diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5524 @@
+{
+  "best_metric": 0.7965201735496521,
+  "best_model_checkpoint": "lora-Vicuna/checkpoint-1000",
+  "epoch": 2.9878273699741795,
+  "global_step": 16200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 2.041,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 1.5971,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.2446,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 1.1737,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0003,
+      "loss": 1.1234,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002996287817855596,
+      "loss": 1.0991,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002992575635711192,
+      "loss": 1.089,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002988863453566788,
+      "loss": 1.0685,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002985151271422384,
+      "loss": 1.0466,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029814390892779805,
+      "loss": 1.0441,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.8649595379829407,
+      "eval_runtime": 1.3492,
+      "eval_samples_per_second": 0.741,
+      "eval_steps_per_second": 0.741,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002977726907133576,
+      "loss": 1.0417,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00029740147249891725,
+      "loss": 1.0269,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002970302542844769,
+      "loss": 1.015,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002966590360700365,
+      "loss": 1.0089,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002962878178555961,
+      "loss": 1.0269,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002959165996411557,
+      "loss": 1.0187,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029554538142671533,
+      "loss": 1.0142,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002951741632122749,
+      "loss": 1.0077,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029480294499783453,
+      "loss": 1.004,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029443172678339416,
+      "loss": 1.0042,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.8426274061203003,
+      "eval_runtime": 1.3288,
+      "eval_samples_per_second": 0.753,
+      "eval_steps_per_second": 0.753,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029406050856895373,
+      "loss": 1.0061,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029368929035451335,
+      "loss": 1.0152,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000293318072140073,
+      "loss": 1.0049,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002929468539256326,
+      "loss": 1.0011,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029257563571119223,
+      "loss": 0.9932,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002922044174967518,
+      "loss": 0.9926,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029183319928231143,
+      "loss": 0.9971,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000291461981067871,
+      "loss": 0.9866,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029109076285343063,
+      "loss": 0.9871,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029071954463899026,
+      "loss": 0.9906,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8352808356285095,
+      "eval_runtime": 1.3169,
+      "eval_samples_per_second": 0.759,
+      "eval_steps_per_second": 0.759,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002903483264245499,
+      "loss": 0.9797,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002899771082101095,
+      "loss": 0.989,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002896058899956691,
+      "loss": 0.9931,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002892346717812287,
+      "loss": 0.9859,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00028886345356678834,
+      "loss": 0.9874,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002884922353523479,
+      "loss": 0.9888,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028812101713790754,
+      "loss": 0.9798,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028774979892346716,
+      "loss": 0.981,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00028737858070902674,
+      "loss": 0.9824,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00028700736249458636,
+      "loss": 0.9922,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.8162809610366821,
+      "eval_runtime": 1.2603,
+      "eval_samples_per_second": 0.793,
+      "eval_steps_per_second": 0.793,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000286636144280146,
+      "loss": 0.9942,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002862649260657056,
+      "loss": 0.9878,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00028589370785126524,
+      "loss": 0.9855,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002855224896368248,
+      "loss": 0.9803,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028515127142238444,
+      "loss": 0.9793,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000284780053207944,
+      "loss": 0.9939,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00028440883499350364,
+      "loss": 0.9806,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00028403761677906327,
+      "loss": 0.9864,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002836663985646229,
+      "loss": 0.9839,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002832951803501825,
+      "loss": 0.9816,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.7965201735496521,
+      "eval_runtime": 1.3017,
+      "eval_samples_per_second": 0.768,
+      "eval_steps_per_second": 0.768,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002829239621357421,
+      "loss": 0.9848,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002825527439213017,
+      "loss": 0.9908,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00028218152570686134,
+      "loss": 0.9885,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002818103074924209,
+      "loss": 0.9772,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00028143908927798054,
+      "loss": 0.9948,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00028106787106354017,
+      "loss": 0.9778,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00028069665284909974,
+      "loss": 0.9909,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00028032543463465937,
+      "loss": 0.9802,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000279954216420219,
+      "loss": 0.9849,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002795829982057786,
+      "loss": 0.9788,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.8706400990486145,
+      "eval_runtime": 1.3141,
+      "eval_samples_per_second": 0.761,
+      "eval_steps_per_second": 0.761,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00027921177999133825,
+      "loss": 0.9766,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002788405617768978,
+      "loss": 0.9853,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00027846934356245745,
+      "loss": 0.9814,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000278098125348017,
+      "loss": 0.9747,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00027772690713357665,
+      "loss": 0.9715,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002773556889191363,
+      "loss": 0.9903,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002769844707046959,
+      "loss": 0.9862,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002766132524902555,
+      "loss": 0.9791,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002762420342758151,
+      "loss": 0.9834,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002758708160613747,
+      "loss": 0.9752,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.8170408010482788,
+      "eval_runtime": 1.2722,
+      "eval_samples_per_second": 0.786,
+      "eval_steps_per_second": 0.786,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00027549959784693435,
+      "loss": 0.9803,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002751283796324939,
+      "loss": 0.9968,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00027475716141805355,
+      "loss": 0.98,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002743859432036132,
+      "loss": 0.9819,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00027401472498917275,
+      "loss": 0.9855,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002736435067747324,
+      "loss": 0.9852,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000273272288560292,
+      "loss": 0.977,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00027290107034585163,
+      "loss": 0.9918,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00027252985213141126,
+      "loss": 0.9965,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00027215863391697083,
+      "loss": 0.9721,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.8074629902839661,
+      "eval_runtime": 1.2561,
+      "eval_samples_per_second": 0.796,
+      "eval_steps_per_second": 0.796,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00027178741570253045,
+      "loss": 0.984,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002714533193095341,
+      "loss": 0.9858,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002711006620058157,
+      "loss": 1.011,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00027072944379137533,
+      "loss": 0.9876,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00027035822557693495,
+      "loss": 0.9941,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002699870073624946,
+      "loss": 0.9935,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00026961578914805415,
+      "loss": 0.9818,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002692445709336138,
+      "loss": 0.9876,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002688733527191734,
+      "loss": 0.9628,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026850213450473303,
+      "loss": 0.983,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.8400484919548035,
+      "eval_runtime": 1.261,
+      "eval_samples_per_second": 0.793,
+      "eval_steps_per_second": 0.793,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002681309162902926,
+      "loss": 0.9845,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00026775969807585223,
+      "loss": 0.98,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002673884798614118,
+      "loss": 0.9649,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00026701726164697143,
+      "loss": 0.9847,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00026664604343253106,
+      "loss": 0.9824,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002662748252180907,
+      "loss": 0.9777,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002659036070036503,
+      "loss": 0.9781,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00026553238878920994,
+      "loss": 0.9795,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002651611705747695,
+      "loss": 0.9875,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00026478995236032914,
+      "loss": 0.9741,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.8619368672370911,
+      "eval_runtime": 1.2019,
+      "eval_samples_per_second": 0.832,
+      "eval_steps_per_second": 0.832,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002644187341458887,
+      "loss": 0.9841,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026404751593144834,
+      "loss": 0.9813,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00026367629771700796,
+      "loss": 0.9777,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002633050795025676,
+      "loss": 0.972,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00026293386128812716,
+      "loss": 0.97,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002625626430736868,
+      "loss": 0.9866,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002621914248592464,
+      "loss": 0.9742,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026182020664480604,
+      "loss": 0.9843,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002614489884303656,
+      "loss": 0.9616,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026107777021592524,
+      "loss": 0.9813,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.8606493473052979,
+      "eval_runtime": 1.2145,
+      "eval_samples_per_second": 0.823,
+      "eval_steps_per_second": 0.823,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002607065520014848,
+      "loss": 0.9716,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026033533378704444,
+      "loss": 0.9822,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025996411557260407,
+      "loss": 0.9857,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002595928973581637,
+      "loss": 0.9977,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002592216791437233,
+      "loss": 0.9715,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025885046092928294,
+      "loss": 0.9623,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002584792427148425,
+      "loss": 0.9637,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025810802450040214,
+      "loss": 0.978,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002577368062859617,
+      "loss": 0.9802,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00025736558807152134,
+      "loss": 0.9807,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.8513050079345703,
+      "eval_runtime": 1.2201,
+      "eval_samples_per_second": 0.82,
+      "eval_steps_per_second": 0.82,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025699436985708097,
+      "loss": 0.9841,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002566231516426406,
+      "loss": 0.9714,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00025625193342820017,
+      "loss": 0.9693,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002558807152137598,
+      "loss": 0.9763,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002555094969993194,
+      "loss": 0.9779,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00025513827878487905,
+      "loss": 0.9736,
+      "step": 2520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002547670605704386,
+      "loss": 0.9788,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00025439584235599825,
+      "loss": 0.9789,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002540246241415578,
+      "loss": 0.9729,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025365340592711745,
+      "loss": 0.9752,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.8586015105247498,
+      "eval_runtime": 1.2961,
+      "eval_samples_per_second": 0.772,
+      "eval_steps_per_second": 0.772,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00025328218771267707,
+      "loss": 0.9734,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002529109694982367,
+      "loss": 0.9767,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002525397512837963,
+      "loss": 0.962,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00025216853306935595,
+      "loss": 0.9706,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002517973148549155,
+      "loss": 0.975,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00025142609664047515,
+      "loss": 0.9653,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002510548784260347,
+      "loss": 0.976,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00025068366021159435,
+      "loss": 0.9768,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000250312441997154,
+      "loss": 0.9764,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002499412237827136,
+      "loss": 0.9741,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.8531299829483032,
+      "eval_runtime": 1.2788,
+      "eval_samples_per_second": 0.782,
+      "eval_steps_per_second": 0.782,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002495700055682732,
+      "loss": 0.9786,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002491987873538328,
+      "loss": 0.9737,
+      "step": 2840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024882756913939243,
+      "loss": 0.9648,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024845635092495206,
+      "loss": 0.9717,
+      "step": 2880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00024808513271051163,
+      "loss": 0.9706,
+      "step": 2900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00024771391449607125,
+      "loss": 0.9662,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002473426962816308,
+      "loss": 0.9719,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00024697147806719045,
+      "loss": 0.9729,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002466002598527501,
+      "loss": 0.9679,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002462290416383097,
+      "loss": 0.9782,
+      "step": 3000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.8357999324798584,
+      "eval_runtime": 1.3862,
+      "eval_samples_per_second": 0.721,
+      "eval_steps_per_second": 0.721,
+      "step": 3000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024585782342386933,
+      "loss": 0.9636,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024548660520942896,
+      "loss": 0.9779,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00024511538699498853,
+      "loss": 0.9718,
+      "step": 3060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024474416878054816,
+      "loss": 0.9756,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00024437295056610773,
+      "loss": 0.983,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024400173235166736,
+      "loss": 0.9735,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00024363051413722698,
+      "loss": 0.976,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002432592959227866,
+      "loss": 0.9691,
+      "step": 3160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002428880777083462,
+      "loss": 0.9644,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002425168594939058,
+      "loss": 0.9768,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.8302448987960815,
+      "eval_runtime": 1.362,
+      "eval_samples_per_second": 0.734,
+      "eval_steps_per_second": 0.734,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002421456412794654,
+      "loss": 0.9709,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00024177442306502504,
+      "loss": 0.9696,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00024140320485058466,
+      "loss": 0.9669,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00024103198663614426,
+      "loss": 0.965,
+      "step": 3280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002406607684217039,
+      "loss": 0.9757,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00024028955020726346,
+      "loss": 0.9717,
+      "step": 3320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002399183319928231,
+      "loss": 0.963,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00023954711377838271,
+      "loss": 0.9519,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002391758955639423,
+      "loss": 0.964,
+      "step": 3380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00023880467734950194,
+      "loss": 0.967,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.87740159034729,
+      "eval_runtime": 1.4495,
+      "eval_samples_per_second": 0.69,
+      "eval_steps_per_second": 0.69,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00023843345913506157,
+      "loss": 0.9666,
+      "step": 3420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00023806224092062114,
+      "loss": 0.9616,
+      "step": 3440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00023769102270618077,
+      "loss": 0.9663,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00023731980449174036,
+      "loss": 0.9665,
+      "step": 3480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002369485862773,
+      "loss": 0.9662,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023657736806285962,
+      "loss": 0.9656,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00023620614984841922,
+      "loss": 0.9587,
+      "step": 3540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023583493163397882,
+      "loss": 0.9639,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023546371341953842,
+      "loss": 0.9727,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00023509249520509804,
+      "loss": 0.954,
+      "step": 3600
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.8274036645889282,
+      "eval_runtime": 1.399,
+      "eval_samples_per_second": 0.715,
+      "eval_steps_per_second": 0.715,
+      "step": 3600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023472127699065767,
+      "loss": 0.9614,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00023435005877621727,
+      "loss": 0.9573,
+      "step": 3640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002339788405617769,
+      "loss": 0.9606,
+      "step": 3660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023360762234733647,
+      "loss": 0.962,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002332364041328961,
+      "loss": 0.9591,
+      "step": 3700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023286518591845572,
+      "loss": 0.9547,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023249396770401532,
+      "loss": 0.9627,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023212274948957495,
+      "loss": 0.9578,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023175153127513457,
+      "loss": 0.9623,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023138031306069415,
+      "loss": 0.946,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.8205245137214661,
+      "eval_runtime": 1.3808,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.724,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00023100909484625377,
+      "loss": 0.9734,
+      "step": 3820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00023063787663181337,
+      "loss": 0.964,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000230266658417373,
+      "loss": 0.9576,
+      "step": 3860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022989544020293263,
+      "loss": 0.9512,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022952422198849222,
+      "loss": 0.9555,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00022915300377405182,
+      "loss": 0.95,
+      "step": 3920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022878178555961142,
+      "loss": 0.9617,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022841056734517105,
+      "loss": 0.9636,
+      "step": 3960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00022803934913073068,
+      "loss": 0.955,
+      "step": 3980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00022766813091629028,
+      "loss": 0.9516,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.8116470575332642,
+      "eval_runtime": 1.3156,
+      "eval_samples_per_second": 0.76,
+      "eval_steps_per_second": 0.76,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002272969127018499,
+      "loss": 0.9517,
+      "step": 4020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00022692569448740948,
+      "loss": 0.9551,
+      "step": 4040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002265544762729691,
+      "loss": 0.9578,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00022618325805852873,
+      "loss": 0.9613,
+      "step": 4080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022581203984408833,
+      "loss": 0.9526,
+      "step": 4100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022544082162964795,
+      "loss": 0.966,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00022506960341520755,
+      "loss": 0.9504,
+      "step": 4140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00022469838520076715,
+      "loss": 0.9458,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00022432716698632678,
+      "loss": 0.9693,
+      "step": 4180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00022395594877188638,
+      "loss": 0.9566,
+      "step": 4200
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.8783697485923767,
+      "eval_runtime": 1.3242,
+      "eval_samples_per_second": 0.755,
+      "eval_steps_per_second": 0.755,
+      "step": 4200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000223584730557446,
+      "loss": 0.9736,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002232135123430056,
+      "loss": 0.9454,
+      "step": 4240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022284229412856523,
+      "loss": 0.9547,
+      "step": 4260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022247107591412483,
+      "loss": 0.9521,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00022209985769968443,
+      "loss": 0.9546,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00022172863948524406,
+      "loss": 0.9557,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00022135742127080366,
+      "loss": 0.947,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00022098620305636328,
+      "loss": 0.9576,
+      "step": 4360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002206149848419229,
+      "loss": 0.9506,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00022024376662748248,
+      "loss": 0.9537,
+      "step": 4400
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.8563671112060547,
+      "eval_runtime": 1.3666,
+      "eval_samples_per_second": 0.732,
+      "eval_steps_per_second": 0.732,
+      "step": 4400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002198725484130421,
+      "loss": 0.955,
+      "step": 4420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002195013301986017,
+      "loss": 0.9653,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00021913011198416133,
+      "loss": 0.9601,
+      "step": 4460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021875889376972096,
+      "loss": 0.9589,
+      "step": 4480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021838767555528056,
+      "loss": 0.9448,
+      "step": 4500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00021801645734084016,
+      "loss": 0.9538,
+      "step": 4520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00021764523912639976,
+      "loss": 0.958,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002172740209119594,
+      "loss": 0.9466,
+      "step": 4560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000216902802697519,
+      "loss": 0.9544,
+      "step": 4580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002165315844830786,
+      "loss": 0.9456,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.8733828663825989,
+      "eval_runtime": 1.3225,
+      "eval_samples_per_second": 0.756,
+      "eval_steps_per_second": 0.756,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00021616036626863824,
+      "loss": 0.956,
+      "step": 4620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002157891480541978,
+      "loss": 0.9614,
+      "step": 4640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00021541792983975744,
+      "loss": 0.9568,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00021504671162531706,
+      "loss": 0.9606,
+      "step": 4680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00021467549341087666,
+      "loss": 0.9375,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002143042751964363,
+      "loss": 0.938,
+      "step": 4720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00021393305698199592,
+      "loss": 0.9647,
+      "step": 4740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002135618387675555,
+      "loss": 0.96,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00021319062055311512,
+      "loss": 0.9458,
+      "step": 4780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00021281940233867472,
+      "loss": 0.9492,
+      "step": 4800
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.869820773601532,
+      "eval_runtime": 1.3761,
+      "eval_samples_per_second": 0.727,
+      "eval_steps_per_second": 0.727,
+      "step": 4800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00021244818412423434,
+      "loss": 0.9536,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00021207696590979397,
+      "loss": 0.9522,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00021170574769535357,
+      "loss": 0.9501,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00021133452948091317,
+      "loss": 0.958,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00021096331126647277,
+      "loss": 0.9585,
+      "step": 4900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002105920930520324,
+      "loss": 0.9529,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00021022087483759202,
+      "loss": 0.9463,
+      "step": 4940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00020984965662315162,
+      "loss": 0.9558,
+      "step": 4960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00020947843840871125,
+      "loss": 0.9476,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00020910722019427087,
+      "loss": 0.9423,
+      "step": 5000
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.8774290084838867,
+      "eval_runtime": 1.3657,
+      "eval_samples_per_second": 0.732,
+      "eval_steps_per_second": 0.732,
+      "step": 5000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00020873600197983045,
+      "loss": 0.9463,
+      "step": 5020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00020836478376539007,
+      "loss": 0.9389,
+      "step": 5040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00020799356555094967,
+      "loss": 0.9443,
+      "step": 5060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002076223473365093,
+      "loss": 0.9412,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00020725112912206892,
+      "loss": 0.9417,
+      "step": 5100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00020687991090762852,
+      "loss": 0.9485,
+      "step": 5120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020650869269318812,
+      "loss": 0.9479,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00020613747447874772,
+      "loss": 0.9541,
+      "step": 5160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00020576625626430735,
+      "loss": 0.9472,
+      "step": 5180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00020539503804986698,
+      "loss": 0.9382,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.9030184745788574,
+      "eval_runtime": 1.4724,
+      "eval_samples_per_second": 0.679,
+      "eval_steps_per_second": 0.679,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00020502381983542658,
+      "loss": 0.9323,
+      "step": 5220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002046526016209862,
+      "loss": 0.9435,
+      "step": 5240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00020428138340654577,
+      "loss": 0.9417,
+      "step": 5260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002039101651921054,
+      "loss": 0.9479,
+      "step": 5280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020353894697766503,
+      "loss": 0.9418,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020316772876322463,
+      "loss": 0.946,
+      "step": 5320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00020279651054878425,
+      "loss": 0.9475,
+      "step": 5340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00020242529233434388,
+      "loss": 0.94,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00020205407411990345,
+      "loss": 0.943,
+      "step": 5380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00020168285590546308,
+      "loss": 0.9358,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.8716119527816772,
+      "eval_runtime": 1.3296,
+      "eval_samples_per_second": 0.752,
+      "eval_steps_per_second": 0.752,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00020131163769102268,
+      "loss": 0.9351,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002009404194765823,
+      "loss": 0.9412,
+      "step": 5440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00020056920126214193,
+      "loss": 0.9346,
+      "step": 5460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00020019798304770153,
+      "loss": 0.9427,
+      "step": 5480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019982676483326113,
+      "loss": 0.9553,
+      "step": 5500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019945554661882073,
+      "loss": 0.9443,
+      "step": 5520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019908432840438036,
+      "loss": 0.9446,
+      "step": 5540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00019871311018993998,
+      "loss": 0.9319,
+      "step": 5560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00019834189197549958,
+      "loss": 0.9333,
+      "step": 5580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001979706737610592,
+      "loss": 0.9405,
+      "step": 5600
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 0.8719614744186401,
+      "eval_runtime": 1.2911,
+      "eval_samples_per_second": 0.775,
+      "eval_steps_per_second": 0.775,
+      "step": 5600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019759945554661878,
+      "loss": 0.9367,
+      "step": 5620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001972282373321784,
+      "loss": 0.9352,
+      "step": 5640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00019685701911773803,
+      "loss": 0.9422,
+      "step": 5660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019648580090329763,
+      "loss": 0.9543,
+      "step": 5680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00019611458268885726,
+      "loss": 0.9308,
+      "step": 5700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001957433644744169,
+      "loss": 0.9471,
+      "step": 5720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00019537214625997646,
+      "loss": 0.933,
+      "step": 5740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00019500092804553609,
+      "loss": 0.9475,
+      "step": 5760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00019462970983109569,
+      "loss": 0.9368,
+      "step": 5780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001942584916166553,
+      "loss": 0.945,
+      "step": 5800
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.9158815145492554,
+      "eval_runtime": 1.3387,
+      "eval_samples_per_second": 0.747,
+      "eval_steps_per_second": 0.747,
+      "step": 5800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00019388727340221494,
+      "loss": 0.9407,
+      "step": 5820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019351605518777454,
+      "loss": 0.9345,
+      "step": 5840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019314483697333414,
+      "loss": 0.9355,
+      "step": 5860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00019277361875889374,
+      "loss": 0.9368,
+      "step": 5880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00019240240054445336,
+      "loss": 0.9339,
+      "step": 5900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000192031182330013,
+      "loss": 0.9315,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001916599641155726,
+      "loss": 0.9382,
+      "step": 5940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00019128874590113222,
+      "loss": 0.9391,
+      "step": 5960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001909175276866918,
+      "loss": 0.9324,
+      "step": 5980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00019054630947225142,
+      "loss": 0.9346,
+      "step": 6000
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 0.8685809969902039,
+      "eval_runtime": 1.2797,
+      "eval_samples_per_second": 0.781,
+      "eval_steps_per_second": 0.781,
+      "step": 6000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00019017509125781104,
+      "loss": 0.9302,
+      "step": 6020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018980387304337064,
+      "loss": 0.9375,
+      "step": 6040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018943265482893027,
+      "loss": 0.9265,
+      "step": 6060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018906143661448987,
+      "loss": 0.9421,
+      "step": 6080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018869021840004947,
+      "loss": 0.9382,
+      "step": 6100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001883190001856091,
+      "loss": 0.9376,
+      "step": 6120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001879477819711687,
+      "loss": 0.9315,
+      "step": 6140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018757656375672832,
+      "loss": 0.9326,
+      "step": 6160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018720534554228792,
+      "loss": 0.9283,
+      "step": 6180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018683412732784755,
+      "loss": 0.9252,
+      "step": 6200
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 0.8812684416770935,
+      "eval_runtime": 1.2442,
+      "eval_samples_per_second": 0.804,
+      "eval_steps_per_second": 0.804,
+      "step": 6200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018646290911340715,
+      "loss": 0.929,
+      "step": 6220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018609169089896674,
+      "loss": 0.9303,
+      "step": 6240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018572047268452637,
+      "loss": 0.9288,
+      "step": 6260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018534925447008597,
+      "loss": 0.9308,
+      "step": 6280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001849780362556456,
+      "loss": 0.9247,
+      "step": 6300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018460681804120522,
+      "loss": 0.939,
+      "step": 6320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001842355998267648,
+      "loss": 0.927,
+      "step": 6340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018386438161232442,
+      "loss": 0.9227,
+      "step": 6360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018349316339788402,
+      "loss": 0.9275,
+      "step": 6380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018312194518344365,
+      "loss": 0.9377,
+      "step": 6400
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 0.9376851916313171,
+      "eval_runtime": 1.3094,
+      "eval_samples_per_second": 0.764,
+      "eval_steps_per_second": 0.764,
+      "step": 6400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018275072696900328,
+      "loss": 0.9386,
+      "step": 6420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018237950875456287,
+      "loss": 0.9389,
+      "step": 6440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018200829054012247,
+      "loss": 0.9323,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018163707232568207,
+      "loss": 0.9304,
+      "step": 6480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001812658541112417,
+      "loss": 0.9382,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018089463589680133,
+      "loss": 0.9344,
+      "step": 6520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018052341768236093,
+      "loss": 0.9382,
+      "step": 6540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018015219946792055,
+      "loss": 0.9298,
+      "step": 6560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017978098125348013,
+      "loss": 0.9246,
+      "step": 6580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017940976303903975,
+      "loss": 0.9349,
+      "step": 6600
+    },
+    {
+      "epoch": 1.22,
+      "eval_loss": 0.8311364650726318,
+      "eval_runtime": 1.2953,
+      "eval_samples_per_second": 0.772,
+      "eval_steps_per_second": 0.772,
+      "step": 6600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017903854482459938,
+      "loss": 0.9309,
+      "step": 6620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017866732661015898,
+      "loss": 0.9311,
+      "step": 6640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001782961083957186,
+      "loss": 0.938,
+      "step": 6660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017792489018127823,
+      "loss": 0.9324,
+      "step": 6680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017755367196683783,
+      "loss": 0.9128,
+      "step": 6700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017718245375239743,
+      "loss": 0.9407,
+      "step": 6720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017681123553795703,
+      "loss": 0.9286,
+      "step": 6740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017644001732351666,
+      "loss": 0.9298,
+      "step": 6760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017606879910907628,
+      "loss": 0.9329,
+      "step": 6780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017569758089463588,
+      "loss": 0.9331,
+      "step": 6800
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 0.8751674890518188,
+      "eval_runtime": 1.3065,
+      "eval_samples_per_second": 0.765,
+      "eval_steps_per_second": 0.765,
+      "step": 6800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001753263626801955,
+      "loss": 0.9264,
+      "step": 6820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017495514446575508,
+      "loss": 0.9345,
+      "step": 6840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001745839262513147,
+      "loss": 0.925,
+      "step": 6860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017421270803687433,
+      "loss": 0.9227,
+      "step": 6880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017384148982243393,
+      "loss": 0.9237,
+      "step": 6900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017347027160799356,
+      "loss": 0.9365,
+      "step": 6920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001730990533935532,
+      "loss": 0.9356,
+      "step": 6940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017272783517911276,
+      "loss": 0.9242,
+      "step": 6960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017235661696467239,
+      "loss": 0.9297,
+      "step": 6980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017198539875023199,
+      "loss": 0.9232,
+      "step": 7000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.8359224796295166,
+      "eval_runtime": 1.2767,
+      "eval_samples_per_second": 0.783,
+      "eval_steps_per_second": 0.783,
+      "step": 7000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001716141805357916,
+      "loss": 0.9324,
+      "step": 7020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017124296232135124,
+      "loss": 0.9222,
+      "step": 7040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017087174410691084,
+      "loss": 0.9214,
+      "step": 7060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017050052589247044,
+      "loss": 0.9214,
+      "step": 7080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017012930767803004,
+      "loss": 0.931,
+      "step": 7100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00016975808946358966,
+      "loss": 0.9218,
+      "step": 7120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001693868712491493,
+      "loss": 0.9136,
+      "step": 7140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001690156530347089,
+      "loss": 0.9284,
+      "step": 7160
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00016864443482026852,
+      "loss": 0.921,
+      "step": 7180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001682732166058281,
+      "loss": 0.9245,
+      "step": 7200
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 0.8561541438102722,
+      "eval_runtime": 1.3583,
+      "eval_samples_per_second": 0.736,
+      "eval_steps_per_second": 0.736,
+      "step": 7200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00016790199839138771,
+      "loss": 0.9259,
+      "step": 7220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016753078017694734,
+      "loss": 0.9238,
+      "step": 7240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016715956196250694,
+      "loss": 0.9237,
+      "step": 7260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00016678834374806657,
+      "loss": 0.9111,
+      "step": 7280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001664171255336262,
+      "loss": 0.9257,
+      "step": 7300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00016604590731918577,
+      "loss": 0.9128,
+      "step": 7320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001656746891047454,
+      "loss": 0.9219,
+      "step": 7340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000165303470890305,
+      "loss": 0.9163,
+      "step": 7360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016493225267586462,
+      "loss": 0.9195,
+      "step": 7380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00016456103446142425,
+      "loss": 0.9283,
+      "step": 7400
+    },
+    {
+      "epoch": 1.36,
+      "eval_loss": 0.8775261044502258,
+      "eval_runtime": 1.3808,
+      "eval_samples_per_second": 0.724,
+      "eval_steps_per_second": 0.724,
+      "step": 7400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00016418981624698385,
+      "loss": 0.9177,
+      "step": 7420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00016381859803254344,
+      "loss": 0.9113,
+      "step": 7440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016344737981810304,
+      "loss": 0.9189,
+      "step": 7460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00016307616160366267,
+      "loss": 0.9096,
+      "step": 7480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001627049433892223,
+      "loss": 0.9191,
+      "step": 7500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001623337251747819,
+      "loss": 0.9107,
+      "step": 7520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00016196250696034152,
+      "loss": 0.9173,
+      "step": 7540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001615912887459011,
+      "loss": 0.9187,
+      "step": 7560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016122007053146072,
+      "loss": 0.9187,
+      "step": 7580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00016084885231702035,
+      "loss": 0.9211,
+      "step": 7600
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 0.8760462999343872,
+      "eval_runtime": 1.4851,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.673,
+      "step": 7600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016047763410257995,
+      "loss": 0.9116,
+      "step": 7620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00016010641588813957,
+      "loss": 0.9117,
+      "step": 7640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001597351976736992,
+      "loss": 0.9184,
+      "step": 7660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00015936397945925877,
+      "loss": 0.9122,
+      "step": 7680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001589927612448184,
+      "loss": 0.9155,
+      "step": 7700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000158621543030378,
+      "loss": 0.9186,
+      "step": 7720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015825032481593763,
+      "loss": 0.9144,
+      "step": 7740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015787910660149725,
+      "loss": 0.9224,
+      "step": 7760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00015750788838705685,
+      "loss": 0.9192,
+      "step": 7780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015713667017261645,
+      "loss": 0.9106,
+      "step": 7800
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 0.8746755719184875,
+      "eval_runtime": 1.3658,
+      "eval_samples_per_second": 0.732,
+      "eval_steps_per_second": 0.732,
+      "step": 7800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00015676545195817605,
+      "loss": 0.9106,
+      "step": 7820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00015639423374373568,
+      "loss": 0.9024,
+      "step": 7840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001560230155292953,
+      "loss": 0.9055,
+      "step": 7860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001556517973148549,
+      "loss": 0.9106,
+      "step": 7880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00015528057910041453,
+      "loss": 0.9219,
+      "step": 7900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001549093608859741,
+      "loss": 0.9015,
+      "step": 7920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00015453814267153373,
+      "loss": 0.9029,
+      "step": 7940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015416692445709336,
+      "loss": 0.9095,
+      "step": 7960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00015379570624265296,
+      "loss": 0.9284,
+      "step": 7980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015342448802821258,
+      "loss": 0.9047,
+      "step": 8000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.8788416981697083,
+      "eval_runtime": 1.3646,
+      "eval_samples_per_second": 0.733,
+      "eval_steps_per_second": 0.733,
+      "step": 8000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015305326981377218,
+      "loss": 0.9151,
+      "step": 8020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00015268205159933178,
+      "loss": 0.9076,
+      "step": 8040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001523108333848914,
+      "loss": 0.9134,
+      "step": 8060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000151939615170451,
+      "loss": 0.9136,
+      "step": 8080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00015156839695601063,
+      "loss": 0.9073,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015119717874157023,
+      "loss": 0.9054,
+      "step": 8120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015082596052712986,
+      "loss": 0.9169,
+      "step": 8140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00015045474231268946,
+      "loss": 0.9249,
+      "step": 8160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015008352409824906,
+      "loss": 0.9032,
+      "step": 8180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00014971230588380869,
+      "loss": 0.9089,
+      "step": 8200
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 0.8750318288803101,
+      "eval_runtime": 1.361,
+      "eval_samples_per_second": 0.735,
+      "eval_steps_per_second": 0.735,
+      "step": 8200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00014934108766936828,
+      "loss": 0.903,
+      "step": 8220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001489698694549279,
+      "loss": 0.9045,
+      "step": 8240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001485986512404875,
+      "loss": 0.9146,
+      "step": 8260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014822743302604714,
+      "loss": 0.9085,
+      "step": 8280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014785621481160674,
+      "loss": 0.9105,
+      "step": 8300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00014748499659716634,
+      "loss": 0.9034,
+      "step": 8320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00014711377838272596,
+      "loss": 0.906,
+      "step": 8340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001467425601682856,
+      "loss": 0.9055,
+      "step": 8360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001463713419538452,
+      "loss": 0.9013,
+      "step": 8380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001460001237394048,
+      "loss": 0.9136,
+      "step": 8400
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 0.8716868162155151,
+      "eval_runtime": 1.3893,
+      "eval_samples_per_second": 0.72,
+      "eval_steps_per_second": 0.72,
+      "step": 8400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00014562890552496441,
+      "loss": 0.8991,
+      "step": 8420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014525768731052401,
+      "loss": 0.9072,
+      "step": 8440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014488646909608364,
+      "loss": 0.9185,
+      "step": 8460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00014451525088164324,
+      "loss": 0.9097,
+      "step": 8480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00014414403266720284,
+      "loss": 0.9044,
+      "step": 8500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00014377281445276247,
+      "loss": 0.9095,
+      "step": 8520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001434015962383221,
+      "loss": 0.9119,
+      "step": 8540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001430303780238817,
+      "loss": 0.8971,
+      "step": 8560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001426591598094413,
+      "loss": 0.9011,
+      "step": 8580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014228794159500092,
+      "loss": 0.9048,
+      "step": 8600
+    },
+    {
+      "epoch": 1.59,
+      "eval_loss": 0.8536492586135864,
+      "eval_runtime": 1.2569,
+      "eval_samples_per_second": 0.796,
+      "eval_steps_per_second": 0.796,
+      "step": 8600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014191672338056052,
+      "loss": 0.9031,
+      "step": 8620
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00014154550516612014,
+      "loss": 0.9078,
+      "step": 8640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00014117428695167974,
+      "loss": 0.9115,
+      "step": 8660
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00014080306873723934,
+      "loss": 0.9018,
+      "step": 8680
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00014043185052279897,
+      "loss": 0.8974,
+      "step": 8700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001400606323083586,
+      "loss": 0.9084,
+      "step": 8720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001396894140939182,
+      "loss": 0.9041,
+      "step": 8740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001393181958794778,
+      "loss": 0.8967,
+      "step": 8760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013894697766503742,
+      "loss": 0.8969,
+      "step": 8780
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013857575945059702,
+      "loss": 0.8993,
+      "step": 8800
+    },
+    {
+      "epoch": 1.62,
+      "eval_loss": 0.8307236433029175,
+      "eval_runtime": 1.2771,
+      "eval_samples_per_second": 0.783,
+      "eval_steps_per_second": 0.783,
+      "step": 8800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013820454123615665,
+      "loss": 0.9031,
+      "step": 8820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013783332302171625,
+      "loss": 0.9008,
+      "step": 8840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013746210480727585,
+      "loss": 0.9039,
+      "step": 8860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013709088659283547,
+      "loss": 0.9063,
+      "step": 8880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001367196683783951,
+      "loss": 0.901,
+      "step": 8900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001363484501639547,
+      "loss": 0.8973,
+      "step": 8920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001359772319495143,
+      "loss": 0.9057,
+      "step": 8940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013560601373507393,
+      "loss": 0.898,
+      "step": 8960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013523479552063353,
+      "loss": 0.9027,
+      "step": 8980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013486357730619315,
+      "loss": 0.9084,
+      "step": 9000
+    },
+    {
+      "epoch": 1.66,
+      "eval_loss": 0.8812193274497986,
+      "eval_runtime": 1.2199,
+      "eval_samples_per_second": 0.82,
+      "eval_steps_per_second": 0.82,
+      "step": 9000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013449235909175275,
+      "loss": 0.9043,
+      "step": 9020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013412114087731235,
+      "loss": 0.9021,
+      "step": 9040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013374992266287198,
+      "loss": 0.8951,
+      "step": 9060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001333787044484316,
+      "loss": 0.8865,
+      "step": 9080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001330074862339912,
+      "loss": 0.9027,
+      "step": 9100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001326362680195508,
+      "loss": 0.8976,
+      "step": 9120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013226504980511043,
+      "loss": 0.8911,
+      "step": 9140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013189383159067003,
+      "loss": 0.8973,
+      "step": 9160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013152261337622966,
+      "loss": 0.9076,
+      "step": 9180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013115139516178925,
+      "loss": 0.899,
+      "step": 9200
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 0.82787024974823,
+      "eval_runtime": 1.2643,
+      "eval_samples_per_second": 0.791,
+      "eval_steps_per_second": 0.791,
+      "step": 9200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013078017694734885,
+      "loss": 0.9022,
+      "step": 9220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013040895873290848,
+      "loss": 0.8969,
+      "step": 9240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001300377405184681,
+      "loss": 0.895,
+      "step": 9260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001296665223040277,
+      "loss": 0.9034,
+      "step": 9280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001292953040895873,
+      "loss": 0.8892,
+      "step": 9300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012892408587514693,
+      "loss": 0.8956,
+      "step": 9320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012855286766070656,
+      "loss": 0.8952,
+      "step": 9340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012818164944626616,
+      "loss": 0.89,
+      "step": 9360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012781043123182576,
+      "loss": 0.9029,
+      "step": 9380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012743921301738539,
+      "loss": 0.8954,
+      "step": 9400
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 0.8729041218757629,
+      "eval_runtime": 1.2221,
+      "eval_samples_per_second": 0.818,
+      "eval_steps_per_second": 0.818,
+      "step": 9400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00012706799480294498,
+      "loss": 0.8805,
+      "step": 9420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001266967765885046,
+      "loss": 0.8936,
+      "step": 9440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001263255583740642,
+      "loss": 0.8893,
+      "step": 9460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001259543401596238,
+      "loss": 0.8953,
+      "step": 9480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012558312194518344,
+      "loss": 0.8943,
+      "step": 9500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012521190373074306,
+      "loss": 0.8766,
+      "step": 9520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012484068551630266,
+      "loss": 0.899,
+      "step": 9540
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012446946730186226,
+      "loss": 0.8949,
+      "step": 9560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001240982490874219,
+      "loss": 0.8849,
+      "step": 9580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001237270308729815,
+      "loss": 0.894,
+      "step": 9600
+    },
+    {
+      "epoch": 1.77,
+      "eval_loss": 0.8376001715660095,
+      "eval_runtime": 1.3277,
+      "eval_samples_per_second": 0.753,
+      "eval_steps_per_second": 0.753,
+      "step": 9600
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012335581265854111,
+      "loss": 0.9029,
+      "step": 9620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012298459444410071,
+      "loss": 0.898,
+      "step": 9640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012261337622966031,
+      "loss": 0.8848,
+      "step": 9660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012224215801521994,
+      "loss": 0.8932,
+      "step": 9680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012187093980077955,
+      "loss": 0.891,
+      "step": 9700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012149972158633915,
+      "loss": 0.8936,
+      "step": 9720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012112850337189877,
+      "loss": 0.8907,
+      "step": 9740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012075728515745839,
+      "loss": 0.8958,
+      "step": 9760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012038606694301799,
+      "loss": 0.8912,
+      "step": 9780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001200148487285776,
+      "loss": 0.9032,
+      "step": 9800
+    },
+    {
+      "epoch": 1.81,
+      "eval_loss": 0.8445655703544617,
+      "eval_runtime": 1.3427,
+      "eval_samples_per_second": 0.745,
+      "eval_steps_per_second": 0.745,
+      "step": 9800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011964363051413722,
+      "loss": 0.8918,
+      "step": 9820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011927241229969682,
+      "loss": 0.8865,
+      "step": 9840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011890119408525644,
+      "loss": 0.8893,
+      "step": 9860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011852997587081606,
+      "loss": 0.8863,
+      "step": 9880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011815875765637566,
+      "loss": 0.8978,
+      "step": 9900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011778753944193527,
+      "loss": 0.886,
+      "step": 9920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001174163212274949,
+      "loss": 0.8789,
+      "step": 9940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001170451030130545,
+      "loss": 0.8944,
+      "step": 9960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011667388479861411,
+      "loss": 0.8964,
+      "step": 9980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011630266658417372,
+      "loss": 0.8921,
+      "step": 10000
+    },
+    {
+      "epoch": 1.84,
+      "eval_loss": 0.8549759387969971,
+      "eval_runtime": 1.2407,
+      "eval_samples_per_second": 0.806,
+      "eval_steps_per_second": 0.806,
+      "step": 10000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011593144836973332,
+      "loss": 0.8918,
+      "step": 10020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011556023015529295,
+      "loss": 0.8966,
+      "step": 10040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011518901194085256,
+      "loss": 0.8919,
+      "step": 10060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011481779372641216,
+      "loss": 0.8963,
+      "step": 10080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011444657551197177,
+      "loss": 0.8805,
+      "step": 10100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001140753572975314,
+      "loss": 0.8837,
+      "step": 10120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000113704139083091,
+      "loss": 0.8919,
+      "step": 10140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00011333292086865061,
+      "loss": 0.8904,
+      "step": 10160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011296170265421023,
+      "loss": 0.8765,
+      "step": 10180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011259048443976982,
+      "loss": 0.8846,
+      "step": 10200
+    },
+    {
+      "epoch": 1.88,
+      "eval_loss": 0.8317739963531494,
+      "eval_runtime": 1.3125,
+      "eval_samples_per_second": 0.762,
+      "eval_steps_per_second": 0.762,
+      "step": 10200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00011221926622532945,
+      "loss": 0.8754,
+      "step": 10220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011184804801088906,
+      "loss": 0.8824,
+      "step": 10240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00011147682979644866,
+      "loss": 0.888,
+      "step": 10260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00011110561158200828,
+      "loss": 0.8986,
+      "step": 10280
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001107343933675679,
+      "loss": 0.8764,
+      "step": 10300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001103631751531275,
+      "loss": 0.8901,
+      "step": 10320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010999195693868712,
+      "loss": 0.8859,
+      "step": 10340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010962073872424673,
+      "loss": 0.8868,
+      "step": 10360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010924952050980633,
+      "loss": 0.8827,
+      "step": 10380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010887830229536595,
+      "loss": 0.8911,
+      "step": 10400
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 0.8604958057403564,
+      "eval_runtime": 1.2674,
+      "eval_samples_per_second": 0.789,
+      "eval_steps_per_second": 0.789,
+      "step": 10400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010850708408092557,
+      "loss": 0.8888,
+      "step": 10420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010813586586648517,
+      "loss": 0.8878,
+      "step": 10440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010776464765204478,
+      "loss": 0.8853,
+      "step": 10460
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001073934294376044,
+      "loss": 0.8831,
+      "step": 10480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000107022211223164,
+      "loss": 0.8906,
+      "step": 10500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010665099300872362,
+      "loss": 0.8869,
+      "step": 10520
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010627977479428323,
+      "loss": 0.882,
+      "step": 10540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010590855657984283,
+      "loss": 0.8789,
+      "step": 10560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010553733836540246,
+      "loss": 0.8819,
+      "step": 10580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010516612015096207,
+      "loss": 0.8765,
+      "step": 10600
+    },
+    {
+      "epoch": 1.95,
+      "eval_loss": 0.8246310949325562,
+      "eval_runtime": 1.3918,
+      "eval_samples_per_second": 0.719,
+      "eval_steps_per_second": 0.719,
+      "step": 10600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010479490193652167,
+      "loss": 0.8754,
+      "step": 10620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010442368372208128,
+      "loss": 0.8876,
+      "step": 10640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001040524655076409,
+      "loss": 0.8871,
+      "step": 10660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010368124729320051,
+      "loss": 0.879,
+      "step": 10680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010331002907876012,
+      "loss": 0.8809,
+      "step": 10700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010293881086431974,
+      "loss": 0.8847,
+      "step": 10720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010256759264987934,
+      "loss": 0.8861,
+      "step": 10740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010219637443543895,
+      "loss": 0.8758,
+      "step": 10760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010182515622099858,
+      "loss": 0.8781,
+      "step": 10780
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010145393800655817,
+      "loss": 0.8726,
+      "step": 10800
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.846712052822113,
+      "eval_runtime": 1.3944,
+      "eval_samples_per_second": 0.717,
+      "eval_steps_per_second": 0.717,
+      "step": 10800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00010108271979211779,
+      "loss": 0.8774,
+      "step": 10820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001007115015776774,
+      "loss": 0.8755,
+      "step": 10840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000100340283363237,
+      "loss": 0.8814,
+      "step": 10860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.996906514879663e-05,
+      "loss": 0.8694,
+      "step": 10880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.959784693435624e-05,
+      "loss": 0.8808,
+      "step": 10900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.922662871991584e-05,
+      "loss": 0.8824,
+      "step": 10920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.885541050547545e-05,
+      "loss": 0.8669,
+      "step": 10940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.848419229103508e-05,
+      "loss": 0.8702,
+      "step": 10960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.811297407659468e-05,
+      "loss": 0.8722,
+      "step": 10980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.774175586215429e-05,
+      "loss": 0.8844,
+      "step": 11000
+    },
+    {
+      "epoch": 2.03,
+      "eval_loss": 0.8575500845909119,
+      "eval_runtime": 1.622,
+      "eval_samples_per_second": 0.617,
+      "eval_steps_per_second": 0.617,
+      "step": 11000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.73705376477139e-05,
+      "loss": 0.877,
+      "step": 11020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.699931943327353e-05,
+      "loss": 0.8635,
+      "step": 11040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.662810121883313e-05,
+      "loss": 0.8729,
+      "step": 11060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.625688300439274e-05,
+      "loss": 0.865,
+      "step": 11080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.588566478995236e-05,
+      "loss": 0.8811,
+      "step": 11100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.551444657551196e-05,
+      "loss": 0.8729,
+      "step": 11120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.514322836107158e-05,
+      "loss": 0.8865,
+      "step": 11140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.47720101466312e-05,
+      "loss": 0.8747,
+      "step": 11160
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.44007919321908e-05,
+      "loss": 0.8779,
+      "step": 11180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.402957371775041e-05,
+      "loss": 0.8764,
+      "step": 11200
+    },
+    {
+      "epoch": 2.07,
+      "eval_loss": 0.8142430186271667,
+      "eval_runtime": 1.4953,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.669,
+      "step": 11200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.365835550331003e-05,
+      "loss": 0.8779,
+      "step": 11220
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.328713728886963e-05,
+      "loss": 0.8702,
+      "step": 11240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.291591907442925e-05,
+      "loss": 0.8832,
+      "step": 11260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.254470085998886e-05,
+      "loss": 0.8718,
+      "step": 11280
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.217348264554846e-05,
+      "loss": 0.8726,
+      "step": 11300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.180226443110809e-05,
+      "loss": 0.8677,
+      "step": 11320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.14310462166677e-05,
+      "loss": 0.8748,
+      "step": 11340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.10598280022273e-05,
+      "loss": 0.8694,
+      "step": 11360
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.068860978778691e-05,
+      "loss": 0.8613,
+      "step": 11380
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.031739157334652e-05,
+      "loss": 0.8687,
+      "step": 11400
+    },
+    {
+      "epoch": 2.1,
+      "eval_loss": 0.827206552028656,
+      "eval_runtime": 1.6259,
+      "eval_samples_per_second": 0.615,
+      "eval_steps_per_second": 0.615,
+      "step": 11400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.994617335890614e-05,
+      "loss": 0.8735,
+      "step": 11420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.957495514446575e-05,
+      "loss": 0.8691,
+      "step": 11440
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.920373693002536e-05,
+      "loss": 0.8605,
+      "step": 11460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.883251871558496e-05,
+      "loss": 0.8715,
+      "step": 11480
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.846130050114459e-05,
+      "loss": 0.8541,
+      "step": 11500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.80900822867042e-05,
+      "loss": 0.8668,
+      "step": 11520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.77188640722638e-05,
+      "loss": 0.8745,
+      "step": 11540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.734764585782342e-05,
+      "loss": 0.8781,
+      "step": 11560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.697642764338303e-05,
+      "loss": 0.8782,
+      "step": 11580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.660520942894264e-05,
+      "loss": 0.8733,
+      "step": 11600
+    },
+    {
+      "epoch": 2.14,
+      "eval_loss": 0.8250707983970642,
+      "eval_runtime": 1.5297,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.654,
+      "step": 11600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.623399121450225e-05,
+      "loss": 0.8771,
+      "step": 11620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.586277300006187e-05,
+      "loss": 0.8809,
+      "step": 11640
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.549155478562147e-05,
+      "loss": 0.8663,
+      "step": 11660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.512033657118108e-05,
+      "loss": 0.8565,
+      "step": 11680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.47491183567407e-05,
+      "loss": 0.8688,
+      "step": 11700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.43779001423003e-05,
+      "loss": 0.8712,
+      "step": 11720
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.400668192785992e-05,
+      "loss": 0.8692,
+      "step": 11740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.363546371341953e-05,
+      "loss": 0.8639,
+      "step": 11760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.326424549897913e-05,
+      "loss": 0.8598,
+      "step": 11780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.289302728453876e-05,
+      "loss": 0.8675,
+      "step": 11800
+    },
+    {
+      "epoch": 2.18,
+      "eval_loss": 0.8465573191642761,
+      "eval_runtime": 1.6002,
+      "eval_samples_per_second": 0.625,
+      "eval_steps_per_second": 0.625,
+      "step": 11800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.252180907009837e-05,
+      "loss": 0.8667,
+      "step": 11820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.215059085565797e-05,
+      "loss": 0.8696,
+      "step": 11840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.177937264121758e-05,
+      "loss": 0.8691,
+      "step": 11860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.140815442677721e-05,
+      "loss": 0.8649,
+      "step": 11880
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.103693621233681e-05,
+      "loss": 0.8605,
+      "step": 11900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.066571799789642e-05,
+      "loss": 0.8698,
+      "step": 11920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.029449978345604e-05,
+      "loss": 0.8595,
+      "step": 11940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.992328156901564e-05,
+      "loss": 0.867,
+      "step": 11960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.955206335457526e-05,
+      "loss": 0.8662,
+      "step": 11980
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.918084514013487e-05,
+      "loss": 0.8747,
+      "step": 12000
+    },
+    {
+      "epoch": 2.21,
+      "eval_loss": 0.8371507525444031,
+      "eval_runtime": 1.5617,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.64,
+      "step": 12000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.880962692569447e-05,
+      "loss": 0.8635,
+      "step": 12020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.843840871125409e-05,
+      "loss": 0.8648,
+      "step": 12040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.806719049681371e-05,
+      "loss": 0.8653,
+      "step": 12060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.769597228237331e-05,
+      "loss": 0.8619,
+      "step": 12080
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.732475406793293e-05,
+      "loss": 0.8616,
+      "step": 12100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.695353585349254e-05,
+      "loss": 0.8593,
+      "step": 12120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.658231763905214e-05,
+      "loss": 0.8683,
+      "step": 12140
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.621109942461177e-05,
+      "loss": 0.8641,
+      "step": 12160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.583988121017138e-05,
+      "loss": 0.8596,
+      "step": 12180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.546866299573098e-05,
+      "loss": 0.8626,
+      "step": 12200
+    },
+    {
+      "epoch": 2.25,
+      "eval_loss": 0.8443911075592041,
+      "eval_runtime": 1.578,
+      "eval_samples_per_second": 0.634,
+      "eval_steps_per_second": 0.634,
+      "step": 12200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.509744478129059e-05,
+      "loss": 0.8568,
+      "step": 12220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.47262265668502e-05,
+      "loss": 0.8688,
+      "step": 12240
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.435500835240982e-05,
+      "loss": 0.8667,
+      "step": 12260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.398379013796943e-05,
+      "loss": 0.8638,
+      "step": 12280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.361257192352904e-05,
+      "loss": 0.8633,
+      "step": 12300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.324135370908866e-05,
+      "loss": 0.8716,
+      "step": 12320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.287013549464827e-05,
+      "loss": 0.862,
+      "step": 12340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.249891728020788e-05,
+      "loss": 0.8673,
+      "step": 12360
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.21276990657675e-05,
+      "loss": 0.8518,
+      "step": 12380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.17564808513271e-05,
+      "loss": 0.8573,
+      "step": 12400
+    },
+    {
+      "epoch": 2.29,
+      "eval_loss": 0.8218865394592285,
+      "eval_runtime": 1.5777,
+      "eval_samples_per_second": 0.634,
+      "eval_steps_per_second": 0.634,
+      "step": 12400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.138526263688672e-05,
+      "loss": 0.8647,
+      "step": 12420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.101404442244632e-05,
+      "loss": 0.8584,
+      "step": 12440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.064282620800593e-05,
+      "loss": 0.8642,
+      "step": 12460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.027160799356555e-05,
+      "loss": 0.8612,
+      "step": 12480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.990038977912516e-05,
+      "loss": 0.8594,
+      "step": 12500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.952917156468477e-05,
+      "loss": 0.856,
+      "step": 12520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.915795335024439e-05,
+      "loss": 0.8527,
+      "step": 12540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.878673513580399e-05,
+      "loss": 0.8721,
+      "step": 12560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.84155169213636e-05,
+      "loss": 0.8555,
+      "step": 12580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.804429870692321e-05,
+      "loss": 0.8666,
+      "step": 12600
+    },
+    {
+      "epoch": 2.32,
+      "eval_loss": 0.8119338750839233,
+      "eval_runtime": 1.4548,
+      "eval_samples_per_second": 0.687,
+      "eval_steps_per_second": 0.687,
+      "step": 12600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.767308049248282e-05,
+      "loss": 0.8531,
+      "step": 12620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.730186227804244e-05,
+      "loss": 0.8558,
+      "step": 12640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.693064406360205e-05,
+      "loss": 0.8634,
+      "step": 12660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.655942584916166e-05,
+      "loss": 0.862,
+      "step": 12680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.618820763472126e-05,
+      "loss": 0.8527,
+      "step": 12700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.581698942028089e-05,
+      "loss": 0.8564,
+      "step": 12720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.544577120584049e-05,
+      "loss": 0.8632,
+      "step": 12740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.50745529914001e-05,
+      "loss": 0.8497,
+      "step": 12760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.470333477695971e-05,
+      "loss": 0.8684,
+      "step": 12780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.433211656251933e-05,
+      "loss": 0.8625,
+      "step": 12800
+    },
+    {
+      "epoch": 2.36,
+      "eval_loss": 0.8092614412307739,
+      "eval_runtime": 1.351,
+      "eval_samples_per_second": 0.74,
+      "eval_steps_per_second": 0.74,
+      "step": 12800
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.396089834807894e-05,
+      "loss": 0.8472,
+      "step": 12820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.358968013363855e-05,
+      "loss": 0.8463,
+      "step": 12840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.321846191919817e-05,
+      "loss": 0.8503,
+      "step": 12860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.284724370475777e-05,
+      "loss": 0.8501,
+      "step": 12880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.247602549031739e-05,
+      "loss": 0.8577,
+      "step": 12900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.210480727587699e-05,
+      "loss": 0.8644,
+      "step": 12920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.17335890614366e-05,
+      "loss": 0.8491,
+      "step": 12940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.136237084699622e-05,
+      "loss": 0.8589,
+      "step": 12960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.099115263255583e-05,
+      "loss": 0.8558,
+      "step": 12980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.061993441811544e-05,
+      "loss": 0.8523,
+      "step": 13000
+    },
+    {
+      "epoch": 2.4,
+      "eval_loss": 0.8120648264884949,
+      "eval_runtime": 1.35,
+      "eval_samples_per_second": 0.741,
+      "eval_steps_per_second": 0.741,
+      "step": 13000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.024871620367506e-05,
+      "loss": 0.8638,
+      "step": 13020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.9877497989234664e-05,
+      "loss": 0.8553,
+      "step": 13040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.950627977479428e-05,
+      "loss": 0.8514,
+      "step": 13060
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.913506156035389e-05,
+      "loss": 0.847,
+      "step": 13080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.87638433459135e-05,
+      "loss": 0.8626,
+      "step": 13100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.839262513147311e-05,
+      "loss": 0.8501,
+      "step": 13120
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.802140691703273e-05,
+      "loss": 0.8546,
+      "step": 13140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7650188702592335e-05,
+      "loss": 0.85,
+      "step": 13160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.727897048815194e-05,
+      "loss": 0.8572,
+      "step": 13180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.690775227371156e-05,
+      "loss": 0.8407,
+      "step": 13200
+    },
+    {
+      "epoch": 2.43,
+      "eval_loss": 0.8303215503692627,
+      "eval_runtime": 1.3376,
+      "eval_samples_per_second": 0.748,
+      "eval_steps_per_second": 0.748,
+      "step": 13200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.653653405927117e-05,
+      "loss": 0.8494,
+      "step": 13220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.616531584483079e-05,
+      "loss": 0.8517,
+      "step": 13240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.579409763039039e-05,
+      "loss": 0.8472,
+      "step": 13260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.542287941595e-05,
+      "loss": 0.8573,
+      "step": 13280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.505166120150962e-05,
+      "loss": 0.8492,
+      "step": 13300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4680442987069226e-05,
+      "loss": 0.8613,
+      "step": 13320
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.430922477262884e-05,
+      "loss": 0.8469,
+      "step": 13340
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.393800655818845e-05,
+      "loss": 0.854,
+      "step": 13360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.3566788343748065e-05,
+      "loss": 0.8404,
+      "step": 13380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.319557012930767e-05,
+      "loss": 0.8415,
+      "step": 13400
+    },
+    {
+      "epoch": 2.47,
+      "eval_loss": 0.7983837723731995,
+      "eval_runtime": 1.3234,
+      "eval_samples_per_second": 0.756,
+      "eval_steps_per_second": 0.756,
+      "step": 13400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.282435191486729e-05,
+      "loss": 0.8442,
+      "step": 13420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.24531337004269e-05,
+      "loss": 0.8515,
+      "step": 13440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.20819154859865e-05,
+      "loss": 0.8472,
+      "step": 13460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.171069727154612e-05,
+      "loss": 0.8557,
+      "step": 13480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.133947905710573e-05,
+      "loss": 0.8531,
+      "step": 13500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.096826084266534e-05,
+      "loss": 0.841,
+      "step": 13520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.0597042628224955e-05,
+      "loss": 0.8574,
+      "step": 13540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.022582441378457e-05,
+      "loss": 0.8484,
+      "step": 13560
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.9854606199344175e-05,
+      "loss": 0.8457,
+      "step": 13580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.9483387984903794e-05,
+      "loss": 0.8457,
+      "step": 13600
+    },
+    {
+      "epoch": 2.51,
+      "eval_loss": 0.823489248752594,
+      "eval_runtime": 1.2716,
+      "eval_samples_per_second": 0.786,
+      "eval_steps_per_second": 0.786,
+      "step": 13600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.91121697704634e-05,
+      "loss": 0.8461,
+      "step": 13620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.874095155602301e-05,
+      "loss": 0.8441,
+      "step": 13640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.836973334158263e-05,
+      "loss": 0.8401,
+      "step": 13660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.799851512714223e-05,
+      "loss": 0.838,
+      "step": 13680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.7627296912701846e-05,
+      "loss": 0.846,
+      "step": 13700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.725607869826146e-05,
+      "loss": 0.8529,
+      "step": 13720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.6884860483821065e-05,
+      "loss": 0.8396,
+      "step": 13740
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.651364226938068e-05,
+      "loss": 0.8367,
+      "step": 13760
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.614242405494029e-05,
+      "loss": 0.8469,
+      "step": 13780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5771205840499904e-05,
+      "loss": 0.8463,
+      "step": 13800
+    },
+    {
+      "epoch": 2.55,
+      "eval_loss": 0.8162570595741272,
+      "eval_runtime": 1.311,
+      "eval_samples_per_second": 0.763,
+      "eval_steps_per_second": 0.763,
+      "step": 13800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.539998762605951e-05,
+      "loss": 0.8409,
+      "step": 13820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.502876941161913e-05,
+      "loss": 0.8465,
+      "step": 13840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.4657551197178737e-05,
+      "loss": 0.8468,
+      "step": 13860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.428633298273834e-05,
+      "loss": 0.8405,
+      "step": 13880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.391511476829796e-05,
+      "loss": 0.8369,
+      "step": 13900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.354389655385757e-05,
+      "loss": 0.8488,
+      "step": 13920
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.317267833941718e-05,
+      "loss": 0.8451,
+      "step": 13940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.2801460124976795e-05,
+      "loss": 0.85,
+      "step": 13960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.243024191053641e-05,
+      "loss": 0.8414,
+      "step": 13980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.2059023696096014e-05,
+      "loss": 0.8397,
+      "step": 14000
+    },
+    {
+      "epoch": 2.58,
+      "eval_loss": 0.8420786261558533,
+      "eval_runtime": 1.2958,
+      "eval_samples_per_second": 0.772,
+      "eval_steps_per_second": 0.772,
+      "step": 14000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1687805481655634e-05,
+      "loss": 0.8465,
+      "step": 14020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.131658726721524e-05,
+      "loss": 0.8441,
+      "step": 14040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.0945369052774846e-05,
+      "loss": 0.8418,
+      "step": 14060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.0574150838334466e-05,
+      "loss": 0.8301,
+      "step": 14080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.020293262389407e-05,
+      "loss": 0.837,
+      "step": 14100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.983171440945369e-05,
+      "loss": 0.8387,
+      "step": 14120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.94604961950133e-05,
+      "loss": 0.8525,
+      "step": 14140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.908927798057291e-05,
+      "loss": 0.8406,
+      "step": 14160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.8718059766132525e-05,
+      "loss": 0.8351,
+      "step": 14180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.834684155169213e-05,
+      "loss": 0.8357,
+      "step": 14200
+    },
+    {
+      "epoch": 2.62,
+      "eval_loss": 0.8169363737106323,
+      "eval_runtime": 1.3056,
+      "eval_samples_per_second": 0.766,
+      "eval_steps_per_second": 0.766,
+      "step": 14200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7975623337251744e-05,
+      "loss": 0.8501,
+      "step": 14220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.760440512281136e-05,
+      "loss": 0.8351,
+      "step": 14240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.723318690837097e-05,
+      "loss": 0.832,
+      "step": 14260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6861968693930576e-05,
+      "loss": 0.8365,
+      "step": 14280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.649075047949019e-05,
+      "loss": 0.8353,
+      "step": 14300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.61195322650498e-05,
+      "loss": 0.8344,
+      "step": 14320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.5748314050609415e-05,
+      "loss": 0.8441,
+      "step": 14340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.537709583616902e-05,
+      "loss": 0.8413,
+      "step": 14360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5005877621728634e-05,
+      "loss": 0.8328,
+      "step": 14380
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.463465940728825e-05,
+      "loss": 0.8473,
+      "step": 14400
+    },
+    {
+      "epoch": 2.66,
+      "eval_loss": 0.8171071410179138,
+      "eval_runtime": 1.2752,
+      "eval_samples_per_second": 0.784,
+      "eval_steps_per_second": 0.784,
+      "step": 14400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.426344119284786e-05,
+      "loss": 0.8369,
+      "step": 14420
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.3892222978407473e-05,
+      "loss": 0.8406,
+      "step": 14440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3521004763967087e-05,
+      "loss": 0.8292,
+      "step": 14460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.314978654952669e-05,
+      "loss": 0.8449,
+      "step": 14480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.2778568335086306e-05,
+      "loss": 0.8399,
+      "step": 14500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.240735012064592e-05,
+      "loss": 0.835,
+      "step": 14520
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2036131906205525e-05,
+      "loss": 0.8416,
+      "step": 14540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.166491369176514e-05,
+      "loss": 0.8369,
+      "step": 14560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.129369547732475e-05,
+      "loss": 0.8347,
+      "step": 14580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.0922477262884364e-05,
+      "loss": 0.8392,
+      "step": 14600
+    },
+    {
+      "epoch": 2.69,
+      "eval_loss": 0.8220583200454712,
+      "eval_runtime": 1.3692,
+      "eval_samples_per_second": 0.73,
+      "eval_steps_per_second": 0.73,
+      "step": 14600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.055125904844398e-05,
+      "loss": 0.8413,
+      "step": 14620
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0180040834003587e-05,
+      "loss": 0.8428,
+      "step": 14640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9808822619563196e-05,
+      "loss": 0.8362,
+      "step": 14660
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.943760440512281e-05,
+      "loss": 0.8324,
+      "step": 14680
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9066386190682422e-05,
+      "loss": 0.8363,
+      "step": 14700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.869516797624203e-05,
+      "loss": 0.841,
+      "step": 14720
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.832394976180164e-05,
+      "loss": 0.8357,
+      "step": 14740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7952731547361255e-05,
+      "loss": 0.8309,
+      "step": 14760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7581513332920868e-05,
+      "loss": 0.8361,
+      "step": 14780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7210295118480477e-05,
+      "loss": 0.8404,
+      "step": 14800
+    },
+    {
+      "epoch": 2.73,
+      "eval_loss": 0.8303062915802002,
+      "eval_runtime": 1.4178,
+      "eval_samples_per_second": 0.705,
+      "eval_steps_per_second": 0.705,
+      "step": 14800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.683907690404009e-05,
+      "loss": 0.8418,
+      "step": 14820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6467858689599703e-05,
+      "loss": 0.8466,
+      "step": 14840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.609664047515931e-05,
+      "loss": 0.8288,
+      "step": 14860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5725422260718923e-05,
+      "loss": 0.8311,
+      "step": 14880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5354204046278536e-05,
+      "loss": 0.8369,
+      "step": 14900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4982985831838145e-05,
+      "loss": 0.8275,
+      "step": 14920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.461176761739776e-05,
+      "loss": 0.8283,
+      "step": 14940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.424054940295737e-05,
+      "loss": 0.8367,
+      "step": 14960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.386933118851698e-05,
+      "loss": 0.8247,
+      "step": 14980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3498112974076594e-05,
+      "loss": 0.8405,
+      "step": 15000
+    },
+    {
+      "epoch": 2.77,
+      "eval_loss": 0.828734815120697,
+      "eval_runtime": 1.3774,
+      "eval_samples_per_second": 0.726,
+      "eval_steps_per_second": 0.726,
+      "step": 15000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3126894759636204e-05,
+      "loss": 0.822,
+      "step": 15020
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2774237455917836e-05,
+      "loss": 0.8458,
+      "step": 15040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.240301924147745e-05,
+      "loss": 0.8282,
+      "step": 15060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2031801027037055e-05,
+      "loss": 0.8241,
+      "step": 15080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.166058281259667e-05,
+      "loss": 0.8322,
+      "step": 15100
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.128936459815628e-05,
+      "loss": 0.8343,
+      "step": 15120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.091814638371589e-05,
+      "loss": 0.8238,
+      "step": 15140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0546928169275504e-05,
+      "loss": 0.8381,
+      "step": 15160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0175709954835117e-05,
+      "loss": 0.8301,
+      "step": 15180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.980449174039473e-05,
+      "loss": 0.8177,
+      "step": 15200
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 0.8177865743637085,
+      "eval_runtime": 1.3723,
+      "eval_samples_per_second": 0.729,
+      "eval_steps_per_second": 0.729,
+      "step": 15200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9433273525954336e-05,
+      "loss": 0.8324,
+      "step": 15220
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.906205531151395e-05,
+      "loss": 0.83,
+      "step": 15240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8690837097073562e-05,
+      "loss": 0.8283,
+      "step": 15260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8319618882633172e-05,
+      "loss": 0.8262,
+      "step": 15280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7948400668192785e-05,
+      "loss": 0.8361,
+      "step": 15300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7577182453752395e-05,
+      "loss": 0.832,
+      "step": 15320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7205964239312008e-05,
+      "loss": 0.8294,
+      "step": 15340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.683474602487162e-05,
+      "loss": 0.8215,
+      "step": 15360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.646352781043123e-05,
+      "loss": 0.8337,
+      "step": 15380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6092309595990843e-05,
+      "loss": 0.8274,
+      "step": 15400
+    },
+    {
+      "epoch": 2.84,
+      "eval_loss": 0.810486376285553,
+      "eval_runtime": 1.3289,
+      "eval_samples_per_second": 0.753,
+      "eval_steps_per_second": 0.753,
+      "step": 15400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5721091381550453e-05,
+      "loss": 0.8162,
+      "step": 15420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5349873167110063e-05,
+      "loss": 0.8229,
+      "step": 15440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4978654952669677e-05,
+      "loss": 0.8272,
+      "step": 15460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4607436738229287e-05,
+      "loss": 0.8319,
+      "step": 15480
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4236218523788898e-05,
+      "loss": 0.8207,
+      "step": 15500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3865000309348511e-05,
+      "loss": 0.8273,
+      "step": 15520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3493782094908123e-05,
+      "loss": 0.825,
+      "step": 15540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3122563880467734e-05,
+      "loss": 0.825,
+      "step": 15560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2751345666027345e-05,
+      "loss": 0.8304,
+      "step": 15580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2380127451586957e-05,
+      "loss": 0.8246,
+      "step": 15600
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 0.8182739615440369,
+      "eval_runtime": 1.4484,
+      "eval_samples_per_second": 0.69,
+      "eval_steps_per_second": 0.69,
+      "step": 15600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.200890923714657e-05,
+      "loss": 0.8231,
+      "step": 15620
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.163769102270618e-05,
+      "loss": 0.8232,
+      "step": 15640
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.126647280826579e-05,
+      "loss": 0.827,
+      "step": 15660
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0895254593825404e-05,
+      "loss": 0.821,
+      "step": 15680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0524036379385013e-05,
+      "loss": 0.8294,
+      "step": 15700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0152818164944625e-05,
+      "loss": 0.8324,
+      "step": 15720
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.781599950504238e-06,
+      "loss": 0.826,
+      "step": 15740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.410381736063849e-06,
+      "loss": 0.8277,
+      "step": 15760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.03916352162346e-06,
+      "loss": 0.8286,
+      "step": 15780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.667945307183072e-06,
+      "loss": 0.8229,
+      "step": 15800
+    },
+    {
+      "epoch": 2.91,
+      "eval_loss": 0.8074170351028442,
+      "eval_runtime": 1.3094,
+      "eval_samples_per_second": 0.764,
+      "eval_steps_per_second": 0.764,
+      "step": 15800
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.296727092742683e-06,
+      "loss": 0.8249,
+      "step": 15820
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.925508878302294e-06,
+      "loss": 0.824,
+      "step": 15840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.554290663861906e-06,
+      "loss": 0.8244,
+      "step": 15860
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.183072449421518e-06,
+      "loss": 0.8257,
+      "step": 15880
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.81185423498113e-06,
+      "loss": 0.8176,
+      "step": 15900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.44063602054074e-06,
+      "loss": 0.8287,
+      "step": 15920
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.0694178061003525e-06,
+      "loss": 0.8272,
+      "step": 15940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.698199591659964e-06,
+      "loss": 0.8193,
+      "step": 15960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.326981377219574e-06,
+      "loss": 0.8067,
+      "step": 15980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.9557631627791865e-06,
+      "loss": 0.8213,
+      "step": 16000
+    },
+    {
+      "epoch": 2.95,
+      "eval_loss": 0.816523551940918,
+      "eval_runtime": 1.3844,
+      "eval_samples_per_second": 0.722,
+      "eval_steps_per_second": 0.722,
+      "step": 16000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.584544948338798e-06,
+      "loss": 0.8285,
+      "step": 16020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.213326733898409e-06,
+      "loss": 0.823,
+      "step": 16040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.842108519458021e-06,
+      "loss": 0.8245,
+      "step": 16060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4708903050176327e-06,
+      "loss": 0.8222,
+      "step": 16080
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0996720905772444e-06,
+      "loss": 0.8244,
+      "step": 16100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.7284538761368553e-06,
+      "loss": 0.8269,
+      "step": 16120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.357235661696467e-06,
+      "loss": 0.831,
+      "step": 16140
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9860174472560784e-06,
+      "loss": 0.8272,
+      "step": 16160
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6147992328156901e-06,
+      "loss": 0.8302,
+      "step": 16180
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2435810183753015e-06,
+      "loss": 0.8074,
+      "step": 16200
+    },
+    {
+      "epoch": 2.99,
+      "eval_loss": 0.8098793625831604,
+      "eval_runtime": 1.5007,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.666,
+      "step": 16200
+    }
+  ],
+  "max_steps": 16263,
+  "num_train_epochs": 3,
+  "total_flos": 1.0744586862539571e+19,
+  "trial_name": null,
+  "trial_params": null
+}