987,
+      "step": 73150
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.8275,
+      "eval_samples_per_second": 389.787,
+      "eval_steps_per_second": 0.78,
+      "step": 73150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.952377513179438e-06,
+      "loss": 0.9887,
+      "step": 73200
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.8387,
+      "eval_samples_per_second": 389.447,
+      "eval_steps_per_second": 0.779,
+      "step": 73200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.957809465032702e-06,
+      "loss": 0.9878,
+      "step": 73250
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 13.07,
+      "eval_samples_per_second": 382.557,
+      "eval_steps_per_second": 0.765,
+      "step": 73250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.963241416885967e-06,
+      "loss": 0.9875,
+      "step": 73300
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.9762,
+      "eval_samples_per_second": 385.32,
+      "eval_steps_per_second": 0.771,
+      "step": 73300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.96867336873923e-06,
+      "loss": 0.9907,
+      "step": 73350
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.9803,
+      "eval_samples_per_second": 385.2,
+      "eval_steps_per_second": 0.77,
+      "step": 73350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.974105320592495e-06,
+      "loss": 0.989,
+      "step": 73400
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.8108,
+      "eval_samples_per_second": 390.295,
+      "eval_steps_per_second": 0.781,
+      "step": 73400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.97953727244576e-06,
+      "loss": 0.9891,
+      "step": 73450
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.811,
+      "eval_samples_per_second": 390.291,
+      "eval_steps_per_second": 0.781,
+      "step": 73450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.984969224299025e-06,
+      "loss": 0.991,
+      "step": 73500
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 12.7855,
+      "eval_samples_per_second": 391.069,
+      "eval_steps_per_second": 0.782,
+      "step": 73500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.99040117615229e-06,
+      "loss": 0.9857,
+      "step": 73550
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 13.0931,
+      "eval_samples_per_second": 381.88,
+      "eval_steps_per_second": 0.764,
+      "step": 73550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.995833128005554e-06,
+      "loss": 0.9924,
+      "step": 73600
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 13.0999,
+      "eval_samples_per_second": 381.683,
+      "eval_steps_per_second": 0.763,
+      "step": 73600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.001265079858819e-06,
+      "loss": 0.9828,
+      "step": 73650
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 13.1331,
+      "eval_samples_per_second": 380.718,
+      "eval_steps_per_second": 0.761,
+      "step": 73650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.006697031712083e-06,
+      "loss": 0.9882,
+      "step": 73700
+    },
+    {
+      "epoch": 0.89,
+      "eval_runtime": 13.1611,
+      "eval_samples_per_second": 379.906,
+      "eval_steps_per_second": 0.76,
+      "step": 73700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.012128983565348e-06,
+      "loss": 0.9881,
+      "step": 73750
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0781,
+      "eval_samples_per_second": 382.318,
+      "eval_steps_per_second": 0.765,
+      "step": 73750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.017560935418613e-06,
+      "loss": 0.9834,
+      "step": 73800
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.2698,
+      "eval_samples_per_second": 376.796,
+      "eval_steps_per_second": 0.754,
+      "step": 73800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.022992887271877e-06,
+      "loss": 0.9857,
+      "step": 73850
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0167,
+      "eval_samples_per_second": 384.121,
+      "eval_steps_per_second": 0.768,
+      "step": 73850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.028424839125142e-06,
+      "loss": 0.9879,
+      "step": 73900
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 12.8826,
+      "eval_samples_per_second": 388.121,
+      "eval_steps_per_second": 0.776,
+      "step": 73900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.033856790978407e-06,
+      "loss": 0.9842,
+      "step": 73950
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.2419,
+      "eval_samples_per_second": 377.59,
+      "eval_steps_per_second": 0.755,
+      "step": 73950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.039288742831672e-06,
+      "loss": 0.9875,
+      "step": 74000
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.1804,
+      "eval_samples_per_second": 379.351,
+      "eval_steps_per_second": 0.759,
+      "step": 74000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.044720694684936e-06,
+      "loss": 0.9926,
+      "step": 74050
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0913,
+      "eval_samples_per_second": 381.933,
+      "eval_steps_per_second": 0.764,
+      "step": 74050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.0501526465382e-06,
+      "loss": 0.9866,
+      "step": 74100
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 12.9684,
+      "eval_samples_per_second": 385.553,
+      "eval_steps_per_second": 0.771,
+      "step": 74100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.055584598391465e-06,
+      "loss": 0.9802,
+      "step": 74150
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0712,
+      "eval_samples_per_second": 382.519,
+      "eval_steps_per_second": 0.765,
+      "step": 74150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.061016550244729e-06,
+      "loss": 0.9879,
+      "step": 74200
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.1549,
+      "eval_samples_per_second": 380.087,
+      "eval_steps_per_second": 0.76,
+      "step": 74200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.066448502097994e-06,
+      "loss": 0.982,
+      "step": 74250
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0504,
+      "eval_samples_per_second": 383.129,
+      "eval_steps_per_second": 0.766,
+      "step": 74250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.07188045395126e-06,
+      "loss": 0.9877,
+      "step": 74300
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0664,
+      "eval_samples_per_second": 382.66,
+      "eval_steps_per_second": 0.765,
+      "step": 74300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.077312405804524e-06,
+      "loss": 0.9833,
+      "step": 74350
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.1774,
+      "eval_samples_per_second": 379.437,
+      "eval_steps_per_second": 0.759,
+      "step": 74350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.082744357657788e-06,
+      "loss": 0.9819,
+      "step": 74400
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.1893,
+      "eval_samples_per_second": 379.095,
+      "eval_steps_per_second": 0.758,
+      "step": 74400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.088176309511053e-06,
+      "loss": 0.9838,
+      "step": 74450
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 12.8618,
+      "eval_samples_per_second": 388.747,
+      "eval_steps_per_second": 0.777,
+      "step": 74450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.093608261364317e-06,
+      "loss": 0.987,
+      "step": 74500
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.06,
+      "eval_samples_per_second": 382.85,
+      "eval_steps_per_second": 0.766,
+      "step": 74500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.099040213217582e-06,
+      "loss": 0.9834,
+      "step": 74550
+    },
+    {
+      "epoch": 0.9,
+      "eval_runtime": 13.0988,
+      "eval_samples_per_second": 381.713,
+      "eval_steps_per_second": 0.763,
+      "step": 74550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.104472165070846e-06,
+      "loss": 0.9822,
+      "step": 74600
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 12.8106,
+      "eval_samples_per_second": 390.301,
+      "eval_steps_per_second": 0.781,
+      "step": 74600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.109904116924112e-06,
+      "loss": 0.979,
+      "step": 74650
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0379,
+      "eval_samples_per_second": 383.499,
+      "eval_steps_per_second": 0.767,
+      "step": 74650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.115336068777376e-06,
+      "loss": 0.9789,
+      "step": 74700
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0918,
+      "eval_samples_per_second": 381.918,
+      "eval_steps_per_second": 0.764,
+      "step": 74700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.12076802063064e-06,
+      "loss": 0.9823,
+      "step": 74750
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0765,
+      "eval_samples_per_second": 382.364,
+      "eval_steps_per_second": 0.765,
+      "step": 74750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.126199972483906e-06,
+      "loss": 0.9862,
+      "step": 74800
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.1164,
+      "eval_samples_per_second": 381.203,
+      "eval_steps_per_second": 0.762,
+      "step": 74800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.13163192433717e-06,
+      "loss": 0.9849,
+      "step": 74850
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.1038,
+      "eval_samples_per_second": 381.568,
+      "eval_steps_per_second": 0.763,
+      "step": 74850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.137063876190434e-06,
+      "loss": 0.9837,
+      "step": 74900
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.1991,
+      "eval_samples_per_second": 378.813,
+      "eval_steps_per_second": 0.758,
+      "step": 74900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.142495828043699e-06,
+      "loss": 0.9839,
+      "step": 74950
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0104,
+      "eval_samples_per_second": 384.308,
+      "eval_steps_per_second": 0.769,
+      "step": 74950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.147927779896964e-06,
+      "loss": 0.9819,
+      "step": 75000
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0544,
+      "eval_samples_per_second": 383.012,
+      "eval_steps_per_second": 0.766,
+      "step": 75000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.153359731750228e-06,
+      "loss": 0.9812,
+      "step": 75050
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.056,
+      "eval_samples_per_second": 382.967,
+      "eval_steps_per_second": 0.766,
+      "step": 75050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.158791683603494e-06,
+      "loss": 0.9784,
+      "step": 75100
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 12.7397,
+      "eval_samples_per_second": 392.475,
+      "eval_steps_per_second": 0.785,
+      "step": 75100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.164223635456758e-06,
+      "loss": 0.9855,
+      "step": 75150
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0291,
+      "eval_samples_per_second": 383.756,
+      "eval_steps_per_second": 0.768,
+      "step": 75150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.169655587310022e-06,
+      "loss": 0.9805,
+      "step": 75200
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 13.0337,
+      "eval_samples_per_second": 383.622,
+      "eval_steps_per_second": 0.767,
+      "step": 75200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.175087539163286e-06,
+      "loss": 0.9817,
+      "step": 75250
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 12.7391,
+      "eval_samples_per_second": 392.492,
+      "eval_steps_per_second": 0.785,
+      "step": 75250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.180519491016552e-06,
+      "loss": 0.9826,
+      "step": 75300
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 12.9479,
+      "eval_samples_per_second": 386.163,
+      "eval_steps_per_second": 0.772,
+      "step": 75300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.185951442869818e-06,
+      "loss": 0.9912,
+      "step": 75350
+    },
+    {
+      "epoch": 0.91,
+      "eval_runtime": 12.8262,
+      "eval_samples_per_second": 389.828,
+      "eval_steps_per_second": 0.78,
+      "step": 75350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.191383394723082e-06,
+      "loss": 0.9854,
+      "step": 75400
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9222,
+      "eval_samples_per_second": 386.93,
+      "eval_steps_per_second": 0.774,
+      "step": 75400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.196815346576346e-06,
+      "loss": 0.98,
+      "step": 75450
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9815,
+      "eval_samples_per_second": 385.163,
+      "eval_steps_per_second": 0.77,
+      "step": 75450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.20224729842961e-06,
+      "loss": 0.977,
+      "step": 75500
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9495,
+      "eval_samples_per_second": 386.116,
+      "eval_steps_per_second": 0.772,
+      "step": 75500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.207679250282874e-06,
+      "loss": 0.9758,
+      "step": 75550
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 13.1355,
+      "eval_samples_per_second": 380.647,
+      "eval_steps_per_second": 0.761,
+      "step": 75550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.21311120213614e-06,
+      "loss": 0.978,
+      "step": 75600
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 13.0153,
+      "eval_samples_per_second": 384.164,
+      "eval_steps_per_second": 0.768,
+      "step": 75600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.218543153989404e-06,
+      "loss": 0.9811,
+      "step": 75650
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9749,
+      "eval_samples_per_second": 385.359,
+      "eval_steps_per_second": 0.771,
+      "step": 75650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.22397510584267e-06,
+      "loss": 0.978,
+      "step": 75700
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9237,
+      "eval_samples_per_second": 386.886,
+      "eval_steps_per_second": 0.774,
+      "step": 75700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.229407057695934e-06,
+      "loss": 0.9779,
+      "step": 75750
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9365,
+      "eval_samples_per_second": 386.503,
+      "eval_steps_per_second": 0.773,
+      "step": 75750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.234839009549198e-06,
+      "loss": 0.9775,
+      "step": 75800
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.7578,
+      "eval_samples_per_second": 391.916,
+      "eval_steps_per_second": 0.784,
+      "step": 75800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.240270961402462e-06,
+      "loss": 0.9775,
+      "step": 75850
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.7458,
+      "eval_samples_per_second": 392.287,
+      "eval_steps_per_second": 0.785,
+      "step": 75850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.245702913255728e-06,
+      "loss": 0.9777,
+      "step": 75900
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 13.1264,
+      "eval_samples_per_second": 380.911,
+      "eval_steps_per_second": 0.762,
+      "step": 75900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.251134865108992e-06,
+      "loss": 0.9753,
+      "step": 75950
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9487,
+      "eval_samples_per_second": 386.141,
+      "eval_steps_per_second": 0.772,
+      "step": 75950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.256566816962256e-06,
+      "loss": 0.981,
+      "step": 76000
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.8876,
+      "eval_samples_per_second": 387.969,
+      "eval_steps_per_second": 0.776,
+      "step": 76000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.261998768815522e-06,
+      "loss": 0.98,
+      "step": 76050
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9615,
+      "eval_samples_per_second": 385.759,
+      "eval_steps_per_second": 0.772,
+      "step": 76050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.267430720668786e-06,
+      "loss": 0.9765,
+      "step": 76100
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 13.0379,
+      "eval_samples_per_second": 383.498,
+      "eval_steps_per_second": 0.767,
+      "step": 76100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.272862672522052e-06,
+      "loss": 0.9798,
+      "step": 76150
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9253,
+      "eval_samples_per_second": 386.837,
+      "eval_steps_per_second": 0.774,
+      "step": 76150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.278294624375316e-06,
+      "loss": 0.9744,
+      "step": 76200
+    },
+    {
+      "epoch": 0.92,
+      "eval_runtime": 12.9812,
+      "eval_samples_per_second": 385.174,
+      "eval_steps_per_second": 0.77,
+      "step": 76200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.28372657622858e-06,
+      "loss": 0.9763,
+      "step": 76250
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 12.6997,
+      "eval_samples_per_second": 393.71,
+      "eval_steps_per_second": 0.787,
+      "step": 76250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.289158528081844e-06,
+      "loss": 0.9773,
+      "step": 76300
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 12.9925,
+      "eval_samples_per_second": 384.838,
+      "eval_steps_per_second": 0.77,
+      "step": 76300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.294590479935108e-06,
+      "loss": 0.9809,
+      "step": 76350
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.0709,
+      "eval_samples_per_second": 382.528,
+      "eval_steps_per_second": 0.765,
+      "step": 76350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.300022431788374e-06,
+      "loss": 0.9733,
+      "step": 76400
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.3701,
+      "eval_samples_per_second": 373.968,
+      "eval_steps_per_second": 0.748,
+      "step": 76400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.30545438364164e-06,
+      "loss": 0.9809,
+      "step": 76450
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.1659,
+      "eval_samples_per_second": 379.768,
+      "eval_steps_per_second": 0.76,
+      "step": 76450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.310886335494904e-06,
+      "loss": 0.9688,
+      "step": 76500
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.0618,
+      "eval_samples_per_second": 382.797,
+      "eval_steps_per_second": 0.766,
+      "step": 76500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.316318287348168e-06,
+      "loss": 0.9706,
+      "step": 76550
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.0337,
+      "eval_samples_per_second": 383.621,
+      "eval_steps_per_second": 0.767,
+      "step": 76550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.321750239201432e-06,
+      "loss": 0.9743,
+      "step": 76600
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.3181,
+      "eval_samples_per_second": 375.43,
+      "eval_steps_per_second": 0.751,
+      "step": 76600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.327182191054696e-06,
+      "loss": 0.9695,
+      "step": 76650
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 12.8261,
+      "eval_samples_per_second": 389.831,
+      "eval_steps_per_second": 0.78,
+      "step": 76650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.332614142907962e-06,
+      "loss": 0.9747,
+      "step": 76700
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.1759,
+      "eval_samples_per_second": 379.48,
+      "eval_steps_per_second": 0.759,
+      "step": 76700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.338046094761226e-06,
+      "loss": 0.9783,
+      "step": 76750
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.0865,
+      "eval_samples_per_second": 382.073,
+      "eval_steps_per_second": 0.764,
+      "step": 76750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.343478046614492e-06,
+      "loss": 0.9702,
+      "step": 76800
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.2533,
+      "eval_samples_per_second": 377.266,
+      "eval_steps_per_second": 0.755,
+      "step": 76800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.348909998467756e-06,
+      "loss": 0.9772,
+      "step": 76850
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.1961,
+      "eval_samples_per_second": 378.9,
+      "eval_steps_per_second": 0.758,
+      "step": 76850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.35434195032102e-06,
+      "loss": 0.971,
+      "step": 76900
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 13.1072,
+      "eval_samples_per_second": 381.471,
+      "eval_steps_per_second": 0.763,
+      "step": 76900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.359773902174286e-06,
+      "loss": 0.9732,
+      "step": 76950
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 12.9801,
+      "eval_samples_per_second": 385.204,
+      "eval_steps_per_second": 0.77,
+      "step": 76950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.36520585402755e-06,
+      "loss": 0.9699,
+      "step": 77000
+    },
+    {
+      "epoch": 0.93,
+      "eval_runtime": 12.9757,
+      "eval_samples_per_second": 385.335,
+      "eval_steps_per_second": 0.771,
+      "step": 77000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.370637805880814e-06,
+      "loss": 0.9699,
+      "step": 77050
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 12.9802,
+      "eval_samples_per_second": 385.202,
+      "eval_steps_per_second": 0.77,
+      "step": 77050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.376069757734078e-06,
+      "loss": 0.9721,
+      "step": 77100
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.2026,
+      "eval_samples_per_second": 378.712,
+      "eval_steps_per_second": 0.757,
+      "step": 77100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.381501709587344e-06,
+      "loss": 0.9751,
+      "step": 77150
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 12.9037,
+      "eval_samples_per_second": 387.486,
+      "eval_steps_per_second": 0.775,
+      "step": 77150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.386933661440608e-06,
+      "loss": 0.9738,
+      "step": 77200
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 12.8099,
+      "eval_samples_per_second": 390.323,
+      "eval_steps_per_second": 0.781,
+      "step": 77200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.392365613293874e-06,
+      "loss": 0.9794,
+      "step": 77250
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.0338,
+      "eval_samples_per_second": 383.618,
+      "eval_steps_per_second": 0.767,
+      "step": 77250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.397797565147138e-06,
+      "loss": 0.9727,
+      "step": 77300
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.1464,
+      "eval_samples_per_second": 380.331,
+      "eval_steps_per_second": 0.761,
+      "step": 77300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.403229517000402e-06,
+      "loss": 0.9669,
+      "step": 77350
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.0298,
+      "eval_samples_per_second": 383.736,
+      "eval_steps_per_second": 0.767,
+      "step": 77350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.408661468853666e-06,
+      "loss": 0.9705,
+      "step": 77400
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.2222,
+      "eval_samples_per_second": 378.153,
+      "eval_steps_per_second": 0.756,
+      "step": 77400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.414093420706932e-06,
+      "loss": 0.9762,
+      "step": 77450
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.1173,
+      "eval_samples_per_second": 381.175,
+      "eval_steps_per_second": 0.762,
+      "step": 77450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.419525372560198e-06,
+      "loss": 0.9679,
+      "step": 77500
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.0948,
+      "eval_samples_per_second": 381.831,
+      "eval_steps_per_second": 0.764,
+      "step": 77500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.424957324413462e-06,
+      "loss": 0.9676,
+      "step": 77550
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.0946,
+      "eval_samples_per_second": 381.837,
+      "eval_steps_per_second": 0.764,
+      "step": 77550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.430389276266726e-06,
+      "loss": 0.9668,
+      "step": 77600
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 12.8307,
+      "eval_samples_per_second": 389.69,
+      "eval_steps_per_second": 0.779,
+      "step": 77600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.43582122811999e-06,
+      "loss": 0.9695,
+      "step": 77650
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.1928,
+      "eval_samples_per_second": 378.994,
+      "eval_steps_per_second": 0.758,
+      "step": 77650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.441253179973254e-06,
+      "loss": 0.9702,
+      "step": 77700
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.0525,
+      "eval_samples_per_second": 383.068,
+      "eval_steps_per_second": 0.766,
+      "step": 77700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.44668513182652e-06,
+      "loss": 0.9704,
+      "step": 77750
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.1072,
+      "eval_samples_per_second": 381.47,
+      "eval_steps_per_second": 0.763,
+      "step": 77750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.452117083679784e-06,
+      "loss": 0.9687,
+      "step": 77800
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 12.8753,
+      "eval_samples_per_second": 388.341,
+      "eval_steps_per_second": 0.777,
+      "step": 77800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.45754903553305e-06,
+      "loss": 0.9736,
+      "step": 77850
+    },
+    {
+      "epoch": 0.94,
+      "eval_runtime": 13.1417,
+      "eval_samples_per_second": 380.468,
+      "eval_steps_per_second": 0.761,
+      "step": 77850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.462980987386314e-06,
+      "loss": 0.9694,
+      "step": 77900
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 12.9709,
+      "eval_samples_per_second": 385.477,
+      "eval_steps_per_second": 0.771,
+      "step": 77900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.468412939239578e-06,
+      "loss": 0.9708,
+      "step": 77950
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.1795,
+      "eval_samples_per_second": 379.377,
+      "eval_steps_per_second": 0.759,
+      "step": 77950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.473844891092842e-06,
+      "loss": 0.9631,
+      "step": 78000
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.1333,
+      "eval_samples_per_second": 380.712,
+      "eval_steps_per_second": 0.761,
+      "step": 78000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.479276842946108e-06,
+      "loss": 0.9717,
+      "step": 78050
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.0745,
+      "eval_samples_per_second": 382.424,
+      "eval_steps_per_second": 0.765,
+      "step": 78050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.484708794799372e-06,
+      "loss": 0.9722,
+      "step": 78100
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.1747,
+      "eval_samples_per_second": 379.514,
+      "eval_steps_per_second": 0.759,
+      "step": 78100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.490140746652636e-06,
+      "loss": 0.9708,
+      "step": 78150
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.1743,
+      "eval_samples_per_second": 379.526,
+      "eval_steps_per_second": 0.759,
+      "step": 78150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.4955726985059e-06,
+      "loss": 0.9701,
+      "step": 78200
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 12.8873,
+      "eval_samples_per_second": 387.978,
+      "eval_steps_per_second": 0.776,
+      "step": 78200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.501004650359166e-06,
+      "loss": 0.9683,
+      "step": 78250
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 12.8431,
+      "eval_samples_per_second": 389.314,
+      "eval_steps_per_second": 0.779,
+      "step": 78250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.506436602212432e-06,
+      "loss": 0.9695,
+      "step": 78300
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.075,
+      "eval_samples_per_second": 382.41,
+      "eval_steps_per_second": 0.765,
+      "step": 78300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.511868554065696e-06,
+      "loss": 0.9699,
+      "step": 78350
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.011,
+      "eval_samples_per_second": 384.289,
+      "eval_steps_per_second": 0.769,
+      "step": 78350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.51730050591896e-06,
+      "loss": 0.9666,
+      "step": 78400
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.0643,
+      "eval_samples_per_second": 382.722,
+      "eval_steps_per_second": 0.765,
+      "step": 78400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.522732457772224e-06,
+      "loss": 0.9648,
+      "step": 78450
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 12.9387,
+      "eval_samples_per_second": 386.438,
+      "eval_steps_per_second": 0.773,
+      "step": 78450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.528164409625488e-06,
+      "loss": 0.9675,
+      "step": 78500
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.0937,
+      "eval_samples_per_second": 381.862,
+      "eval_steps_per_second": 0.764,
+      "step": 78500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.533596361478754e-06,
+      "loss": 0.9728,
+      "step": 78550
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 12.7864,
+      "eval_samples_per_second": 391.04,
+      "eval_steps_per_second": 0.782,
+      "step": 78550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.53902831333202e-06,
+      "loss": 0.9659,
+      "step": 78600
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 13.0284,
+      "eval_samples_per_second": 383.776,
+      "eval_steps_per_second": 0.768,
+      "step": 78600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.544460265185284e-06,
+      "loss": 0.9708,
+      "step": 78650
+    },
+    {
+      "epoch": 0.95,
+      "eval_runtime": 12.7254,
+      "eval_samples_per_second": 392.914,
+      "eval_steps_per_second": 0.786,
+      "step": 78650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.549892217038548e-06,
+      "loss": 0.9684,
+      "step": 78700
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0692,
+      "eval_samples_per_second": 382.578,
+      "eval_steps_per_second": 0.765,
+      "step": 78700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.555324168891812e-06,
+      "loss": 0.9669,
+      "step": 78750
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0726,
+      "eval_samples_per_second": 382.479,
+      "eval_steps_per_second": 0.765,
+      "step": 78750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.560756120745076e-06,
+      "loss": 0.9658,
+      "step": 78800
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0994,
+      "eval_samples_per_second": 381.698,
+      "eval_steps_per_second": 0.763,
+      "step": 78800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.566188072598342e-06,
+      "loss": 0.9626,
+      "step": 78850
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 12.9891,
+      "eval_samples_per_second": 384.937,
+      "eval_steps_per_second": 0.77,
+      "step": 78850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.571620024451606e-06,
+      "loss": 0.9627,
+      "step": 78900
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0855,
+      "eval_samples_per_second": 382.102,
+      "eval_steps_per_second": 0.764,
+      "step": 78900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.577051976304872e-06,
+      "loss": 0.9669,
+      "step": 78950
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 12.9958,
+      "eval_samples_per_second": 384.739,
+      "eval_steps_per_second": 0.769,
+      "step": 78950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.582483928158136e-06,
+      "loss": 0.961,
+      "step": 79000
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 12.9372,
+      "eval_samples_per_second": 386.484,
+      "eval_steps_per_second": 0.773,
+      "step": 79000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.5879158800114e-06,
+      "loss": 0.9622,
+      "step": 79050
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 12.9409,
+      "eval_samples_per_second": 386.373,
+      "eval_steps_per_second": 0.773,
+      "step": 79050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.593347831864666e-06,
+      "loss": 0.9643,
+      "step": 79100
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.106,
+      "eval_samples_per_second": 381.505,
+      "eval_steps_per_second": 0.763,
+      "step": 79100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.59877978371793e-06,
+      "loss": 0.961,
+      "step": 79150
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 12.9598,
+      "eval_samples_per_second": 385.809,
+      "eval_steps_per_second": 0.772,
+      "step": 79150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.604211735571194e-06,
+      "loss": 0.9571,
+      "step": 79200
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0206,
+      "eval_samples_per_second": 384.006,
+      "eval_steps_per_second": 0.768,
+      "step": 79200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.609643687424458e-06,
+      "loss": 0.9676,
+      "step": 79250
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.1282,
+      "eval_samples_per_second": 380.86,
+      "eval_steps_per_second": 0.762,
+      "step": 79250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.615075639277724e-06,
+      "loss": 0.9586,
+      "step": 79300
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0247,
+      "eval_samples_per_second": 383.887,
+      "eval_steps_per_second": 0.768,
+      "step": 79300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.620507591130988e-06,
+      "loss": 0.9654,
+      "step": 79350
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.1777,
+      "eval_samples_per_second": 379.43,
+      "eval_steps_per_second": 0.759,
+      "step": 79350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.625939542984253e-06,
+      "loss": 0.9595,
+      "step": 79400
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.0825,
+      "eval_samples_per_second": 382.19,
+      "eval_steps_per_second": 0.764,
+      "step": 79400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.631371494837518e-06,
+      "loss": 0.964,
+      "step": 79450
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 13.2454,
+      "eval_samples_per_second": 377.488,
+      "eval_steps_per_second": 0.755,
+      "step": 79450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.636803446690782e-06,
+      "loss": 0.9642,
+      "step": 79500
+    },
+    {
+      "epoch": 0.96,
+      "eval_runtime": 12.8727,
+      "eval_samples_per_second": 388.418,
+      "eval_steps_per_second": 0.777,
+      "step": 79500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.642235398544046e-06,
+      "loss": 0.9571,
+      "step": 79550
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.1572,
+      "eval_samples_per_second": 380.021,
+      "eval_steps_per_second": 0.76,
+      "step": 79550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.647667350397312e-06,
+      "loss": 0.9638,
+      "step": 79600
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.2593,
+      "eval_samples_per_second": 377.094,
+      "eval_steps_per_second": 0.754,
+      "step": 79600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.653099302250577e-06,
+      "loss": 0.9626,
+      "step": 79650
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.2665,
+      "eval_samples_per_second": 376.89,
+      "eval_steps_per_second": 0.754,
+      "step": 79650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.658531254103841e-06,
+      "loss": 0.96,
+      "step": 79700
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.1685,
+      "eval_samples_per_second": 379.694,
+      "eval_steps_per_second": 0.759,
+      "step": 79700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.663963205957106e-06,
+      "loss": 0.9609,
+      "step": 79750
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.1336,
+      "eval_samples_per_second": 380.704,
+      "eval_steps_per_second": 0.761,
+      "step": 79750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.66939515781037e-06,
+      "loss": 0.9633,
+      "step": 79800
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.0975,
+      "eval_samples_per_second": 381.752,
+      "eval_steps_per_second": 0.764,
+      "step": 79800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.674827109663634e-06,
+      "loss": 0.9626,
+      "step": 79850
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.1409,
+      "eval_samples_per_second": 380.492,
+      "eval_steps_per_second": 0.761,
+      "step": 79850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.6802590615169e-06,
+      "loss": 0.9536,
+      "step": 79900
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 12.8462,
+      "eval_samples_per_second": 389.221,
+      "eval_steps_per_second": 0.778,
+      "step": 79900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.685691013370164e-06,
+      "loss": 0.9545,
+      "step": 79950
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 13.339,
+      "eval_samples_per_second": 374.839,
+      "eval_steps_per_second": 0.75,
+      "step": 79950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.69112296522343e-06,
+      "loss": 0.9662,
+      "step": 80000
+    },
+    {
+      "epoch": 0.97,
+      "eval_runtime": 12.8538,
+      "eval_samples_per_second": 388.99,
+      "eval_steps_per_second": 0.778,
+      "step": 80000
+    }
+  ],
+  "max_steps": 7670662,
+  "num_train_epochs": 94,
+  "total_flos": 2.73294686158848e+18,
+  "trial_name": null,
+  "trial_params": null
+}