diff --git "a/check-22000/trainer_state.json" "b/check-22000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/check-22000/trainer_state.json"
@@ -0,0 +1,13612 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9986835534976621,
+  "global_step": 22000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999092105860457e-05,
+      "loss": 2.4805,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9981842117209137e-05,
+      "loss": 2.4355,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99727631758137e-05,
+      "loss": 1.317,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.996368423441827e-05,
+      "loss": 1.5764,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9954605293022836e-05,
+      "loss": 2.4654,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9945526351627404e-05,
+      "loss": 3.0094,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.993644741023197e-05,
+      "loss": 1.9016,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9927368468836535e-05,
+      "loss": 2.7273,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9918289527441103e-05,
+      "loss": 2.0789,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9909210586045667e-05,
+      "loss": 1.8132,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9900131644650234e-05,
+      "loss": 2.7362,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9891052703254802e-05,
+      "loss": 1.1584,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9881973761859366e-05,
+      "loss": 1.8474,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9872894820463934e-05,
+      "loss": 1.2248,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98638158790685e-05,
+      "loss": 1.8344,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.985473693767307e-05,
+      "loss": 1.5881,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9845657996277636e-05,
+      "loss": 1.9684,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98365790548822e-05,
+      "loss": 1.9283,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9827500113486768e-05,
+      "loss": 1.776,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9818421172091336e-05,
+      "loss": 1.5498,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9809342230695903e-05,
+      "loss": 2.0404,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.980026328930047e-05,
+      "loss": 2.0609,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9791184347905038e-05,
+      "loss": 1.2291,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9782105406509602e-05,
+      "loss": 0.6943,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.977302646511417e-05,
+      "loss": 1.5242,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9763947523718737e-05,
+      "loss": 1.5855,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9754868582323305e-05,
+      "loss": 1.6128,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.974578964092787e-05,
+      "loss": 1.4307,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9736710699532437e-05,
+      "loss": 1.402,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9727631758137e-05,
+      "loss": 1.4633,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971855281674157e-05,
+      "loss": 1.0023,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9709473875346136e-05,
+      "loss": 1.2508,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9700394933950703e-05,
+      "loss": 1.4191,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9691315992555268e-05,
+      "loss": 1.0203,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9682237051159835e-05,
+      "loss": 1.2482,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9673158109764403e-05,
+      "loss": 1.108,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.966407916836897e-05,
+      "loss": 0.8581,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9655000226973538e-05,
+      "loss": 1.307,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9645921285578102e-05,
+      "loss": 1.0133,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.963684234418267e-05,
+      "loss": 0.9578,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9627763402787237e-05,
+      "loss": 1.2262,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9618684461391805e-05,
+      "loss": 1.075,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9609605519996372e-05,
+      "loss": 1.1379,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9600526578600936e-05,
+      "loss": 1.1068,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9591447637205504e-05,
+      "loss": 0.9898,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.958236869581007e-05,
+      "loss": 0.7063,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9573289754414636e-05,
+      "loss": 1.0111,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9564210813019203e-05,
+      "loss": 0.9605,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.955513187162377e-05,
+      "loss": 0.8791,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9546052930228335e-05,
+      "loss": 0.8611,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "eval_accuracy": 0.5282472235634958,
+      "eval_loss": 0.9476923942565918,
+      "eval_runtime": 73.7994,
+      "eval_samples_per_second": 56.125,
+      "eval_steps_per_second": 14.038,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9536973988832902e-05,
+      "loss": 0.7687,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.952789504743747e-05,
+      "loss": 1.0031,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9518816106042037e-05,
+      "loss": 1.0047,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.95097371646466e-05,
+      "loss": 1.198,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.950065822325117e-05,
+      "loss": 1.1266,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9491579281855737e-05,
+      "loss": 0.9148,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9482500340460304e-05,
+      "loss": 0.9363,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9473421399064872e-05,
+      "loss": 1.1197,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.946434245766944e-05,
+      "loss": 0.8326,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9455263516274003e-05,
+      "loss": 1.0715,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.944618457487857e-05,
+      "loss": 0.8307,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.943710563348314e-05,
+      "loss": 0.6955,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9428026692087706e-05,
+      "loss": 0.9908,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9418947750692274e-05,
+      "loss": 0.7029,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9409868809296838e-05,
+      "loss": 0.8758,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9400789867901405e-05,
+      "loss": 0.858,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.939171092650597e-05,
+      "loss": 0.6072,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9382631985110537e-05,
+      "loss": 0.7605,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9373553043715105e-05,
+      "loss": 0.6867,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.936447410231967e-05,
+      "loss": 0.8187,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9355395160924236e-05,
+      "loss": 0.7312,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9346316219528804e-05,
+      "loss": 0.7791,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.933723727813337e-05,
+      "loss": 0.8156,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.932815833673794e-05,
+      "loss": 0.9059,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9319079395342503e-05,
+      "loss": 0.7408,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.931000045394707e-05,
+      "loss": 0.8867,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9300921512551638e-05,
+      "loss": 0.6566,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9291842571156206e-05,
+      "loss": 0.7961,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9282763629760773e-05,
+      "loss": 0.7789,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9273684688365337e-05,
+      "loss": 0.9086,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9264605746969905e-05,
+      "loss": 0.7547,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9255526805574472e-05,
+      "loss": 0.7371,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.924644786417904e-05,
+      "loss": 0.6797,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9237368922783604e-05,
+      "loss": 0.6729,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.922828998138817e-05,
+      "loss": 0.6578,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.921921103999274e-05,
+      "loss": 0.7986,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9210132098597303e-05,
+      "loss": 0.691,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.920105315720187e-05,
+      "loss": 0.7988,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.919197421580644e-05,
+      "loss": 0.7607,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9182895274411003e-05,
+      "loss": 0.6561,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.917381633301557e-05,
+      "loss": 0.71,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9164737391620138e-05,
+      "loss": 0.7926,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9155658450224705e-05,
+      "loss": 0.8121,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9146579508829273e-05,
+      "loss": 0.6301,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.913750056743384e-05,
+      "loss": 0.6631,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9128421626038404e-05,
+      "loss": 0.774,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9119342684642972e-05,
+      "loss": 0.6559,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.911026374324754e-05,
+      "loss": 0.7158,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9101184801852107e-05,
+      "loss": 0.8125,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9092105860456675e-05,
+      "loss": 0.8473,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "eval_accuracy": 0.5444229840656688,
+      "eval_loss": 0.7574031949043274,
+      "eval_runtime": 73.6682,
+      "eval_samples_per_second": 56.225,
+      "eval_steps_per_second": 14.063,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.908302691906124e-05,
+      "loss": 0.6639,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9073947977665806e-05,
+      "loss": 0.7871,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9064869036270374e-05,
+      "loss": 0.651,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9055790094874938e-05,
+      "loss": 0.7703,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9046711153479506e-05,
+      "loss": 0.7203,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9037632212084073e-05,
+      "loss": 0.7766,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9028553270688637e-05,
+      "loss": 0.6488,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9019474329293205e-05,
+      "loss": 0.9973,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9010395387897772e-05,
+      "loss": 0.7613,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.900131644650234e-05,
+      "loss": 0.6891,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8992237505106904e-05,
+      "loss": 0.6699,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.898315856371147e-05,
+      "loss": 0.6811,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.897407962231604e-05,
+      "loss": 0.8492,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8965000680920607e-05,
+      "loss": 0.7957,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8955921739525174e-05,
+      "loss": 0.759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.894684279812974e-05,
+      "loss": 0.8492,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8937763856734306e-05,
+      "loss": 0.7164,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8928684915338873e-05,
+      "loss": 0.6832,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.891960597394344e-05,
+      "loss": 0.793,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.891052703254801e-05,
+      "loss": 0.725,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8901448091152573e-05,
+      "loss": 0.6547,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.889236914975714e-05,
+      "loss": 0.6254,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8883290208361708e-05,
+      "loss": 0.7918,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8874211266966272e-05,
+      "loss": 0.8402,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.886513232557084e-05,
+      "loss": 0.6934,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8856053384175407e-05,
+      "loss": 0.691,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.884697444277997e-05,
+      "loss": 0.6693,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.883789550138454e-05,
+      "loss": 0.827,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8828816559989106e-05,
+      "loss": 0.6457,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8819737618593674e-05,
+      "loss": 0.6223,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8810658677198238e-05,
+      "loss": 0.7271,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8801579735802806e-05,
+      "loss": 0.85,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8792500794407373e-05,
+      "loss": 0.7082,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.878342185301194e-05,
+      "loss": 0.7145,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8774342911616508e-05,
+      "loss": 0.8406,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8765263970221076e-05,
+      "loss": 0.6855,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.875618502882564e-05,
+      "loss": 0.6941,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8747106087430207e-05,
+      "loss": 0.6742,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8738027146034775e-05,
+      "loss": 0.6928,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8728948204639342e-05,
+      "loss": 0.7762,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8719869263243907e-05,
+      "loss": 0.8305,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8710790321848474e-05,
+      "loss": 0.6861,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8701711380453042e-05,
+      "loss": 0.7648,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8692632439057606e-05,
+      "loss": 0.7822,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8683553497662173e-05,
+      "loss": 0.7717,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.867447455626674e-05,
+      "loss": 0.6504,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8665395614871305e-05,
+      "loss": 0.7363,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8656316673475873e-05,
+      "loss": 0.6836,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.864723773208044e-05,
+      "loss": 0.8811,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8638158790685008e-05,
+      "loss": 0.7086,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "eval_accuracy": 0.5596330275229358,
+      "eval_loss": 0.7176767587661743,
+      "eval_runtime": 74.531,
+      "eval_samples_per_second": 55.574,
+      "eval_steps_per_second": 13.9,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8629079849289575e-05,
+      "loss": 0.6834,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.862000090789414e-05,
+      "loss": 0.7777,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8610921966498707e-05,
+      "loss": 0.6586,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8601843025103275e-05,
+      "loss": 0.7246,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8592764083707842e-05,
+      "loss": 0.7203,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.858368514231241e-05,
+      "loss": 0.7113,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8574606200916974e-05,
+      "loss": 0.7828,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.856552725952154e-05,
+      "loss": 0.7518,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.855644831812611e-05,
+      "loss": 0.693,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8547369376730676e-05,
+      "loss": 0.7,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.853829043533524e-05,
+      "loss": 0.65,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8529211493939808e-05,
+      "loss": 0.6734,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8520132552544376e-05,
+      "loss": 0.7648,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.851105361114894e-05,
+      "loss": 0.6742,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8501974669753507e-05,
+      "loss": 0.7734,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8492895728358075e-05,
+      "loss": 0.7621,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.848381678696264e-05,
+      "loss": 0.6535,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8474737845567207e-05,
+      "loss": 0.7879,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8465658904171774e-05,
+      "loss": 0.6699,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.845657996277634e-05,
+      "loss": 0.6672,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.844750102138091e-05,
+      "loss": 0.759,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8438422079985477e-05,
+      "loss": 0.7328,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842934313859004e-05,
+      "loss": 0.7902,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842026419719461e-05,
+      "loss": 0.7342,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8411185255799176e-05,
+      "loss": 0.7047,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8402106314403744e-05,
+      "loss": 0.7137,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.839302737300831e-05,
+      "loss": 0.6871,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8383948431612875e-05,
+      "loss": 0.6461,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8374869490217443e-05,
+      "loss": 0.7117,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.836579054882201e-05,
+      "loss": 0.7719,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8356711607426574e-05,
+      "loss": 0.65,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8347632666031142e-05,
+      "loss": 0.7473,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.833855372463571e-05,
+      "loss": 0.7457,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8329474783240274e-05,
+      "loss": 0.7027,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.832039584184484e-05,
+      "loss": 0.7066,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.831131690044941e-05,
+      "loss": 0.7414,
+      "step": 1860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8302237959053976e-05,
+      "loss": 0.7328,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.829315901765854e-05,
+      "loss": 0.7215,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8284080076263108e-05,
+      "loss": 0.623,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8275001134867676e-05,
+      "loss": 0.7512,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8265922193472243e-05,
+      "loss": 0.7398,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.825684325207681e-05,
+      "loss": 0.7441,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8247764310681375e-05,
+      "loss": 0.6504,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8238685369285942e-05,
+      "loss": 0.7625,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.822960642789051e-05,
+      "loss": 0.668,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8220527486495077e-05,
+      "loss": 0.6961,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8211448545099645e-05,
+      "loss": 0.7477,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.820236960370421e-05,
+      "loss": 0.7844,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8193290662308777e-05,
+      "loss": 0.6893,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8184211720913344e-05,
+      "loss": 0.7223,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "eval_accuracy": 0.5642201834862385,
+      "eval_loss": 0.7053582072257996,
+      "eval_runtime": 73.9274,
+      "eval_samples_per_second": 56.028,
+      "eval_steps_per_second": 14.014,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.817513277951791e-05,
+      "loss": 0.6996,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8166053838122476e-05,
+      "loss": 0.6939,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8156974896727043e-05,
+      "loss": 0.7418,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8147895955331608e-05,
+      "loss": 0.7295,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8138817013936175e-05,
+      "loss": 0.6215,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8129738072540743e-05,
+      "loss": 0.6334,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.812065913114531e-05,
+      "loss": 0.7109,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8111580189749878e-05,
+      "loss": 0.6764,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8102501248354442e-05,
+      "loss": 0.6371,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.809342230695901e-05,
+      "loss": 0.6674,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8084343365563577e-05,
+      "loss": 0.6689,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8075264424168145e-05,
+      "loss": 0.6523,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8066185482772712e-05,
+      "loss": 0.6008,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8057106541377276e-05,
+      "loss": 0.742,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8048027599981844e-05,
+      "loss": 0.7527,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.803894865858641e-05,
+      "loss": 0.7352,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.802986971719098e-05,
+      "loss": 0.6502,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8020790775795543e-05,
+      "loss": 0.7156,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.801171183440011e-05,
+      "loss": 0.7008,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8002632893004678e-05,
+      "loss": 0.6953,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7993553951609242e-05,
+      "loss": 0.7641,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.798447501021381e-05,
+      "loss": 0.7266,
+      "step": 2220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7975396068818377e-05,
+      "loss": 0.657,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.796631712742294e-05,
+      "loss": 0.6883,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.795723818602751e-05,
+      "loss": 0.699,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7948159244632077e-05,
+      "loss": 0.6229,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7939080303236644e-05,
+      "loss": 0.6969,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7930001361841212e-05,
+      "loss": 0.7721,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7920922420445776e-05,
+      "loss": 0.7125,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7911843479050343e-05,
+      "loss": 0.7691,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.790276453765491e-05,
+      "loss": 0.7262,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.789368559625948e-05,
+      "loss": 0.7363,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7884606654864046e-05,
+      "loss": 0.6568,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.787552771346861e-05,
+      "loss": 0.6848,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7866448772073178e-05,
+      "loss": 0.623,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7857369830677745e-05,
+      "loss": 0.7602,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7848290889282313e-05,
+      "loss": 0.5975,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7839211947886877e-05,
+      "loss": 0.6773,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7830133006491445e-05,
+      "loss": 0.7605,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7821054065096012e-05,
+      "loss": 0.7703,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7811975123700576e-05,
+      "loss": 0.7121,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7802896182305144e-05,
+      "loss": 0.7508,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.779381724090971e-05,
+      "loss": 0.6875,
+      "step": 2430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7784738299514275e-05,
+      "loss": 0.7035,
+      "step": 2440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7775659358118843e-05,
+      "loss": 0.7422,
+      "step": 2450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.776658041672341e-05,
+      "loss": 0.6859,
+      "step": 2460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7757501475327978e-05,
+      "loss": 0.7377,
+      "step": 2470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7748422533932546e-05,
+      "loss": 0.7625,
+      "step": 2480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7739343592537113e-05,
+      "loss": 0.6867,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7730264651141677e-05,
+      "loss": 0.7092,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "eval_accuracy": 0.5864316755190729,
+      "eval_loss": 0.69435054063797,
+      "eval_runtime": 74.0387,
+      "eval_samples_per_second": 55.944,
+      "eval_steps_per_second": 13.993,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7721185709746245e-05,
+      "loss": 0.7135,
+      "step": 2510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7712106768350812e-05,
+      "loss": 0.6902,
+      "step": 2520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.770302782695538e-05,
+      "loss": 0.6463,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7693948885559948e-05,
+      "loss": 0.7086,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.768486994416451e-05,
+      "loss": 0.743,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.767579100276908e-05,
+      "loss": 0.727,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7666712061373647e-05,
+      "loss": 0.7523,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.765763311997821e-05,
+      "loss": 0.6906,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.764855417858278e-05,
+      "loss": 0.7262,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7639475237187346e-05,
+      "loss": 0.6633,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.763039629579191e-05,
+      "loss": 0.6121,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7621317354396478e-05,
+      "loss": 0.6309,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7612238413001045e-05,
+      "loss": 0.6721,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7603159471605613e-05,
+      "loss": 0.6562,
+      "step": 2640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7594080530210177e-05,
+      "loss": 0.6633,
+      "step": 2650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7585001588814745e-05,
+      "loss": 0.7188,
+      "step": 2660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7575922647419312e-05,
+      "loss": 0.7297,
+      "step": 2670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.756684370602388e-05,
+      "loss": 0.6438,
+      "step": 2680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7557764764628447e-05,
+      "loss": 0.6961,
+      "step": 2690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.754868582323301e-05,
+      "loss": 0.6992,
+      "step": 2700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.753960688183758e-05,
+      "loss": 0.5748,
+      "step": 2710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7530527940442146e-05,
+      "loss": 0.5975,
+      "step": 2720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7521448999046714e-05,
+      "loss": 0.7008,
+      "step": 2730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.751237005765128e-05,
+      "loss": 0.8043,
+      "step": 2740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7503291116255846e-05,
+      "loss": 0.7314,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7494212174860413e-05,
+      "loss": 0.6992,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.748513323346498e-05,
+      "loss": 0.6832,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7476054292069545e-05,
+      "loss": 0.7281,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7466975350674112e-05,
+      "loss": 0.7684,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.745789640927868e-05,
+      "loss": 0.693,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7448817467883244e-05,
+      "loss": 0.7365,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.743973852648781e-05,
+      "loss": 0.6813,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.743065958509238e-05,
+      "loss": 0.6824,
+      "step": 2830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7421580643696947e-05,
+      "loss": 0.7217,
+      "step": 2840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7412501702301514e-05,
+      "loss": 0.7031,
+      "step": 2850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.740342276090608e-05,
+      "loss": 0.7348,
+      "step": 2860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7394343819510646e-05,
+      "loss": 0.6928,
+      "step": 2870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7385264878115214e-05,
+      "loss": 0.7035,
+      "step": 2880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.737618593671978e-05,
+      "loss": 0.6838,
+      "step": 2890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.736710699532435e-05,
+      "loss": 0.6586,
+      "step": 2900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7358028053928913e-05,
+      "loss": 0.6711,
+      "step": 2910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.734894911253348e-05,
+      "loss": 0.693,
+      "step": 2920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7339870171138048e-05,
+      "loss": 0.701,
+      "step": 2930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7330791229742615e-05,
+      "loss": 0.598,
+      "step": 2940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.732171228834718e-05,
+      "loss": 0.673,
+      "step": 2950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7312633346951747e-05,
+      "loss": 0.6502,
+      "step": 2960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7303554405556315e-05,
+      "loss": 0.7301,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.729447546416088e-05,
+      "loss": 0.6961,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7285396522765446e-05,
+      "loss": 0.7199,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7276317581370014e-05,
+      "loss": 0.6605,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "eval_accuracy": 0.5806373732496378,
+      "eval_loss": 0.7019612193107605,
+      "eval_runtime": 75.2508,
+      "eval_samples_per_second": 55.043,
+      "eval_steps_per_second": 13.767,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7267238639974578e-05,
+      "loss": 0.6375,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7258159698579146e-05,
+      "loss": 0.7408,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7249080757183713e-05,
+      "loss": 0.6504,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.724000181578828e-05,
+      "loss": 0.7404,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7230922874392848e-05,
+      "loss": 0.7684,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7221843932997412e-05,
+      "loss": 0.6043,
+      "step": 3060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.721276499160198e-05,
+      "loss": 0.6562,
+      "step": 3070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7203686050206547e-05,
+      "loss": 0.7309,
+      "step": 3080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7194607108811115e-05,
+      "loss": 0.6188,
+      "step": 3090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7185528167415683e-05,
+      "loss": 0.767,
+      "step": 3100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7176449226020247e-05,
+      "loss": 0.7379,
+      "step": 3110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7167370284624814e-05,
+      "loss": 0.732,
+      "step": 3120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7158291343229382e-05,
+      "loss": 0.7145,
+      "step": 3130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.714921240183395e-05,
+      "loss": 0.7035,
+      "step": 3140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7140133460438513e-05,
+      "loss": 0.759,
+      "step": 3150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.713105451904308e-05,
+      "loss": 0.7232,
+      "step": 3160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.712197557764765e-05,
+      "loss": 0.6621,
+      "step": 3170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7112896636252213e-05,
+      "loss": 0.6104,
+      "step": 3180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.710381769485678e-05,
+      "loss": 0.7688,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7094738753461348e-05,
+      "loss": 0.6383,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7085659812065915e-05,
+      "loss": 0.6279,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.707658087067048e-05,
+      "loss": 0.8711,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7067501929275047e-05,
+      "loss": 0.7449,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7058422987879615e-05,
+      "loss": 0.6668,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7049344046484182e-05,
+      "loss": 0.6992,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.704026510508875e-05,
+      "loss": 0.6529,
+      "step": 3260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7031186163693314e-05,
+      "loss": 0.6301,
+      "step": 3270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.702210722229788e-05,
+      "loss": 0.7,
+      "step": 3280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.701302828090245e-05,
+      "loss": 0.7004,
+      "step": 3290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7003949339507016e-05,
+      "loss": 0.6744,
+      "step": 3300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6994870398111584e-05,
+      "loss": 0.6875,
+      "step": 3310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6985791456716148e-05,
+      "loss": 0.7172,
+      "step": 3320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6976712515320716e-05,
+      "loss": 0.7213,
+      "step": 3330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6967633573925283e-05,
+      "loss": 0.6945,
+      "step": 3340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6958554632529847e-05,
+      "loss": 0.6578,
+      "step": 3350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6949475691134415e-05,
+      "loss": 0.7834,
+      "step": 3360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6940396749738982e-05,
+      "loss": 0.6871,
+      "step": 3370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6931317808343547e-05,
+      "loss": 0.6965,
+      "step": 3380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6922238866948114e-05,
+      "loss": 0.6832,
+      "step": 3390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6913159925552682e-05,
+      "loss": 0.6949,
+      "step": 3400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.690408098415725e-05,
+      "loss": 0.7303,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6895002042761813e-05,
+      "loss": 0.7215,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.688592310136638e-05,
+      "loss": 0.6664,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.687684415997095e-05,
+      "loss": 0.7107,
+      "step": 3440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6867765218575516e-05,
+      "loss": 0.7289,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6858686277180084e-05,
+      "loss": 0.6801,
+      "step": 3460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6849607335784648e-05,
+      "loss": 0.6475,
+      "step": 3470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6840528394389215e-05,
+      "loss": 0.6285,
+      "step": 3480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6831449452993783e-05,
+      "loss": 0.6699,
+      "step": 3490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.682237051159835e-05,
+      "loss": 0.7424,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "eval_accuracy": 0.5782230806373733,
+      "eval_loss": 0.6958387494087219,
+      "eval_runtime": 74.8034,
+      "eval_samples_per_second": 55.372,
+      "eval_steps_per_second": 13.85,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6813291570202918e-05,
+      "loss": 0.676,
+      "step": 3510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6804212628807482e-05,
+      "loss": 0.5826,
+      "step": 3520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.679513368741205e-05,
+      "loss": 0.7164,
+      "step": 3530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6786054746016617e-05,
+      "loss": 0.8461,
+      "step": 3540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.677697580462118e-05,
+      "loss": 0.7219,
+      "step": 3550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.676789686322575e-05,
+      "loss": 0.7793,
+      "step": 3560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6758817921830316e-05,
+      "loss": 0.6773,
+      "step": 3570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.674973898043488e-05,
+      "loss": 0.6342,
+      "step": 3580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6740660039039448e-05,
+      "loss": 0.6426,
+      "step": 3590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6731581097644016e-05,
+      "loss": 0.8352,
+      "step": 3600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6722502156248583e-05,
+      "loss": 0.673,
+      "step": 3610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.671342321485315e-05,
+      "loss": 0.6305,
+      "step": 3620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6704344273457715e-05,
+      "loss": 0.6246,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6695265332062282e-05,
+      "loss": 0.6191,
+      "step": 3640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.668618639066685e-05,
+      "loss": 0.6887,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6677107449271418e-05,
+      "loss": 0.6418,
+      "step": 3660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6668028507875985e-05,
+      "loss": 0.727,
+      "step": 3670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.665894956648055e-05,
+      "loss": 0.6848,
+      "step": 3680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6649870625085117e-05,
+      "loss": 0.7105,
+      "step": 3690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6640791683689684e-05,
+      "loss": 0.7939,
+      "step": 3700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6631712742294252e-05,
+      "loss": 0.6701,
+      "step": 3710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6622633800898816e-05,
+      "loss": 0.7344,
+      "step": 3720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6613554859503384e-05,
+      "loss": 0.6816,
+      "step": 3730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.660447591810795e-05,
+      "loss": 0.6516,
+      "step": 3740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6595396976712515e-05,
+      "loss": 0.6641,
+      "step": 3750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6586318035317083e-05,
+      "loss": 0.6396,
+      "step": 3760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.657723909392165e-05,
+      "loss": 0.6596,
+      "step": 3770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6568160152526214e-05,
+      "loss": 0.6672,
+      "step": 3780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6559081211130782e-05,
+      "loss": 0.7094,
+      "step": 3790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.655000226973535e-05,
+      "loss": 0.7512,
+      "step": 3800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6540923328339917e-05,
+      "loss": 0.7082,
+      "step": 3810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6531844386944485e-05,
+      "loss": 0.6738,
+      "step": 3820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.652276544554905e-05,
+      "loss": 0.659,
+      "step": 3830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6513686504153616e-05,
+      "loss": 0.725,
+      "step": 3840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6504607562758184e-05,
+      "loss": 0.6949,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.649552862136275e-05,
+      "loss": 0.7621,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.648644967996732e-05,
+      "loss": 0.5955,
+      "step": 3870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6477370738571883e-05,
+      "loss": 0.7156,
+      "step": 3880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.646829179717645e-05,
+      "loss": 0.7133,
+      "step": 3890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6459212855781018e-05,
+      "loss": 0.5943,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6450133914385586e-05,
+      "loss": 0.7309,
+      "step": 3910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.644105497299015e-05,
+      "loss": 0.6729,
+      "step": 3920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6431976031594717e-05,
+      "loss": 0.6658,
+      "step": 3930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6422897090199285e-05,
+      "loss": 0.592,
+      "step": 3940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.641381814880385e-05,
+      "loss": 0.6889,
+      "step": 3950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6404739207408417e-05,
+      "loss": 0.6473,
+      "step": 3960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6395660266012984e-05,
+      "loss": 0.6818,
+      "step": 3970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6386581324617552e-05,
+      "loss": 0.6381,
+      "step": 3980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6377502383222116e-05,
+      "loss": 0.7305,
+      "step": 3990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6368423441826683e-05,
+      "loss": 0.6549,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "eval_accuracy": 0.5782230806373733,
+      "eval_loss": 0.6909931898117065,
+      "eval_runtime": 72.1497,
+      "eval_samples_per_second": 57.408,
+      "eval_steps_per_second": 14.359,
+      "step": 4000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.635934450043125e-05,
+      "loss": 0.715,
+      "step": 4010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.635026555903582e-05,
+      "loss": 0.668,
+      "step": 4020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6341186617640386e-05,
+      "loss": 0.6637,
+      "step": 4030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.633210767624495e-05,
+      "loss": 0.6697,
+      "step": 4040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6323028734849518e-05,
+      "loss": 0.5869,
+      "step": 4050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6313949793454085e-05,
+      "loss": 0.6004,
+      "step": 4060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6304870852058653e-05,
+      "loss": 0.775,
+      "step": 4070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.629579191066322e-05,
+      "loss": 0.6416,
+      "step": 4080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6286712969267785e-05,
+      "loss": 0.641,
+      "step": 4090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6277634027872352e-05,
+      "loss": 0.7406,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.626855508647692e-05,
+      "loss": 0.6365,
+      "step": 4110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6259476145081484e-05,
+      "loss": 0.6926,
+      "step": 4120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.625039720368605e-05,
+      "loss": 0.6426,
+      "step": 4130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.624131826229062e-05,
+      "loss": 0.7107,
+      "step": 4140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6232239320895183e-05,
+      "loss": 0.65,
+      "step": 4150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.622316037949975e-05,
+      "loss": 0.7195,
+      "step": 4160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6214081438104318e-05,
+      "loss": 0.5832,
+      "step": 4170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6205002496708886e-05,
+      "loss": 0.7441,
+      "step": 4180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.619592355531345e-05,
+      "loss": 0.627,
+      "step": 4190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6186844613918017e-05,
+      "loss": 0.6598,
+      "step": 4200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6177765672522585e-05,
+      "loss": 0.6814,
+      "step": 4210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6168686731127152e-05,
+      "loss": 0.7156,
+      "step": 4220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.615960778973172e-05,
+      "loss": 0.6285,
+      "step": 4230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6150528848336284e-05,
+      "loss": 0.7143,
+      "step": 4240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6141449906940852e-05,
+      "loss": 0.6299,
+      "step": 4250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.613237096554542e-05,
+      "loss": 0.6852,
+      "step": 4260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6123292024149987e-05,
+      "loss": 0.7059,
+      "step": 4270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6114213082754554e-05,
+      "loss": 0.6936,
+      "step": 4280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.610513414135912e-05,
+      "loss": 0.5746,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6096055199963686e-05,
+      "loss": 0.7225,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6086976258568254e-05,
+      "loss": 0.6955,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6077897317172818e-05,
+      "loss": 0.7551,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6068818375777385e-05,
+      "loss": 0.6752,
+      "step": 4330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6059739434381953e-05,
+      "loss": 0.7883,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6050660492986517e-05,
+      "loss": 0.7078,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6041581551591085e-05,
+      "loss": 0.8699,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6032502610195652e-05,
+      "loss": 0.6383,
+      "step": 4370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.602342366880022e-05,
+      "loss": 0.6402,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6014344727404787e-05,
+      "loss": 0.6916,
+      "step": 4390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.600526578600935e-05,
+      "loss": 0.6859,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.599618684461392e-05,
+      "loss": 0.6668,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5987107903218486e-05,
+      "loss": 0.7312,
+      "step": 4420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5978028961823054e-05,
+      "loss": 0.6523,
+      "step": 4430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.596895002042762e-05,
+      "loss": 0.6686,
+      "step": 4440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5959871079032186e-05,
+      "loss": 0.7184,
+      "step": 4450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5950792137636753e-05,
+      "loss": 0.65,
+      "step": 4460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.594171319624132e-05,
+      "loss": 0.7178,
+      "step": 4470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5932634254845888e-05,
+      "loss": 0.6918,
+      "step": 4480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5923555313450452e-05,
+      "loss": 0.5949,
+      "step": 4490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.591447637205502e-05,
+      "loss": 0.6268,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2,
+      "eval_accuracy": 0.5929502655721873,
+      "eval_loss": 0.6875754594802856,
+      "eval_runtime": 74.4346,
+      "eval_samples_per_second": 55.646,
+      "eval_steps_per_second": 13.918,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5905397430659588e-05,
+      "loss": 0.6125,
+      "step": 4510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.589631848926415e-05,
+      "loss": 0.8043,
+      "step": 4520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.588723954786872e-05,
+      "loss": 0.6482,
+      "step": 4530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5878160606473287e-05,
+      "loss": 0.7039,
+      "step": 4540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.586908166507785e-05,
+      "loss": 0.6203,
+      "step": 4550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.586000272368242e-05,
+      "loss": 0.6557,
+      "step": 4560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5850923782286986e-05,
+      "loss": 0.6105,
+      "step": 4570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5841844840891554e-05,
+      "loss": 0.8234,
+      "step": 4580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.583276589949612e-05,
+      "loss": 0.6199,
+      "step": 4590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5823686958100685e-05,
+      "loss": 0.675,
+      "step": 4600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5814608016705253e-05,
+      "loss": 0.6316,
+      "step": 4610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.580552907530982e-05,
+      "loss": 0.7779,
+      "step": 4620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5796450133914388e-05,
+      "loss": 0.5781,
+      "step": 4630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5787371192518955e-05,
+      "loss": 0.7367,
+      "step": 4640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.577829225112352e-05,
+      "loss": 0.7154,
+      "step": 4650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5769213309728087e-05,
+      "loss": 0.7148,
+      "step": 4660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5760134368332655e-05,
+      "loss": 0.6227,
+      "step": 4670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5751055426937222e-05,
+      "loss": 0.5682,
+      "step": 4680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5741976485541786e-05,
+      "loss": 0.6184,
+      "step": 4690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5732897544146354e-05,
+      "loss": 0.6588,
+      "step": 4700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.572381860275092e-05,
+      "loss": 0.6627,
+      "step": 4710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5714739661355486e-05,
+      "loss": 0.6479,
+      "step": 4720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5705660719960053e-05,
+      "loss": 0.7141,
+      "step": 4730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.569658177856462e-05,
+      "loss": 0.6545,
+      "step": 4740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5687502837169188e-05,
+      "loss": 0.6789,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5678423895773752e-05,
+      "loss": 0.6906,
+      "step": 4760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.566934495437832e-05,
+      "loss": 0.7527,
+      "step": 4770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5660266012982887e-05,
+      "loss": 0.599,
+      "step": 4780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5651187071587455e-05,
+      "loss": 0.7096,
+      "step": 4790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5642108130192023e-05,
+      "loss": 0.6578,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5633029188796587e-05,
+      "loss": 0.6953,
+      "step": 4810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5623950247401154e-05,
+      "loss": 0.7516,
+      "step": 4820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5614871306005722e-05,
+      "loss": 0.6023,
+      "step": 4830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.560579236461029e-05,
+      "loss": 0.6684,
+      "step": 4840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5596713423214857e-05,
+      "loss": 0.748,
+      "step": 4850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.558763448181942e-05,
+      "loss": 0.6287,
+      "step": 4860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.557855554042399e-05,
+      "loss": 0.7781,
+      "step": 4870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5569476599028556e-05,
+      "loss": 0.7213,
+      "step": 4880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.556039765763312e-05,
+      "loss": 0.673,
+      "step": 4890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5551318716237688e-05,
+      "loss": 0.8391,
+      "step": 4900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5542239774842255e-05,
+      "loss": 0.7537,
+      "step": 4910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.553316083344682e-05,
+      "loss": 0.6406,
+      "step": 4920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5524081892051387e-05,
+      "loss": 0.6082,
+      "step": 4930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5515002950655955e-05,
+      "loss": 0.7135,
+      "step": 4940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5505924009260522e-05,
+      "loss": 0.6555,
+      "step": 4950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5496845067865086e-05,
+      "loss": 0.6604,
+      "step": 4960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5487766126469654e-05,
+      "loss": 0.7117,
+      "step": 4970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.547868718507422e-05,
+      "loss": 0.7215,
+      "step": 4980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.546960824367879e-05,
+      "loss": 0.6865,
+      "step": 4990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5460529302283356e-05,
+      "loss": 0.7523,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "eval_accuracy": 0.59753742153549,
+      "eval_loss": 0.6818556189537048,
+      "eval_runtime": 74.991,
+      "eval_samples_per_second": 55.233,
+      "eval_steps_per_second": 13.815,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.545145036088792e-05,
+      "loss": 0.7436,
+      "step": 5010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5442371419492488e-05,
+      "loss": 0.7418,
+      "step": 5020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5433292478097056e-05,
+      "loss": 0.6656,
+      "step": 5030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5424213536701623e-05,
+      "loss": 0.7031,
+      "step": 5040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.541513459530619e-05,
+      "loss": 0.6496,
+      "step": 5050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5406055653910755e-05,
+      "loss": 0.7084,
+      "step": 5060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5396976712515323e-05,
+      "loss": 0.6781,
+      "step": 5070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.538789777111989e-05,
+      "loss": 0.7336,
+      "step": 5080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5378818829724454e-05,
+      "loss": 0.6762,
+      "step": 5090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5369739888329022e-05,
+      "loss": 0.6391,
+      "step": 5100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.536066094693359e-05,
+      "loss": 0.7719,
+      "step": 5110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5351582005538153e-05,
+      "loss": 0.6418,
+      "step": 5120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.534250306414272e-05,
+      "loss": 0.6543,
+      "step": 5130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.533342412274729e-05,
+      "loss": 0.6639,
+      "step": 5140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5324345181351856e-05,
+      "loss": 0.6256,
+      "step": 5150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5315266239956424e-05,
+      "loss": 0.6721,
+      "step": 5160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5306187298560988e-05,
+      "loss": 0.7191,
+      "step": 5170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5297108357165555e-05,
+      "loss": 0.7355,
+      "step": 5180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5288029415770123e-05,
+      "loss": 0.6605,
+      "step": 5190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.527895047437469e-05,
+      "loss": 0.6436,
+      "step": 5200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5269871532979258e-05,
+      "loss": 0.6848,
+      "step": 5210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5260792591583822e-05,
+      "loss": 0.7281,
+      "step": 5220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.525171365018839e-05,
+      "loss": 0.6137,
+      "step": 5230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5242634708792955e-05,
+      "loss": 0.7535,
+      "step": 5240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5233555767397523e-05,
+      "loss": 0.7076,
+      "step": 5250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.522447682600209e-05,
+      "loss": 0.6793,
+      "step": 5260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5215397884606655e-05,
+      "loss": 0.7406,
+      "step": 5270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5206318943211222e-05,
+      "loss": 0.7,
+      "step": 5280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.519724000181579e-05,
+      "loss": 0.6547,
+      "step": 5290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5188161060420357e-05,
+      "loss": 0.6844,
+      "step": 5300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5179082119024923e-05,
+      "loss": 0.6543,
+      "step": 5310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5170003177629489e-05,
+      "loss": 0.7021,
+      "step": 5320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5160924236234057e-05,
+      "loss": 0.6594,
+      "step": 5330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5151845294838622e-05,
+      "loss": 0.7711,
+      "step": 5340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.514276635344319e-05,
+      "loss": 0.6973,
+      "step": 5350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5133687412047758e-05,
+      "loss": 0.6928,
+      "step": 5360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5124608470652322e-05,
+      "loss": 0.7063,
+      "step": 5370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.511552952925689e-05,
+      "loss": 0.6875,
+      "step": 5380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5106450587861457e-05,
+      "loss": 0.65,
+      "step": 5390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5097371646466024e-05,
+      "loss": 0.6594,
+      "step": 5400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.508829270507059e-05,
+      "loss": 0.6527,
+      "step": 5410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5079213763675156e-05,
+      "loss": 0.6918,
+      "step": 5420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5070134822279724e-05,
+      "loss": 0.7523,
+      "step": 5430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.506105588088429e-05,
+      "loss": 0.6365,
+      "step": 5440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5051976939488857e-05,
+      "loss": 0.7805,
+      "step": 5450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5042897998093424e-05,
+      "loss": 0.6734,
+      "step": 5460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5033819056697992e-05,
+      "loss": 0.748,
+      "step": 5470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5024740115302556e-05,
+      "loss": 0.6609,
+      "step": 5480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5015661173907124e-05,
+      "loss": 0.6475,
+      "step": 5490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5006582232511691e-05,
+      "loss": 0.6582,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25,
+      "eval_accuracy": 0.6028488652824723,
+      "eval_loss": 0.6788424849510193,
+      "eval_runtime": 72.2831,
+      "eval_samples_per_second": 57.302,
+      "eval_steps_per_second": 14.333,
+      "step": 5500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4997503291116257e-05,
+      "loss": 0.6637,
+      "step": 5510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4988424349720825e-05,
+      "loss": 0.6816,
+      "step": 5520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.497934540832539e-05,
+      "loss": 0.6617,
+      "step": 5530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4970266466929956e-05,
+      "loss": 0.6871,
+      "step": 5540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4961187525534524e-05,
+      "loss": 0.6793,
+      "step": 5550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4952108584139091e-05,
+      "loss": 0.6922,
+      "step": 5560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4943029642743659e-05,
+      "loss": 0.5648,
+      "step": 5570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4933950701348223e-05,
+      "loss": 0.6439,
+      "step": 5580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.492487175995279e-05,
+      "loss": 0.7004,
+      "step": 5590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4915792818557358e-05,
+      "loss": 0.8055,
+      "step": 5600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4906713877161924e-05,
+      "loss": 0.5594,
+      "step": 5610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4897634935766492e-05,
+      "loss": 0.8199,
+      "step": 5620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4888555994371057e-05,
+      "loss": 0.6451,
+      "step": 5630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4879477052975623e-05,
+      "loss": 0.6434,
+      "step": 5640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4870398111580191e-05,
+      "loss": 0.6396,
+      "step": 5650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4861319170184758e-05,
+      "loss": 0.5916,
+      "step": 5660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4852240228789326e-05,
+      "loss": 0.7096,
+      "step": 5670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.484316128739389e-05,
+      "loss": 0.6758,
+      "step": 5680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4834082345998458e-05,
+      "loss": 0.6438,
+      "step": 5690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4825003404603025e-05,
+      "loss": 0.6777,
+      "step": 5700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4815924463207591e-05,
+      "loss": 0.7172,
+      "step": 5710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4806845521812159e-05,
+      "loss": 0.7588,
+      "step": 5720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4797766580416724e-05,
+      "loss": 0.651,
+      "step": 5730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.478868763902129e-05,
+      "loss": 0.6287,
+      "step": 5740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4779608697625858e-05,
+      "loss": 0.6875,
+      "step": 5750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4770529756230425e-05,
+      "loss": 0.6582,
+      "step": 5760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4761450814834993e-05,
+      "loss": 0.6656,
+      "step": 5770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4752371873439557e-05,
+      "loss": 0.6037,
+      "step": 5780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4743292932044125e-05,
+      "loss": 0.6973,
+      "step": 5790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4734213990648692e-05,
+      "loss": 0.6377,
+      "step": 5800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4725135049253258e-05,
+      "loss": 0.7617,
+      "step": 5810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4716056107857826e-05,
+      "loss": 0.6621,
+      "step": 5820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4706977166462391e-05,
+      "loss": 0.6861,
+      "step": 5830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4697898225066957e-05,
+      "loss": 0.6504,
+      "step": 5840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4688819283671525e-05,
+      "loss": 0.7766,
+      "step": 5850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4679740342276092e-05,
+      "loss": 0.6512,
+      "step": 5860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.467066140088066e-05,
+      "loss": 0.698,
+      "step": 5870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4661582459485226e-05,
+      "loss": 0.5723,
+      "step": 5880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4652503518089792e-05,
+      "loss": 0.6986,
+      "step": 5890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4643424576694359e-05,
+      "loss": 0.6633,
+      "step": 5900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4634345635298925e-05,
+      "loss": 0.5979,
+      "step": 5910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4625266693903493e-05,
+      "loss": 0.6746,
+      "step": 5920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.461618775250806e-05,
+      "loss": 0.7547,
+      "step": 5930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4607108811112624e-05,
+      "loss": 0.7641,
+      "step": 5940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4598029869717192e-05,
+      "loss": 0.6393,
+      "step": 5950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.458895092832176e-05,
+      "loss": 0.6977,
+      "step": 5960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4579871986926327e-05,
+      "loss": 0.6777,
+      "step": 5970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4570793045530893e-05,
+      "loss": 0.6963,
+      "step": 5980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4561714104135459e-05,
+      "loss": 0.6664,
+      "step": 5990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4552635162740026e-05,
+      "loss": 0.5879,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27,
+      "eval_accuracy": 0.6057460164171897,
+      "eval_loss": 0.6761283278465271,
+      "eval_runtime": 72.9085,
+      "eval_samples_per_second": 56.811,
+      "eval_steps_per_second": 14.21,
+      "step": 6000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4543556221344592e-05,
+      "loss": 0.6695,
+      "step": 6010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.453447727994916e-05,
+      "loss": 0.5879,
+      "step": 6020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4525398338553727e-05,
+      "loss": 0.7484,
+      "step": 6030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4516319397158291e-05,
+      "loss": 0.6438,
+      "step": 6040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4507240455762859e-05,
+      "loss": 0.6422,
+      "step": 6050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4498161514367426e-05,
+      "loss": 0.718,
+      "step": 6060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4489082572971994e-05,
+      "loss": 0.6305,
+      "step": 6070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.448000363157656e-05,
+      "loss": 0.6605,
+      "step": 6080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4470924690181126e-05,
+      "loss": 0.6598,
+      "step": 6090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4461845748785693e-05,
+      "loss": 0.7359,
+      "step": 6100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4452766807390259e-05,
+      "loss": 0.7461,
+      "step": 6110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4443687865994826e-05,
+      "loss": 0.5742,
+      "step": 6120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4434608924599394e-05,
+      "loss": 0.6814,
+      "step": 6130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4425529983203958e-05,
+      "loss": 0.6605,
+      "step": 6140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4416451041808526e-05,
+      "loss": 0.6927,
+      "step": 6150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4407372100413093e-05,
+      "loss": 0.6391,
+      "step": 6160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.439829315901766e-05,
+      "loss": 0.6,
+      "step": 6170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4389214217622227e-05,
+      "loss": 0.6977,
+      "step": 6180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4380135276226792e-05,
+      "loss": 0.6307,
+      "step": 6190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.437105633483136e-05,
+      "loss": 0.642,
+      "step": 6200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4361977393435926e-05,
+      "loss": 0.6969,
+      "step": 6210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4352898452040493e-05,
+      "loss": 0.6883,
+      "step": 6220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4343819510645061e-05,
+      "loss": 0.6189,
+      "step": 6230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4334740569249628e-05,
+      "loss": 0.7242,
+      "step": 6240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4325661627854193e-05,
+      "loss": 0.7031,
+      "step": 6250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.431658268645876e-05,
+      "loss": 0.6645,
+      "step": 6260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4307503745063328e-05,
+      "loss": 0.6244,
+      "step": 6270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4298424803667894e-05,
+      "loss": 0.6785,
+      "step": 6280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4289345862272461e-05,
+      "loss": 0.6316,
+      "step": 6290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4280266920877027e-05,
+      "loss": 0.5918,
+      "step": 6300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4271187979481593e-05,
+      "loss": 0.6477,
+      "step": 6310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.426210903808616e-05,
+      "loss": 0.6242,
+      "step": 6320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4253030096690728e-05,
+      "loss": 0.7172,
+      "step": 6330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4243951155295295e-05,
+      "loss": 0.5865,
+      "step": 6340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.423487221389986e-05,
+      "loss": 0.5662,
+      "step": 6350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4225793272504427e-05,
+      "loss": 0.7059,
+      "step": 6360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4216714331108995e-05,
+      "loss": 0.6787,
+      "step": 6370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.420763538971356e-05,
+      "loss": 0.7768,
+      "step": 6380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4198556448318128e-05,
+      "loss": 0.6779,
+      "step": 6390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4189477506922694e-05,
+      "loss": 0.7148,
+      "step": 6400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.418039856552726e-05,
+      "loss": 0.6711,
+      "step": 6410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4171319624131827e-05,
+      "loss": 0.5986,
+      "step": 6420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4162240682736395e-05,
+      "loss": 0.7102,
+      "step": 6430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4153161741340962e-05,
+      "loss": 0.6748,
+      "step": 6440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4144082799945527e-05,
+      "loss": 0.7035,
+      "step": 6450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4135003858550094e-05,
+      "loss": 0.6783,
+      "step": 6460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4125924917154662e-05,
+      "loss": 0.64,
+      "step": 6470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4116845975759228e-05,
+      "loss": 0.5697,
+      "step": 6480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4107767034363795e-05,
+      "loss": 0.6951,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4098688092968361e-05,
+      "loss": 0.6676,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "eval_accuracy": 0.6064703042008691,
+      "eval_loss": 0.6781559586524963,
+      "eval_runtime": 74.4693,
+      "eval_samples_per_second": 55.62,
+      "eval_steps_per_second": 13.912,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4089609151572927e-05,
+      "loss": 0.7078,
+      "step": 6510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4080530210177494e-05,
+      "loss": 0.7605,
+      "step": 6520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4071451268782062e-05,
+      "loss": 0.7063,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.406237232738663e-05,
+      "loss": 0.6715,
+      "step": 6540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4053293385991194e-05,
+      "loss": 0.7559,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4044214444595761e-05,
+      "loss": 0.6426,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4035135503200329e-05,
+      "loss": 0.6973,
+      "step": 6570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4026056561804894e-05,
+      "loss": 0.6502,
+      "step": 6580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4016977620409462e-05,
+      "loss": 0.6734,
+      "step": 6590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4007898679014026e-05,
+      "loss": 0.6262,
+      "step": 6600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3998819737618594e-05,
+      "loss": 0.7363,
+      "step": 6610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3989740796223161e-05,
+      "loss": 0.6805,
+      "step": 6620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3980661854827729e-05,
+      "loss": 0.6609,
+      "step": 6630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3971582913432296e-05,
+      "loss": 0.6365,
+      "step": 6640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3962503972036862e-05,
+      "loss": 0.6311,
+      "step": 6650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3953425030641428e-05,
+      "loss": 0.7283,
+      "step": 6660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3944346089245996e-05,
+      "loss": 0.6016,
+      "step": 6670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3935267147850561e-05,
+      "loss": 0.7059,
+      "step": 6680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3926188206455129e-05,
+      "loss": 0.5814,
+      "step": 6690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3917109265059697e-05,
+      "loss": 0.6375,
+      "step": 6700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.390803032366426e-05,
+      "loss": 0.7088,
+      "step": 6710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3898951382268828e-05,
+      "loss": 0.6859,
+      "step": 6720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3889872440873396e-05,
+      "loss": 0.7002,
+      "step": 6730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3880793499477963e-05,
+      "loss": 0.7227,
+      "step": 6740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3871714558082529e-05,
+      "loss": 0.6623,
+      "step": 6750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3862635616687095e-05,
+      "loss": 0.685,
+      "step": 6760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3853556675291663e-05,
+      "loss": 0.6824,
+      "step": 6770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3844477733896228e-05,
+      "loss": 0.6699,
+      "step": 6780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3835398792500796e-05,
+      "loss": 0.7406,
+      "step": 6790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3826319851105363e-05,
+      "loss": 0.6277,
+      "step": 6800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3817240909709928e-05,
+      "loss": 0.7102,
+      "step": 6810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3808161968314495e-05,
+      "loss": 0.6322,
+      "step": 6820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3799083026919063e-05,
+      "loss": 0.7041,
+      "step": 6830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.379000408552363e-05,
+      "loss": 0.6141,
+      "step": 6840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3780925144128196e-05,
+      "loss": 0.6508,
+      "step": 6850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3771846202732762e-05,
+      "loss": 0.7113,
+      "step": 6860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.376276726133733e-05,
+      "loss": 0.7309,
+      "step": 6870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3753688319941895e-05,
+      "loss": 0.6883,
+      "step": 6880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3744609378546463e-05,
+      "loss": 0.7211,
+      "step": 6890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.373553043715103e-05,
+      "loss": 0.7285,
+      "step": 6900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3726451495755595e-05,
+      "loss": 0.7053,
+      "step": 6910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3717372554360162e-05,
+      "loss": 0.652,
+      "step": 6920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.370829361296473e-05,
+      "loss": 0.6754,
+      "step": 6930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3699214671569297e-05,
+      "loss": 0.7719,
+      "step": 6940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3690135730173863e-05,
+      "loss": 0.6084,
+      "step": 6950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3681056788778429e-05,
+      "loss": 0.6129,
+      "step": 6960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3671977847382995e-05,
+      "loss": 0.6258,
+      "step": 6970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3662898905987562e-05,
+      "loss": 0.6885,
+      "step": 6980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.365381996459213e-05,
+      "loss": 0.6445,
+      "step": 6990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3644741023196697e-05,
+      "loss": 0.7045,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "eval_accuracy": 0.6038145823273781,
+      "eval_loss": 0.6830967664718628,
+      "eval_runtime": 72.7514,
+      "eval_samples_per_second": 56.934,
+      "eval_steps_per_second": 14.24,
+      "step": 7000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3635662081801265e-05,
+      "loss": 0.6984,
+      "step": 7010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3626583140405829e-05,
+      "loss": 0.6678,
+      "step": 7020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3617504199010397e-05,
+      "loss": 0.7355,
+      "step": 7030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3608425257614964e-05,
+      "loss": 0.6607,
+      "step": 7040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.359934631621953e-05,
+      "loss": 0.6746,
+      "step": 7050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3590267374824098e-05,
+      "loss": 0.5451,
+      "step": 7060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3581188433428662e-05,
+      "loss": 0.6838,
+      "step": 7070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.357210949203323e-05,
+      "loss": 0.6322,
+      "step": 7080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3563030550637797e-05,
+      "loss": 0.5895,
+      "step": 7090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3553951609242364e-05,
+      "loss": 0.6818,
+      "step": 7100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3544872667846932e-05,
+      "loss": 0.777,
+      "step": 7110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3535793726451496e-05,
+      "loss": 0.7098,
+      "step": 7120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3526714785056064e-05,
+      "loss": 0.7652,
+      "step": 7130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3517635843660631e-05,
+      "loss": 0.6576,
+      "step": 7140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3508556902265197e-05,
+      "loss": 0.6236,
+      "step": 7150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3499477960869765e-05,
+      "loss": 0.6613,
+      "step": 7160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3490399019474329e-05,
+      "loss": 0.6992,
+      "step": 7170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3481320078078896e-05,
+      "loss": 0.5818,
+      "step": 7180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3472241136683464e-05,
+      "loss": 0.6623,
+      "step": 7190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3463162195288031e-05,
+      "loss": 0.6654,
+      "step": 7200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3454083253892599e-05,
+      "loss": 0.7695,
+      "step": 7210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3445004312497163e-05,
+      "loss": 0.6055,
+      "step": 7220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.343592537110173e-05,
+      "loss": 0.7094,
+      "step": 7230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3426846429706298e-05,
+      "loss": 0.684,
+      "step": 7240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3417767488310864e-05,
+      "loss": 0.7111,
+      "step": 7250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3408688546915431e-05,
+      "loss": 0.7848,
+      "step": 7260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3399609605519996e-05,
+      "loss": 0.7582,
+      "step": 7270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3390530664124563e-05,
+      "loss": 0.7223,
+      "step": 7280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.338145172272913e-05,
+      "loss": 0.6324,
+      "step": 7290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3372372781333698e-05,
+      "loss": 0.673,
+      "step": 7300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3363293839938266e-05,
+      "loss": 0.6451,
+      "step": 7310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.335421489854283e-05,
+      "loss": 0.6449,
+      "step": 7320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3345135957147398e-05,
+      "loss": 0.7188,
+      "step": 7330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3336057015751963e-05,
+      "loss": 0.7484,
+      "step": 7340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3326978074356531e-05,
+      "loss": 0.6664,
+      "step": 7350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3317899132961098e-05,
+      "loss": 0.7406,
+      "step": 7360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3308820191565666e-05,
+      "loss": 0.7078,
+      "step": 7370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.329974125017023e-05,
+      "loss": 0.5885,
+      "step": 7380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3290662308774798e-05,
+      "loss": 0.7395,
+      "step": 7390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3281583367379365e-05,
+      "loss": 0.6184,
+      "step": 7400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3272504425983933e-05,
+      "loss": 0.7012,
+      "step": 7410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3263425484588499e-05,
+      "loss": 0.7332,
+      "step": 7420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3254346543193064e-05,
+      "loss": 0.5527,
+      "step": 7430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.324526760179763e-05,
+      "loss": 0.6139,
+      "step": 7440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3236188660402198e-05,
+      "loss": 0.6439,
+      "step": 7450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3227109719006765e-05,
+      "loss": 0.6129,
+      "step": 7460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3218030777611333e-05,
+      "loss": 0.6891,
+      "step": 7470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3208951836215897e-05,
+      "loss": 0.7117,
+      "step": 7480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3199872894820465e-05,
+      "loss": 0.7484,
+      "step": 7490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3190793953425032e-05,
+      "loss": 0.6268,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "eval_accuracy": 0.6071945919845485,
+      "eval_loss": 0.6775137186050415,
+      "eval_runtime": 73.7281,
+      "eval_samples_per_second": 56.179,
+      "eval_steps_per_second": 14.052,
+      "step": 7500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.31817150120296e-05,
+      "loss": 0.7041,
+      "step": 7510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3172636070634166e-05,
+      "loss": 0.7602,
+      "step": 7520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3163557129238731e-05,
+      "loss": 0.6477,
+      "step": 7530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3154478187843297e-05,
+      "loss": 0.6014,
+      "step": 7540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3145399246447865e-05,
+      "loss": 0.6334,
+      "step": 7550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3136320305052432e-05,
+      "loss": 0.7918,
+      "step": 7560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3127241363657e-05,
+      "loss": 0.6686,
+      "step": 7570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3118162422261564e-05,
+      "loss": 0.6744,
+      "step": 7580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3109083480866132e-05,
+      "loss": 0.6256,
+      "step": 7590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.31000045394707e-05,
+      "loss": 0.675,
+      "step": 7600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3090925598075267e-05,
+      "loss": 0.7121,
+      "step": 7610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3081846656679833e-05,
+      "loss": 0.6578,
+      "step": 7620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3072767715284398e-05,
+      "loss": 0.6576,
+      "step": 7630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3063688773888964e-05,
+      "loss": 0.6482,
+      "step": 7640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3054609832493532e-05,
+      "loss": 0.5342,
+      "step": 7650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.30455308910981e-05,
+      "loss": 0.6906,
+      "step": 7660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3036451949702667e-05,
+      "loss": 0.6277,
+      "step": 7670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3027373008307231e-05,
+      "loss": 0.5578,
+      "step": 7680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3018294066911799e-05,
+      "loss": 0.5428,
+      "step": 7690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3009215125516366e-05,
+      "loss": 0.6779,
+      "step": 7700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3000136184120932e-05,
+      "loss": 0.7418,
+      "step": 7710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.29910572427255e-05,
+      "loss": 0.6016,
+      "step": 7720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2981978301330065e-05,
+      "loss": 0.7863,
+      "step": 7730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2972899359934631e-05,
+      "loss": 0.6191,
+      "step": 7740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2963820418539199e-05,
+      "loss": 0.6687,
+      "step": 7750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2954741477143766e-05,
+      "loss": 0.6523,
+      "step": 7760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2945662535748334e-05,
+      "loss": 0.7125,
+      "step": 7770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2936583594352901e-05,
+      "loss": 0.5959,
+      "step": 7780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2927504652957466e-05,
+      "loss": 0.6693,
+      "step": 7790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2918425711562033e-05,
+      "loss": 0.6715,
+      "step": 7800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2909346770166599e-05,
+      "loss": 0.7223,
+      "step": 7810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2900267828771166e-05,
+      "loss": 0.6242,
+      "step": 7820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2891188887375734e-05,
+      "loss": 0.7363,
+      "step": 7830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2882109945980298e-05,
+      "loss": 0.6469,
+      "step": 7840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2873031004584866e-05,
+      "loss": 0.632,
+      "step": 7850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2863952063189433e-05,
+      "loss": 0.69,
+      "step": 7860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2854873121794e-05,
+      "loss": 0.7389,
+      "step": 7870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2845794180398568e-05,
+      "loss": 0.693,
+      "step": 7880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2836715239003133e-05,
+      "loss": 0.6895,
+      "step": 7890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.28276362976077e-05,
+      "loss": 0.6652,
+      "step": 7900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2818557356212266e-05,
+      "loss": 0.6574,
+      "step": 7910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2809478414816833e-05,
+      "loss": 0.7434,
+      "step": 7920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2800399473421401e-05,
+      "loss": 0.6785,
+      "step": 7930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2791320532025965e-05,
+      "loss": 0.6727,
+      "step": 7940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2782241590630533e-05,
+      "loss": 0.6723,
+      "step": 7950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.27731626492351e-05,
+      "loss": 0.7227,
+      "step": 7960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2764083707839668e-05,
+      "loss": 0.6693,
+      "step": 7970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2755004766444235e-05,
+      "loss": 0.5281,
+      "step": 7980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.27459258250488e-05,
+      "loss": 0.7488,
+      "step": 7990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2736846883653367e-05,
+      "loss": 0.5973,
+      "step": 8000
+    },
+    {
+      "epoch": 0.36,
+      "eval_accuracy": 0.6120231772090777,
+      "eval_loss": 0.6748174428939819,
+      "eval_runtime": 73.8533,
+      "eval_samples_per_second": 56.084,
+      "eval_steps_per_second": 14.028,
+      "step": 8000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2727767942257933e-05,
+      "loss": 0.6354,
+      "step": 8010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.27186890008625e-05,
+      "loss": 0.6363,
+      "step": 8020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2709610059467068e-05,
+      "loss": 0.7402,
+      "step": 8030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2700531118071632e-05,
+      "loss": 0.675,
+      "step": 8040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.26914521766762e-05,
+      "loss": 0.7207,
+      "step": 8050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2682373235280767e-05,
+      "loss": 0.6342,
+      "step": 8060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2673294293885335e-05,
+      "loss": 0.6756,
+      "step": 8070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.26642153524899e-05,
+      "loss": 0.7234,
+      "step": 8080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2655136411094466e-05,
+      "loss": 0.6652,
+      "step": 8090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2646057469699034e-05,
+      "loss": 0.6008,
+      "step": 8100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.26369785283036e-05,
+      "loss": 0.7371,
+      "step": 8110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2627899586908167e-05,
+      "loss": 0.6217,
+      "step": 8120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2618820645512735e-05,
+      "loss": 0.6926,
+      "step": 8130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2609741704117302e-05,
+      "loss": 0.6658,
+      "step": 8140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2600662762721867e-05,
+      "loss": 0.6504,
+      "step": 8150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2591583821326434e-05,
+      "loss": 0.8324,
+      "step": 8160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2582504879931002e-05,
+      "loss": 0.6648,
+      "step": 8170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2573425938535568e-05,
+      "loss": 0.6914,
+      "step": 8180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2564346997140135e-05,
+      "loss": 0.6416,
+      "step": 8190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2555268055744701e-05,
+      "loss": 0.5684,
+      "step": 8200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2546189114349267e-05,
+      "loss": 0.6676,
+      "step": 8210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2537110172953834e-05,
+      "loss": 0.6582,
+      "step": 8220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2528031231558402e-05,
+      "loss": 0.7184,
+      "step": 8230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.251895229016297e-05,
+      "loss": 0.6281,
+      "step": 8240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2509873348767534e-05,
+      "loss": 0.7473,
+      "step": 8250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2500794407372101e-05,
+      "loss": 0.6387,
+      "step": 8260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2491715465976669e-05,
+      "loss": 0.7762,
+      "step": 8270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2482636524581235e-05,
+      "loss": 0.6533,
+      "step": 8280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2473557583185802e-05,
+      "loss": 0.7139,
+      "step": 8290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2464478641790368e-05,
+      "loss": 0.6283,
+      "step": 8300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2455399700394934e-05,
+      "loss": 0.5828,
+      "step": 8310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2446320758999501e-05,
+      "loss": 0.6191,
+      "step": 8320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2437241817604069e-05,
+      "loss": 0.6521,
+      "step": 8330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2428162876208636e-05,
+      "loss": 0.7365,
+      "step": 8340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.24190839348132e-05,
+      "loss": 0.6521,
+      "step": 8350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2410004993417768e-05,
+      "loss": 0.7143,
+      "step": 8360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2400926052022336e-05,
+      "loss": 0.5857,
+      "step": 8370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2391847110626901e-05,
+      "loss": 0.6746,
+      "step": 8380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2382768169231469e-05,
+      "loss": 0.7582,
+      "step": 8390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2373689227836035e-05,
+      "loss": 0.6582,
+      "step": 8400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.23646102864406e-05,
+      "loss": 0.6844,
+      "step": 8410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2355531345045168e-05,
+      "loss": 0.6281,
+      "step": 8420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2346452403649736e-05,
+      "loss": 0.6521,
+      "step": 8430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2337373462254303e-05,
+      "loss": 0.6479,
+      "step": 8440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2328294520858867e-05,
+      "loss": 0.7461,
+      "step": 8450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2319215579463435e-05,
+      "loss": 0.6861,
+      "step": 8460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2310136638068003e-05,
+      "loss": 0.6818,
+      "step": 8470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2301057696672568e-05,
+      "loss": 0.6973,
+      "step": 8480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2291978755277136e-05,
+      "loss": 0.6914,
+      "step": 8490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2282899813881704e-05,
+      "loss": 0.6592,
+      "step": 8500
+    },
+    {
+      "epoch": 0.39,
+      "eval_accuracy": 0.613713182037663,
+      "eval_loss": 0.6704452633857727,
+      "eval_runtime": 73.8233,
+      "eval_samples_per_second": 56.107,
+      "eval_steps_per_second": 14.034,
+      "step": 8500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2273820872486268e-05,
+      "loss": 0.6002,
+      "step": 8510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2264741931090835e-05,
+      "loss": 0.5867,
+      "step": 8520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2255662989695403e-05,
+      "loss": 0.6746,
+      "step": 8530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.224658404829997e-05,
+      "loss": 0.7316,
+      "step": 8540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2237505106904536e-05,
+      "loss": 0.7844,
+      "step": 8550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2228426165509102e-05,
+      "loss": 0.7021,
+      "step": 8560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.221934722411367e-05,
+      "loss": 0.6369,
+      "step": 8570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2210268282718235e-05,
+      "loss": 0.6469,
+      "step": 8580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2201189341322803e-05,
+      "loss": 0.7189,
+      "step": 8590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.219211039992737e-05,
+      "loss": 0.668,
+      "step": 8600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2183031458531935e-05,
+      "loss": 0.6777,
+      "step": 8610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2173952517136502e-05,
+      "loss": 0.5765,
+      "step": 8620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.216487357574107e-05,
+      "loss": 0.6963,
+      "step": 8630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2155794634345637e-05,
+      "loss": 0.6502,
+      "step": 8640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2146715692950203e-05,
+      "loss": 0.7574,
+      "step": 8650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2137636751554769e-05,
+      "loss": 0.7723,
+      "step": 8660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2128557810159336e-05,
+      "loss": 0.6846,
+      "step": 8670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2119478868763902e-05,
+      "loss": 0.6258,
+      "step": 8680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.211039992736847e-05,
+      "loss": 0.6141,
+      "step": 8690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2101320985973037e-05,
+      "loss": 0.576,
+      "step": 8700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2092242044577602e-05,
+      "loss": 0.7633,
+      "step": 8710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2083163103182169e-05,
+      "loss": 0.6477,
+      "step": 8720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2074084161786737e-05,
+      "loss": 0.7104,
+      "step": 8730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2065005220391304e-05,
+      "loss": 0.673,
+      "step": 8740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.205592627899587e-05,
+      "loss": 0.6516,
+      "step": 8750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2046847337600436e-05,
+      "loss": 0.7199,
+      "step": 8760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2037768396205003e-05,
+      "loss": 0.7305,
+      "step": 8770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.202868945480957e-05,
+      "loss": 0.6078,
+      "step": 8780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2019610513414137e-05,
+      "loss": 0.6326,
+      "step": 8790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2010531572018704e-05,
+      "loss": 0.6406,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2001452630623269e-05,
+      "loss": 0.7246,
+      "step": 8810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1992373689227836e-05,
+      "loss": 0.7004,
+      "step": 8820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1983294747832404e-05,
+      "loss": 0.6844,
+      "step": 8830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1974215806436971e-05,
+      "loss": 0.7195,
+      "step": 8840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1965136865041537e-05,
+      "loss": 0.6477,
+      "step": 8850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1956057923646103e-05,
+      "loss": 0.7027,
+      "step": 8860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.194697898225067e-05,
+      "loss": 0.6791,
+      "step": 8870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1937900040855236e-05,
+      "loss": 0.5969,
+      "step": 8880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1928821099459804e-05,
+      "loss": 0.674,
+      "step": 8890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1919742158064371e-05,
+      "loss": 0.6832,
+      "step": 8900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1910663216668939e-05,
+      "loss": 0.6639,
+      "step": 8910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1901584275273503e-05,
+      "loss": 0.6672,
+      "step": 8920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.189250533387807e-05,
+      "loss": 0.6502,
+      "step": 8930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1883426392482638e-05,
+      "loss": 0.6389,
+      "step": 8940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1874347451087204e-05,
+      "loss": 0.6297,
+      "step": 8950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1865268509691772e-05,
+      "loss": 0.6564,
+      "step": 8960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1856189568296337e-05,
+      "loss": 0.66,
+      "step": 8970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1847110626900903e-05,
+      "loss": 0.8164,
+      "step": 8980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.183803168550547e-05,
+      "loss": 0.6418,
+      "step": 8990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1828952744110038e-05,
+      "loss": 0.6805,
+      "step": 9000
+    },
+    {
+      "epoch": 0.41,
+      "eval_accuracy": 0.6139546112988894,
+      "eval_loss": 0.6687760353088379,
+      "eval_runtime": 74.2738,
+      "eval_samples_per_second": 55.767,
+      "eval_steps_per_second": 13.948,
+      "step": 9000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1819873802714606e-05,
+      "loss": 0.6879,
+      "step": 9010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.181079486131917e-05,
+      "loss": 0.7094,
+      "step": 9020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1801715919923738e-05,
+      "loss": 0.6699,
+      "step": 9030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1792636978528305e-05,
+      "loss": 0.7035,
+      "step": 9040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1783558037132871e-05,
+      "loss": 0.7768,
+      "step": 9050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1774479095737438e-05,
+      "loss": 0.6312,
+      "step": 9060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1765400154342004e-05,
+      "loss": 0.658,
+      "step": 9070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.175632121294657e-05,
+      "loss": 0.6699,
+      "step": 9080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1747242271551138e-05,
+      "loss": 0.7354,
+      "step": 9090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1738163330155705e-05,
+      "loss": 0.6662,
+      "step": 9100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1729084388760273e-05,
+      "loss": 0.6908,
+      "step": 9110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1720005447364837e-05,
+      "loss": 0.7621,
+      "step": 9120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1710926505969405e-05,
+      "loss": 0.7016,
+      "step": 9130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1701847564573972e-05,
+      "loss": 0.6984,
+      "step": 9140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1692768623178538e-05,
+      "loss": 0.6508,
+      "step": 9150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1683689681783105e-05,
+      "loss": 0.6754,
+      "step": 9160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1674610740387671e-05,
+      "loss": 0.6082,
+      "step": 9170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1665531798992237e-05,
+      "loss": 0.651,
+      "step": 9180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1656452857596805e-05,
+      "loss": 0.7121,
+      "step": 9190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1647373916201372e-05,
+      "loss": 0.793,
+      "step": 9200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.163829497480594e-05,
+      "loss": 0.5602,
+      "step": 9210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1629216033410504e-05,
+      "loss": 0.6713,
+      "step": 9220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1620137092015071e-05,
+      "loss": 0.7727,
+      "step": 9230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1611058150619639e-05,
+      "loss": 0.6102,
+      "step": 9240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1601979209224205e-05,
+      "loss": 0.6553,
+      "step": 9250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1592900267828772e-05,
+      "loss": 0.6861,
+      "step": 9260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.158382132643334e-05,
+      "loss": 0.6576,
+      "step": 9270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1574742385037904e-05,
+      "loss": 0.6504,
+      "step": 9280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1565663443642472e-05,
+      "loss": 0.6322,
+      "step": 9290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.155658450224704e-05,
+      "loss": 0.6512,
+      "step": 9300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1547505560851607e-05,
+      "loss": 0.6479,
+      "step": 9310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1538426619456173e-05,
+      "loss": 0.6904,
+      "step": 9320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1529347678060738e-05,
+      "loss": 0.7137,
+      "step": 9330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1520268736665306e-05,
+      "loss": 0.6318,
+      "step": 9340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1511189795269872e-05,
+      "loss": 0.7457,
+      "step": 9350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.150211085387444e-05,
+      "loss": 0.6027,
+      "step": 9360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1493031912479007e-05,
+      "loss": 0.6516,
+      "step": 9370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1483952971083571e-05,
+      "loss": 0.6967,
+      "step": 9380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1474874029688139e-05,
+      "loss": 0.7232,
+      "step": 9390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1465795088292706e-05,
+      "loss": 0.7355,
+      "step": 9400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1456716146897274e-05,
+      "loss": 0.8191,
+      "step": 9410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.144763720550184e-05,
+      "loss": 0.6443,
+      "step": 9420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1438558264106405e-05,
+      "loss": 0.6482,
+      "step": 9430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1429479322710973e-05,
+      "loss": 0.7633,
+      "step": 9440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1420400381315539e-05,
+      "loss": 0.6391,
+      "step": 9450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1411321439920106e-05,
+      "loss": 0.7008,
+      "step": 9460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1402242498524674e-05,
+      "loss": 0.6922,
+      "step": 9470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1393163557129238e-05,
+      "loss": 0.5777,
+      "step": 9480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1384084615733806e-05,
+      "loss": 0.6977,
+      "step": 9490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1375005674338373e-05,
+      "loss": 0.5785,
+      "step": 9500
+    },
+    {
+      "epoch": 0.43,
+      "eval_accuracy": 0.6129888942539836,
+      "eval_loss": 0.6736442446708679,
+      "eval_runtime": 72.7311,
+      "eval_samples_per_second": 56.949,
+      "eval_steps_per_second": 14.244,
+      "step": 9500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.136592673294294e-05,
+      "loss": 0.5643,
+      "step": 9510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1356847791547507e-05,
+      "loss": 0.6428,
+      "step": 9520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1347768850152072e-05,
+      "loss": 0.6371,
+      "step": 9530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.133868990875664e-05,
+      "loss": 0.633,
+      "step": 9540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1329610967361206e-05,
+      "loss": 0.7713,
+      "step": 9550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1320532025965773e-05,
+      "loss": 0.7596,
+      "step": 9560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.131145308457034e-05,
+      "loss": 0.635,
+      "step": 9570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1302374143174905e-05,
+      "loss": 0.6986,
+      "step": 9580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1293295201779473e-05,
+      "loss": 0.6287,
+      "step": 9590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.128421626038404e-05,
+      "loss": 0.6723,
+      "step": 9600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1275137318988608e-05,
+      "loss": 0.5947,
+      "step": 9610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1266058377593173e-05,
+      "loss": 0.757,
+      "step": 9620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1256979436197741e-05,
+      "loss": 0.5848,
+      "step": 9630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1247900494802307e-05,
+      "loss": 0.7131,
+      "step": 9640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1238821553406873e-05,
+      "loss": 0.6102,
+      "step": 9650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.122974261201144e-05,
+      "loss": 0.7449,
+      "step": 9660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1220663670616008e-05,
+      "loss": 0.6512,
+      "step": 9670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1211584729220575e-05,
+      "loss": 0.6586,
+      "step": 9680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.120250578782514e-05,
+      "loss": 0.6793,
+      "step": 9690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1193426846429707e-05,
+      "loss": 0.6764,
+      "step": 9700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1184347905034275e-05,
+      "loss": 0.7092,
+      "step": 9710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.117526896363884e-05,
+      "loss": 0.6264,
+      "step": 9720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1166190022243408e-05,
+      "loss": 0.6221,
+      "step": 9730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1157111080847974e-05,
+      "loss": 0.6859,
+      "step": 9740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.114803213945254e-05,
+      "loss": 0.6084,
+      "step": 9750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1138953198057107e-05,
+      "loss": 0.6539,
+      "step": 9760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1129874256661675e-05,
+      "loss": 0.6131,
+      "step": 9770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1120795315266242e-05,
+      "loss": 0.6937,
+      "step": 9780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1111716373870806e-05,
+      "loss": 0.7943,
+      "step": 9790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1102637432475374e-05,
+      "loss": 0.6664,
+      "step": 9800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1093558491079942e-05,
+      "loss": 0.7092,
+      "step": 9810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1084479549684507e-05,
+      "loss": 0.6586,
+      "step": 9820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1075400608289075e-05,
+      "loss": 0.6746,
+      "step": 9830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.106632166689364e-05,
+      "loss": 0.6631,
+      "step": 9840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1057242725498207e-05,
+      "loss": 0.6672,
+      "step": 9850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1048163784102774e-05,
+      "loss": 0.6475,
+      "step": 9860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1039084842707342e-05,
+      "loss": 0.7523,
+      "step": 9870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.103000590131191e-05,
+      "loss": 0.6203,
+      "step": 9880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1020926959916473e-05,
+      "loss": 0.6355,
+      "step": 9890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1011848018521041e-05,
+      "loss": 0.5984,
+      "step": 9900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1002769077125609e-05,
+      "loss": 0.7189,
+      "step": 9910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0993690135730174e-05,
+      "loss": 0.7266,
+      "step": 9920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0984611194334742e-05,
+      "loss": 0.7332,
+      "step": 9930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0975532252939308e-05,
+      "loss": 0.6303,
+      "step": 9940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0966453311543874e-05,
+      "loss": 0.5779,
+      "step": 9950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0957374370148441e-05,
+      "loss": 0.648,
+      "step": 9960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0948295428753009e-05,
+      "loss": 0.6674,
+      "step": 9970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0939216487357576e-05,
+      "loss": 0.7422,
+      "step": 9980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.093013754596214e-05,
+      "loss": 0.7232,
+      "step": 9990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0921058604566708e-05,
+      "loss": 0.742,
+      "step": 10000
+    },
+    {
+      "epoch": 0.45,
+      "eval_accuracy": 0.6108160309029455,
+      "eval_loss": 0.6693569421768188,
+      "eval_runtime": 72.9525,
+      "eval_samples_per_second": 56.777,
+      "eval_steps_per_second": 14.201,
+      "step": 10000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0911979663171275e-05,
+      "loss": 0.5943,
+      "step": 10010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0902900721775841e-05,
+      "loss": 0.6332,
+      "step": 10020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0893821780380409e-05,
+      "loss": 0.7078,
+      "step": 10030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0884742838984976e-05,
+      "loss": 0.6781,
+      "step": 10040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.087566389758954e-05,
+      "loss": 0.7174,
+      "step": 10050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0866584956194108e-05,
+      "loss": 0.727,
+      "step": 10060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0857506014798676e-05,
+      "loss": 0.6676,
+      "step": 10070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0848427073403243e-05,
+      "loss": 0.6469,
+      "step": 10080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0839348132007809e-05,
+      "loss": 0.7363,
+      "step": 10090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0830269190612375e-05,
+      "loss": 0.768,
+      "step": 10100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0821190249216942e-05,
+      "loss": 0.6092,
+      "step": 10110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0812111307821508e-05,
+      "loss": 0.6672,
+      "step": 10120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0803032366426076e-05,
+      "loss": 0.6822,
+      "step": 10130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0793953425030643e-05,
+      "loss": 0.6617,
+      "step": 10140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0784874483635208e-05,
+      "loss": 0.6945,
+      "step": 10150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0775795542239775e-05,
+      "loss": 0.6867,
+      "step": 10160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0766716600844343e-05,
+      "loss": 0.6332,
+      "step": 10170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.075763765944891e-05,
+      "loss": 0.6752,
+      "step": 10180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0748558718053476e-05,
+      "loss": 0.6369,
+      "step": 10190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0739479776658042e-05,
+      "loss": 0.6684,
+      "step": 10200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.073040083526261e-05,
+      "loss": 0.5461,
+      "step": 10210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0721321893867175e-05,
+      "loss": 0.7184,
+      "step": 10220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0712242952471743e-05,
+      "loss": 0.7496,
+      "step": 10230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.070316401107631e-05,
+      "loss": 0.6715,
+      "step": 10240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0694085069680874e-05,
+      "loss": 0.6252,
+      "step": 10250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0685006128285442e-05,
+      "loss": 0.6961,
+      "step": 10260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.067592718689001e-05,
+      "loss": 0.716,
+      "step": 10270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0666848245494577e-05,
+      "loss": 0.7164,
+      "step": 10280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0657769304099143e-05,
+      "loss": 0.6246,
+      "step": 10290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0648690362703709e-05,
+      "loss": 0.6184,
+      "step": 10300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0639611421308276e-05,
+      "loss": 0.6963,
+      "step": 10310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0630532479912842e-05,
+      "loss": 0.6121,
+      "step": 10320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.062145353851741e-05,
+      "loss": 0.652,
+      "step": 10330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0612374597121977e-05,
+      "loss": 0.6271,
+      "step": 10340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0603295655726541e-05,
+      "loss": 0.624,
+      "step": 10350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0594216714331109e-05,
+      "loss": 0.6789,
+      "step": 10360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0585137772935677e-05,
+      "loss": 0.615,
+      "step": 10370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0576058831540244e-05,
+      "loss": 0.6789,
+      "step": 10380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.056697989014481e-05,
+      "loss": 0.6871,
+      "step": 10390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0557900948749377e-05,
+      "loss": 0.6904,
+      "step": 10400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0548822007353943e-05,
+      "loss": 0.6643,
+      "step": 10410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.053974306595851e-05,
+      "loss": 0.6654,
+      "step": 10420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0530664124563077e-05,
+      "loss": 0.7045,
+      "step": 10430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0521585183167644e-05,
+      "loss": 0.6736,
+      "step": 10440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0512506241772212e-05,
+      "loss": 0.7609,
+      "step": 10450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0503427300376776e-05,
+      "loss": 0.6813,
+      "step": 10460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0494348358981343e-05,
+      "loss": 0.6705,
+      "step": 10470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0485269417585911e-05,
+      "loss": 0.6914,
+      "step": 10480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.7188,
+      "step": 10490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0467111534795044e-05,
+      "loss": 0.6529,
+      "step": 10500
+    },
+    {
+      "epoch": 0.48,
+      "eval_accuracy": 0.6166103331723805,
+      "eval_loss": 0.6667417883872986,
+      "eval_runtime": 72.4713,
+      "eval_samples_per_second": 57.154,
+      "eval_steps_per_second": 14.295,
+      "step": 10500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.045803259339961e-05,
+      "loss": 0.6578,
+      "step": 10510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0448953652004176e-05,
+      "loss": 0.7453,
+      "step": 10520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0439874710608744e-05,
+      "loss": 0.6691,
+      "step": 10530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0430795769213311e-05,
+      "loss": 0.6914,
+      "step": 10540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0421716827817879e-05,
+      "loss": 0.7102,
+      "step": 10550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0412637886422443e-05,
+      "loss": 0.6609,
+      "step": 10560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.040355894502701e-05,
+      "loss": 0.6842,
+      "step": 10570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0394480003631578e-05,
+      "loss": 0.6832,
+      "step": 10580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0385401062236144e-05,
+      "loss": 0.6773,
+      "step": 10590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0376322120840711e-05,
+      "loss": 0.6449,
+      "step": 10600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0367243179445277e-05,
+      "loss": 0.6217,
+      "step": 10610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0358164238049843e-05,
+      "loss": 0.6234,
+      "step": 10620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.034908529665441e-05,
+      "loss": 0.5896,
+      "step": 10630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0340006355258978e-05,
+      "loss": 0.6473,
+      "step": 10640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0330927413863546e-05,
+      "loss": 0.7066,
+      "step": 10650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.032184847246811e-05,
+      "loss": 0.5883,
+      "step": 10660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0312769531072677e-05,
+      "loss": 0.6541,
+      "step": 10670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0303690589677245e-05,
+      "loss": 0.6373,
+      "step": 10680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.029461164828181e-05,
+      "loss": 0.7643,
+      "step": 10690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0285532706886378e-05,
+      "loss": 0.6492,
+      "step": 10700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0276453765490944e-05,
+      "loss": 0.7371,
+      "step": 10710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.026737482409551e-05,
+      "loss": 0.698,
+      "step": 10720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0258295882700078e-05,
+      "loss": 0.7586,
+      "step": 10730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0249216941304645e-05,
+      "loss": 0.6699,
+      "step": 10740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0240137999909213e-05,
+      "loss": 0.708,
+      "step": 10750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0231059058513779e-05,
+      "loss": 0.6816,
+      "step": 10760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0221980117118344e-05,
+      "loss": 0.6832,
+      "step": 10770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0212901175722912e-05,
+      "loss": 0.7,
+      "step": 10780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0203822234327478e-05,
+      "loss": 0.6459,
+      "step": 10790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0194743292932045e-05,
+      "loss": 0.6645,
+      "step": 10800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0185664351536613e-05,
+      "loss": 0.6486,
+      "step": 10810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0176585410141177e-05,
+      "loss": 0.6078,
+      "step": 10820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0167506468745745e-05,
+      "loss": 0.6418,
+      "step": 10830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0158427527350312e-05,
+      "loss": 0.6291,
+      "step": 10840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.014934858595488e-05,
+      "loss": 0.6855,
+      "step": 10850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0140269644559445e-05,
+      "loss": 0.6988,
+      "step": 10860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0131190703164011e-05,
+      "loss": 0.6133,
+      "step": 10870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0122111761768579e-05,
+      "loss": 0.607,
+      "step": 10880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0113032820373145e-05,
+      "loss": 0.7082,
+      "step": 10890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0103953878977712e-05,
+      "loss": 0.6539,
+      "step": 10900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009487493758228e-05,
+      "loss": 0.6391,
+      "step": 10910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0085795996186844e-05,
+      "loss": 0.7664,
+      "step": 10920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076717054791412e-05,
+      "loss": 0.7229,
+      "step": 10930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0067638113395979e-05,
+      "loss": 0.6549,
+      "step": 10940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0058559172000547e-05,
+      "loss": 0.7496,
+      "step": 10950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0049480230605112e-05,
+      "loss": 0.743,
+      "step": 10960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0040401289209678e-05,
+      "loss": 0.7393,
+      "step": 10970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0031322347814246e-05,
+      "loss": 0.7133,
+      "step": 10980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0022243406418812e-05,
+      "loss": 0.6711,
+      "step": 10990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.001316446502338e-05,
+      "loss": 0.6803,
+      "step": 11000
+    },
+    {
+      "epoch": 0.5,
+      "eval_accuracy": 0.6214389183969097,
+      "eval_loss": 0.6656657457351685,
+      "eval_runtime": 72.6193,
+      "eval_samples_per_second": 57.037,
+      "eval_steps_per_second": 14.266,
+      "step": 11000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0004085523627947e-05,
+      "loss": 0.6867,
+      "step": 11010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.995006582232513e-06,
+      "loss": 0.6211,
+      "step": 11020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.985927640837078e-06,
+      "loss": 0.7129,
+      "step": 11030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.976848699441646e-06,
+      "loss": 0.7109,
+      "step": 11040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.967769758046214e-06,
+      "loss": 0.6867,
+      "step": 11050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.95869081665078e-06,
+      "loss": 0.6592,
+      "step": 11060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.949611875255345e-06,
+      "loss": 0.6264,
+      "step": 11070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.940532933859913e-06,
+      "loss": 0.6008,
+      "step": 11080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.931453992464479e-06,
+      "loss": 0.684,
+      "step": 11090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.922375051069046e-06,
+      "loss": 0.6029,
+      "step": 11100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.913296109673612e-06,
+      "loss": 0.6953,
+      "step": 11110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90421716827818e-06,
+      "loss": 0.6785,
+      "step": 11120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.895138226882747e-06,
+      "loss": 0.5875,
+      "step": 11130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.886059285487313e-06,
+      "loss": 0.6793,
+      "step": 11140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.87698034409188e-06,
+      "loss": 0.6586,
+      "step": 11150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.867901402696446e-06,
+      "loss": 0.607,
+      "step": 11160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.858822461301012e-06,
+      "loss": 0.5693,
+      "step": 11170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.84974351990558e-06,
+      "loss": 0.6781,
+      "step": 11180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.840664578510146e-06,
+      "loss": 0.6658,
+      "step": 11190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.831585637114713e-06,
+      "loss": 0.7047,
+      "step": 11200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.822506695719279e-06,
+      "loss": 0.7633,
+      "step": 11210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.813427754323847e-06,
+      "loss": 0.6619,
+      "step": 11220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.804348812928414e-06,
+      "loss": 0.6832,
+      "step": 11230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.79526987153298e-06,
+      "loss": 0.6477,
+      "step": 11240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.786190930137547e-06,
+      "loss": 0.6107,
+      "step": 11250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.777111988742113e-06,
+      "loss": 0.658,
+      "step": 11260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76803304734668e-06,
+      "loss": 0.6137,
+      "step": 11270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.758954105951247e-06,
+      "loss": 0.5883,
+      "step": 11280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.749875164555813e-06,
+      "loss": 0.683,
+      "step": 11290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.74079622316038e-06,
+      "loss": 0.6953,
+      "step": 11300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.731717281764948e-06,
+      "loss": 0.5729,
+      "step": 11310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.722638340369514e-06,
+      "loss": 0.7637,
+      "step": 11320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.713559398974081e-06,
+      "loss": 0.7246,
+      "step": 11330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.704480457578647e-06,
+      "loss": 0.6527,
+      "step": 11340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695401516183214e-06,
+      "loss": 0.7195,
+      "step": 11350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68632257478778e-06,
+      "loss": 0.6072,
+      "step": 11360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677243633392346e-06,
+      "loss": 0.6926,
+      "step": 11370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.668164691996914e-06,
+      "loss": 0.649,
+      "step": 11380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.65908575060148e-06,
+      "loss": 0.5658,
+      "step": 11390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.650006809206047e-06,
+      "loss": 0.7312,
+      "step": 11400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.640927867810615e-06,
+      "loss": 0.6996,
+      "step": 11410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.63184892641518e-06,
+      "loss": 0.6453,
+      "step": 11420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.622769985019748e-06,
+      "loss": 0.6148,
+      "step": 11430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.613691043624314e-06,
+      "loss": 0.734,
+      "step": 11440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.604612102228881e-06,
+      "loss": 0.633,
+      "step": 11450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.595533160833447e-06,
+      "loss": 0.6434,
+      "step": 11460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.586454219438013e-06,
+      "loss": 0.6508,
+      "step": 11470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.57737527804258e-06,
+      "loss": 0.7098,
+      "step": 11480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.568296336647148e-06,
+      "loss": 0.6662,
+      "step": 11490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.559217395251714e-06,
+      "loss": 0.6588,
+      "step": 11500
+    },
+    {
+      "epoch": 0.52,
+      "eval_accuracy": 0.6219217769193627,
+      "eval_loss": 0.6640219688415527,
+      "eval_runtime": 74.4423,
+      "eval_samples_per_second": 55.64,
+      "eval_steps_per_second": 13.917,
+      "step": 11500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.550138453856282e-06,
+      "loss": 0.7092,
+      "step": 11510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.541059512460847e-06,
+      "loss": 0.6582,
+      "step": 11520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.531980571065415e-06,
+      "loss": 0.6588,
+      "step": 11530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.52290162966998e-06,
+      "loss": 0.7258,
+      "step": 11540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.513822688274548e-06,
+      "loss": 0.7121,
+      "step": 11550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.504743746879114e-06,
+      "loss": 0.6375,
+      "step": 11560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49566480548368e-06,
+      "loss": 0.6449,
+      "step": 11570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486585864088248e-06,
+      "loss": 0.7689,
+      "step": 11580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477506922692815e-06,
+      "loss": 0.6449,
+      "step": 11590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468427981297381e-06,
+      "loss": 0.7828,
+      "step": 11600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459349039901949e-06,
+      "loss": 0.5143,
+      "step": 11610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450270098506516e-06,
+      "loss": 0.6818,
+      "step": 11620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441191157111082e-06,
+      "loss": 0.6859,
+      "step": 11630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432112215715648e-06,
+      "loss": 0.5936,
+      "step": 11640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423033274320215e-06,
+      "loss": 0.6316,
+      "step": 11650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413954332924781e-06,
+      "loss": 0.6494,
+      "step": 11660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404875391529349e-06,
+      "loss": 0.6939,
+      "step": 11670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395796450133915e-06,
+      "loss": 0.6666,
+      "step": 11680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386717508738482e-06,
+      "loss": 0.6594,
+      "step": 11690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377638567343048e-06,
+      "loss": 0.6082,
+      "step": 11700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368559625947616e-06,
+      "loss": 0.6211,
+      "step": 11710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359480684552183e-06,
+      "loss": 0.7973,
+      "step": 11720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.350401743156749e-06,
+      "loss": 0.7477,
+      "step": 11730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.341322801761315e-06,
+      "loss": 0.5236,
+      "step": 11740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.332243860365882e-06,
+      "loss": 0.6607,
+      "step": 11750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.323164918970448e-06,
+      "loss": 0.6092,
+      "step": 11760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.314085977575016e-06,
+      "loss": 0.6771,
+      "step": 11770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.305007036179582e-06,
+      "loss": 0.6631,
+      "step": 11780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295928094784149e-06,
+      "loss": 0.6256,
+      "step": 11790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286849153388717e-06,
+      "loss": 0.7223,
+      "step": 11800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277770211993282e-06,
+      "loss": 0.759,
+      "step": 11810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26869127059785e-06,
+      "loss": 0.6131,
+      "step": 11820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259612329202416e-06,
+      "loss": 0.6332,
+      "step": 11830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.250533387806982e-06,
+      "loss": 0.6002,
+      "step": 11840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24145444641155e-06,
+      "loss": 0.5533,
+      "step": 11850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232375505016115e-06,
+      "loss": 0.6191,
+      "step": 11860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223296563620683e-06,
+      "loss": 0.7561,
+      "step": 11870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214217622225248e-06,
+      "loss": 0.6852,
+      "step": 11880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205138680829816e-06,
+      "loss": 0.7559,
+      "step": 11890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196059739434384e-06,
+      "loss": 0.6805,
+      "step": 11900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18698079803895e-06,
+      "loss": 0.6937,
+      "step": 11910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177901856643517e-06,
+      "loss": 0.766,
+      "step": 11920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168822915248083e-06,
+      "loss": 0.6945,
+      "step": 11930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159743973852649e-06,
+      "loss": 0.5963,
+      "step": 11940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150665032457216e-06,
+      "loss": 0.5949,
+      "step": 11950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141586091061782e-06,
+      "loss": 0.7,
+      "step": 11960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13250714966635e-06,
+      "loss": 0.6602,
+      "step": 11970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123428208270917e-06,
+      "loss": 0.6168,
+      "step": 11980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114349266875483e-06,
+      "loss": 0.65,
+      "step": 11990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10527032548005e-06,
+      "loss": 0.5652,
+      "step": 12000
+    },
+    {
+      "epoch": 0.54,
+      "eval_accuracy": 0.6262675036214389,
+      "eval_loss": 0.6643086671829224,
+      "eval_runtime": 74.1561,
+      "eval_samples_per_second": 55.855,
+      "eval_steps_per_second": 13.971,
+      "step": 12000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.096191384084616e-06,
+      "loss": 0.5828,
+      "step": 12010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.087112442689184e-06,
+      "loss": 0.6129,
+      "step": 12020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.07803350129375e-06,
+      "loss": 0.6373,
+      "step": 12030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.068954559898316e-06,
+      "loss": 0.5529,
+      "step": 12040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.059875618502883e-06,
+      "loss": 0.5791,
+      "step": 12050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.050796677107449e-06,
+      "loss": 0.7039,
+      "step": 12060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.041717735712017e-06,
+      "loss": 0.6314,
+      "step": 12070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.032638794316584e-06,
+      "loss": 0.8344,
+      "step": 12080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02355985292115e-06,
+      "loss": 0.6568,
+      "step": 12090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014480911525718e-06,
+      "loss": 0.6854,
+      "step": 12100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005401970130283e-06,
+      "loss": 0.7133,
+      "step": 12110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996323028734851e-06,
+      "loss": 0.59,
+      "step": 12120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987244087339417e-06,
+      "loss": 0.7049,
+      "step": 12130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978165145943983e-06,
+      "loss": 0.5949,
+      "step": 12140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96908620454855e-06,
+      "loss": 0.6137,
+      "step": 12150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960007263153116e-06,
+      "loss": 0.8039,
+      "step": 12160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950928321757684e-06,
+      "loss": 0.5955,
+      "step": 12170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941849380362251e-06,
+      "loss": 0.558,
+      "step": 12180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932770438966817e-06,
+      "loss": 0.6453,
+      "step": 12190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923691497571384e-06,
+      "loss": 0.5607,
+      "step": 12200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91461255617595e-06,
+      "loss": 0.6797,
+      "step": 12210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905533614780518e-06,
+      "loss": 0.6105,
+      "step": 12220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.896454673385084e-06,
+      "loss": 0.607,
+      "step": 12230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.88737573198965e-06,
+      "loss": 0.7182,
+      "step": 12240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.878296790594217e-06,
+      "loss": 0.6537,
+      "step": 12250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.869217849198785e-06,
+      "loss": 0.8063,
+      "step": 12260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.86013890780335e-06,
+      "loss": 0.5801,
+      "step": 12270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.851059966407918e-06,
+      "loss": 0.707,
+      "step": 12280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.841981025012484e-06,
+      "loss": 0.7332,
+      "step": 12290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.832902083617051e-06,
+      "loss": 0.7797,
+      "step": 12300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.823823142221617e-06,
+      "loss": 0.7146,
+      "step": 12310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.814744200826185e-06,
+      "loss": 0.667,
+      "step": 12320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.80566525943075e-06,
+      "loss": 0.7193,
+      "step": 12330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.796586318035317e-06,
+      "loss": 0.7027,
+      "step": 12340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.787507376639884e-06,
+      "loss": 0.6092,
+      "step": 12350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.778428435244452e-06,
+      "loss": 0.6668,
+      "step": 12360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.769349493849017e-06,
+      "loss": 0.7051,
+      "step": 12370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.760270552453585e-06,
+      "loss": 0.7043,
+      "step": 12380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.751191611058153e-06,
+      "loss": 0.7746,
+      "step": 12390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.742112669662718e-06,
+      "loss": 0.5826,
+      "step": 12400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.733033728267284e-06,
+      "loss": 0.6295,
+      "step": 12410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.723954786871852e-06,
+      "loss": 0.6553,
+      "step": 12420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.714875845476418e-06,
+      "loss": 0.6766,
+      "step": 12430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705796904080985e-06,
+      "loss": 0.5719,
+      "step": 12440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.696717962685551e-06,
+      "loss": 0.6652,
+      "step": 12450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.687639021290119e-06,
+      "loss": 0.7088,
+      "step": 12460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.678560079894684e-06,
+      "loss": 0.6367,
+      "step": 12470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.669481138499252e-06,
+      "loss": 0.7176,
+      "step": 12480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.66040219710382e-06,
+      "loss": 0.8223,
+      "step": 12490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.651323255708385e-06,
+      "loss": 0.7406,
+      "step": 12500
+    },
+    {
+      "epoch": 0.57,
+      "eval_accuracy": 0.6284403669724771,
+      "eval_loss": 0.6642463803291321,
+      "eval_runtime": 74.8866,
+      "eval_samples_per_second": 55.31,
+      "eval_steps_per_second": 13.834,
+      "step": 12500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.642244314312951e-06,
+      "loss": 0.7641,
+      "step": 12510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.633165372917519e-06,
+      "loss": 0.6031,
+      "step": 12520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.624086431522085e-06,
+      "loss": 0.6365,
+      "step": 12530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.615007490126652e-06,
+      "loss": 0.5977,
+      "step": 12540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.605928548731218e-06,
+      "loss": 0.6783,
+      "step": 12550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.596849607335786e-06,
+      "loss": 0.7094,
+      "step": 12560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.587770665940353e-06,
+      "loss": 0.6758,
+      "step": 12570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.578691724544919e-06,
+      "loss": 0.7918,
+      "step": 12580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.569612783149486e-06,
+      "loss": 0.6793,
+      "step": 12590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.560533841754052e-06,
+      "loss": 0.7588,
+      "step": 12600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.551454900358618e-06,
+      "loss": 0.6875,
+      "step": 12610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.542375958963186e-06,
+      "loss": 0.5982,
+      "step": 12620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.533297017567752e-06,
+      "loss": 0.7289,
+      "step": 12630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.524218076172319e-06,
+      "loss": 0.641,
+      "step": 12640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.515139134776885e-06,
+      "loss": 0.6434,
+      "step": 12650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.506060193381452e-06,
+      "loss": 0.6051,
+      "step": 12660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.49698125198602e-06,
+      "loss": 0.7082,
+      "step": 12670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.487902310590586e-06,
+      "loss": 0.5811,
+      "step": 12680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.478823369195153e-06,
+      "loss": 0.573,
+      "step": 12690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.46974442779972e-06,
+      "loss": 0.6701,
+      "step": 12700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.460665486404285e-06,
+      "loss": 0.6824,
+      "step": 12710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.451586545008853e-06,
+      "loss": 0.5939,
+      "step": 12720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.442507603613419e-06,
+      "loss": 0.5402,
+      "step": 12730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.433428662217986e-06,
+      "loss": 0.7176,
+      "step": 12740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.424349720822554e-06,
+      "loss": 0.7102,
+      "step": 12750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.41527077942712e-06,
+      "loss": 0.7031,
+      "step": 12760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.406191838031687e-06,
+      "loss": 0.6232,
+      "step": 12770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.397112896636253e-06,
+      "loss": 0.7289,
+      "step": 12780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.38803395524082e-06,
+      "loss": 0.6281,
+      "step": 12790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.378955013845386e-06,
+      "loss": 0.6742,
+      "step": 12800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.369876072449952e-06,
+      "loss": 0.5912,
+      "step": 12810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.36079713105452e-06,
+      "loss": 0.6687,
+      "step": 12820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.351718189659085e-06,
+      "loss": 0.6396,
+      "step": 12830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.342639248263653e-06,
+      "loss": 0.6611,
+      "step": 12840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.33356030686822e-06,
+      "loss": 0.7992,
+      "step": 12850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.324481365472786e-06,
+      "loss": 0.5539,
+      "step": 12860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.315402424077354e-06,
+      "loss": 0.6166,
+      "step": 12870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.30632348268192e-06,
+      "loss": 0.6248,
+      "step": 12880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.297244541286487e-06,
+      "loss": 0.6965,
+      "step": 12890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.288165599891053e-06,
+      "loss": 0.6883,
+      "step": 12900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.279086658495619e-06,
+      "loss": 0.6459,
+      "step": 12910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.270007717100187e-06,
+      "loss": 0.6266,
+      "step": 12920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.260928775704754e-06,
+      "loss": 0.6408,
+      "step": 12930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.25184983430932e-06,
+      "loss": 0.6605,
+      "step": 12940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.242770892913888e-06,
+      "loss": 0.6299,
+      "step": 12950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.233691951518453e-06,
+      "loss": 0.7291,
+      "step": 12960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.224613010123021e-06,
+      "loss": 0.5949,
+      "step": 12970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.215534068727587e-06,
+      "loss": 0.6488,
+      "step": 12980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.206455127332154e-06,
+      "loss": 0.7426,
+      "step": 12990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.19737618593672e-06,
+      "loss": 0.6742,
+      "step": 13000
+    },
+    {
+      "epoch": 0.59,
+      "eval_accuracy": 0.6238532110091743,
+      "eval_loss": 0.6639795303344727,
+      "eval_runtime": 73.0379,
+      "eval_samples_per_second": 56.71,
+      "eval_steps_per_second": 14.184,
+      "step": 13000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.188297244541286e-06,
+      "loss": 0.6525,
+      "step": 13010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.179218303145854e-06,
+      "loss": 0.7201,
+      "step": 13020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.170139361750421e-06,
+      "loss": 0.5959,
+      "step": 13030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.161060420354987e-06,
+      "loss": 0.6248,
+      "step": 13040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.151981478959554e-06,
+      "loss": 0.7281,
+      "step": 13050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.14290253756412e-06,
+      "loss": 0.7008,
+      "step": 13060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.133823596168688e-06,
+      "loss": 0.6934,
+      "step": 13070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.124744654773254e-06,
+      "loss": 0.7391,
+      "step": 13080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.115665713377821e-06,
+      "loss": 0.6283,
+      "step": 13090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.106586771982387e-06,
+      "loss": 0.634,
+      "step": 13100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.097507830586955e-06,
+      "loss": 0.6631,
+      "step": 13110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.08842888919152e-06,
+      "loss": 0.6844,
+      "step": 13120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.079349947796088e-06,
+      "loss": 0.5783,
+      "step": 13130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.070271006400654e-06,
+      "loss": 0.657,
+      "step": 13140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.061192065005221e-06,
+      "loss": 0.7016,
+      "step": 13150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.052113123609789e-06,
+      "loss": 0.684,
+      "step": 13160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.043034182214355e-06,
+      "loss": 0.6338,
+      "step": 13170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.03395524081892e-06,
+      "loss": 0.6326,
+      "step": 13180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.024876299423488e-06,
+      "loss": 0.6875,
+      "step": 13190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.015797358028054e-06,
+      "loss": 0.6289,
+      "step": 13200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.006718416632622e-06,
+      "loss": 0.7225,
+      "step": 13210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.997639475237187e-06,
+      "loss": 0.7109,
+      "step": 13220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.988560533841755e-06,
+      "loss": 0.6988,
+      "step": 13230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.979481592446321e-06,
+      "loss": 0.6207,
+      "step": 13240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.970402651050888e-06,
+      "loss": 0.6271,
+      "step": 13250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.961323709655456e-06,
+      "loss": 0.6178,
+      "step": 13260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.952244768260022e-06,
+      "loss": 0.6682,
+      "step": 13270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.943165826864588e-06,
+      "loss": 0.6662,
+      "step": 13280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.934086885469155e-06,
+      "loss": 0.7354,
+      "step": 13290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.925007944073721e-06,
+      "loss": 0.6856,
+      "step": 13300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.915929002678289e-06,
+      "loss": 0.6242,
+      "step": 13310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.906850061282854e-06,
+      "loss": 0.6432,
+      "step": 13320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.897771119887422e-06,
+      "loss": 0.6941,
+      "step": 13330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.88869217849199e-06,
+      "loss": 0.7207,
+      "step": 13340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.879613237096555e-06,
+      "loss": 0.6047,
+      "step": 13350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.870534295701123e-06,
+      "loss": 0.6664,
+      "step": 13360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.861455354305689e-06,
+      "loss": 0.5578,
+      "step": 13370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.852376412910255e-06,
+      "loss": 0.6244,
+      "step": 13380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.843297471514822e-06,
+      "loss": 0.7566,
+      "step": 13390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.834218530119388e-06,
+      "loss": 0.666,
+      "step": 13400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.825139588723956e-06,
+      "loss": 0.5805,
+      "step": 13410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.816060647328521e-06,
+      "loss": 0.6395,
+      "step": 13420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.806981705933089e-06,
+      "loss": 0.7125,
+      "step": 13430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.797902764537656e-06,
+      "loss": 0.702,
+      "step": 13440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.788823823142222e-06,
+      "loss": 0.6018,
+      "step": 13450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.77974488174679e-06,
+      "loss": 0.7211,
+      "step": 13460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.770665940351356e-06,
+      "loss": 0.6887,
+      "step": 13470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.761586998955922e-06,
+      "loss": 0.666,
+      "step": 13480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.752508057560489e-06,
+      "loss": 0.6154,
+      "step": 13490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.743429116165055e-06,
+      "loss": 0.7082,
+      "step": 13500
+    },
+    {
+      "epoch": 0.61,
+      "eval_accuracy": 0.6262675036214389,
+      "eval_loss": 0.66704922914505,
+      "eval_runtime": 72.8876,
+      "eval_samples_per_second": 56.827,
+      "eval_steps_per_second": 14.214,
+      "step": 13500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.734350174769623e-06,
+      "loss": 0.6029,
+      "step": 13510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.72527123337419e-06,
+      "loss": 0.6395,
+      "step": 13520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.716192291978756e-06,
+      "loss": 0.6648,
+      "step": 13530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.707113350583323e-06,
+      "loss": 0.7348,
+      "step": 13540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.69803440918789e-06,
+      "loss": 0.583,
+      "step": 13550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.688955467792457e-06,
+      "loss": 0.707,
+      "step": 13560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.679876526397023e-06,
+      "loss": 0.625,
+      "step": 13570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.670797585001589e-06,
+      "loss": 0.6125,
+      "step": 13580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.661718643606156e-06,
+      "loss": 0.5121,
+      "step": 13590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.652639702210722e-06,
+      "loss": 0.7041,
+      "step": 13600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.64356076081529e-06,
+      "loss": 0.6555,
+      "step": 13610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.634481819419857e-06,
+      "loss": 0.5404,
+      "step": 13620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.625402878024423e-06,
+      "loss": 0.7129,
+      "step": 13630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.6163239366289896e-06,
+      "loss": 0.7699,
+      "step": 13640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.607244995233556e-06,
+      "loss": 0.6877,
+      "step": 13650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.598166053838123e-06,
+      "loss": 0.735,
+      "step": 13660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.5890871124426905e-06,
+      "loss": 0.7822,
+      "step": 13670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.580008171047256e-06,
+      "loss": 0.6674,
+      "step": 13680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.570929229651823e-06,
+      "loss": 0.6441,
+      "step": 13690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.561850288256391e-06,
+      "loss": 0.643,
+      "step": 13700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.5527713468609564e-06,
+      "loss": 0.6641,
+      "step": 13710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.543692405465524e-06,
+      "loss": 0.6915,
+      "step": 13720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.53461346407009e-06,
+      "loss": 0.7867,
+      "step": 13730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.5255345226746565e-06,
+      "loss": 0.7236,
+      "step": 13740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.516455581279224e-06,
+      "loss": 0.7305,
+      "step": 13750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.50737663988379e-06,
+      "loss": 0.6984,
+      "step": 13760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.4982976984883575e-06,
+      "loss": 0.7041,
+      "step": 13770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.489218757092923e-06,
+      "loss": 0.6264,
+      "step": 13780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.48013981569749e-06,
+      "loss": 0.6533,
+      "step": 13790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.4710608743020575e-06,
+      "loss": 0.7535,
+      "step": 13800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.461981932906623e-06,
+      "loss": 0.7496,
+      "step": 13810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.452902991511191e-06,
+      "loss": 0.7844,
+      "step": 13820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.443824050115757e-06,
+      "loss": 0.7195,
+      "step": 13830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.4347451087203235e-06,
+      "loss": 0.7344,
+      "step": 13840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.425666167324891e-06,
+      "loss": 0.623,
+      "step": 13850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.416587225929457e-06,
+      "loss": 0.6908,
+      "step": 13860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.407508284534024e-06,
+      "loss": 0.6525,
+      "step": 13870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.398429343138591e-06,
+      "loss": 0.6301,
+      "step": 13880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.389350401743157e-06,
+      "loss": 0.7547,
+      "step": 13890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.3802714603477245e-06,
+      "loss": 0.6453,
+      "step": 13900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.37119251895229e-06,
+      "loss": 0.8059,
+      "step": 13910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.362113577556858e-06,
+      "loss": 0.6408,
+      "step": 13920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.353034636161425e-06,
+      "loss": 0.6986,
+      "step": 13930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.3439556947659904e-06,
+      "loss": 0.5945,
+      "step": 13940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.334876753370558e-06,
+      "loss": 0.6809,
+      "step": 13950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.325797811975124e-06,
+      "loss": 0.7059,
+      "step": 13960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.316718870579691e-06,
+      "loss": 0.6023,
+      "step": 13970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.307639929184258e-06,
+      "loss": 0.5889,
+      "step": 13980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.298560987788824e-06,
+      "loss": 0.6645,
+      "step": 13990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.2894820463933915e-06,
+      "loss": 0.5768,
+      "step": 14000
+    },
+    {
+      "epoch": 0.64,
+      "eval_accuracy": 0.6269917914051183,
+      "eval_loss": 0.6619028449058533,
+      "eval_runtime": 73.4725,
+      "eval_samples_per_second": 56.375,
+      "eval_steps_per_second": 14.101,
+      "step": 14000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.280403104997957e-06,
+      "loss": 0.6783,
+      "step": 14010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.271324163602525e-06,
+      "loss": 0.59,
+      "step": 14020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.2622452222070916e-06,
+      "loss": 0.682,
+      "step": 14030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.253166280811657e-06,
+      "loss": 0.6873,
+      "step": 14040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.244087339416225e-06,
+      "loss": 0.7082,
+      "step": 14050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.235008398020792e-06,
+      "loss": 0.8027,
+      "step": 14060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.225929456625358e-06,
+      "loss": 0.6777,
+      "step": 14070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.216850515229925e-06,
+      "loss": 0.6312,
+      "step": 14080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.207771573834491e-06,
+      "loss": 0.6381,
+      "step": 14090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.1986926324390584e-06,
+      "loss": 0.5955,
+      "step": 14100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.189613691043625e-06,
+      "loss": 0.6518,
+      "step": 14110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.180534749648192e-06,
+      "loss": 0.5652,
+      "step": 14120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.1714558082527585e-06,
+      "loss": 0.7027,
+      "step": 14130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.162376866857324e-06,
+      "loss": 0.615,
+      "step": 14140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.153297925461892e-06,
+      "loss": 0.568,
+      "step": 14150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.144218984066459e-06,
+      "loss": 0.6701,
+      "step": 14160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.135140042671025e-06,
+      "loss": 0.7063,
+      "step": 14170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.126061101275592e-06,
+      "loss": 0.6428,
+      "step": 14180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.116982159880158e-06,
+      "loss": 0.6539,
+      "step": 14190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.107903218484725e-06,
+      "loss": 0.7867,
+      "step": 14200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.098824277089292e-06,
+      "loss": 0.6078,
+      "step": 14210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.089745335693859e-06,
+      "loss": 0.5559,
+      "step": 14220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.0806663942984255e-06,
+      "loss": 0.5851,
+      "step": 14230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.071587452902993e-06,
+      "loss": 0.5799,
+      "step": 14240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.062508511507559e-06,
+      "loss": 0.5656,
+      "step": 14250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.0534295701121256e-06,
+      "loss": 0.5715,
+      "step": 14260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.044350628716692e-06,
+      "loss": 0.7535,
+      "step": 14270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.035271687321259e-06,
+      "loss": 0.607,
+      "step": 14280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.0261927459258265e-06,
+      "loss": 0.6488,
+      "step": 14290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.017113804530392e-06,
+      "loss": 0.5721,
+      "step": 14300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.008034863134959e-06,
+      "loss": 0.7064,
+      "step": 14310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.998955921739526e-06,
+      "loss": 0.5836,
+      "step": 14320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.9898769803440924e-06,
+      "loss": 0.624,
+      "step": 14330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.98079803894866e-06,
+      "loss": 0.5609,
+      "step": 14340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.971719097553226e-06,
+      "loss": 0.6859,
+      "step": 14350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.9626401561577925e-06,
+      "loss": 0.652,
+      "step": 14360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.953561214762358e-06,
+      "loss": 0.8596,
+      "step": 14370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.944482273366926e-06,
+      "loss": 0.641,
+      "step": 14380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.9354033319714935e-06,
+      "loss": 0.7641,
+      "step": 14390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.926324390576059e-06,
+      "loss": 0.7662,
+      "step": 14400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.917245449180626e-06,
+      "loss": 0.698,
+      "step": 14410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.908166507785192e-06,
+      "loss": 0.6398,
+      "step": 14420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.899087566389759e-06,
+      "loss": 0.7016,
+      "step": 14430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.890008624994327e-06,
+      "loss": 0.6867,
+      "step": 14440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.880929683598893e-06,
+      "loss": 0.6895,
+      "step": 14450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.8718507422034595e-06,
+      "loss": 0.6836,
+      "step": 14460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.862771800808027e-06,
+      "loss": 0.6588,
+      "step": 14470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.853692859412593e-06,
+      "loss": 0.7346,
+      "step": 14480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.84461391801716e-06,
+      "loss": 0.6236,
+      "step": 14490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.835534976621726e-06,
+      "loss": 0.6082,
+      "step": 14500
+    },
+    {
+      "epoch": 0.66,
+      "eval_accuracy": 0.6154031868662482,
+      "eval_loss": 0.6647462248802185,
+      "eval_runtime": 73.4637,
+      "eval_samples_per_second": 56.382,
+      "eval_steps_per_second": 14.102,
+      "step": 14500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.826456035226293e-06,
+      "loss": 0.7277,
+      "step": 14510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.8173770938308605e-06,
+      "loss": 0.7398,
+      "step": 14520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.808298152435426e-06,
+      "loss": 0.702,
+      "step": 14530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.799219211039994e-06,
+      "loss": 0.6396,
+      "step": 14540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.79014026964456e-06,
+      "loss": 0.6211,
+      "step": 14550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.7810613282491264e-06,
+      "loss": 0.716,
+      "step": 14560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.771982386853694e-06,
+      "loss": 0.6299,
+      "step": 14570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.76290344545826e-06,
+      "loss": 0.6096,
+      "step": 14580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.753824504062827e-06,
+      "loss": 0.6121,
+      "step": 14590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.744745562667393e-06,
+      "loss": 0.65,
+      "step": 14600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.73566662127196e-06,
+      "loss": 0.7004,
+      "step": 14610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.7265876798765275e-06,
+      "loss": 0.5912,
+      "step": 14620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.717508738481093e-06,
+      "loss": 0.6816,
+      "step": 14630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.708429797085661e-06,
+      "loss": 0.7867,
+      "step": 14640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6993508556902276e-06,
+      "loss": 0.6977,
+      "step": 14650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.690271914294793e-06,
+      "loss": 0.7098,
+      "step": 14660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.681192972899361e-06,
+      "loss": 0.749,
+      "step": 14670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.672114031503927e-06,
+      "loss": 0.5707,
+      "step": 14680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.663035090108494e-06,
+      "loss": 0.7465,
+      "step": 14690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.653956148713061e-06,
+      "loss": 0.6639,
+      "step": 14700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.644877207317627e-06,
+      "loss": 0.7758,
+      "step": 14710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6357982659221944e-06,
+      "loss": 0.6439,
+      "step": 14720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.62671932452676e-06,
+      "loss": 0.5809,
+      "step": 14730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.617640383131327e-06,
+      "loss": 0.6887,
+      "step": 14740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6085614417358945e-06,
+      "loss": 0.6861,
+      "step": 14750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.59948250034046e-06,
+      "loss": 0.6762,
+      "step": 14760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.590403558945028e-06,
+      "loss": 0.698,
+      "step": 14770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.581324617549594e-06,
+      "loss": 0.7,
+      "step": 14780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.5722456761541605e-06,
+      "loss": 0.652,
+      "step": 14790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.563166734758728e-06,
+      "loss": 0.6787,
+      "step": 14800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.554087793363294e-06,
+      "loss": 0.6215,
+      "step": 14810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.545008851967861e-06,
+      "loss": 0.8277,
+      "step": 14820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.535929910572428e-06,
+      "loss": 0.6615,
+      "step": 14830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.526850969176994e-06,
+      "loss": 0.6617,
+      "step": 14840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.5177720277815615e-06,
+      "loss": 0.6766,
+      "step": 14850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.508693086386127e-06,
+      "loss": 0.7777,
+      "step": 14860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.499614144990695e-06,
+      "loss": 0.7141,
+      "step": 14870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.4905352035952616e-06,
+      "loss": 0.6637,
+      "step": 14880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.481456262199827e-06,
+      "loss": 0.6723,
+      "step": 14890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.472377320804395e-06,
+      "loss": 0.6975,
+      "step": 14900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.463298379408961e-06,
+      "loss": 0.7087,
+      "step": 14910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.454219438013528e-06,
+      "loss": 0.6736,
+      "step": 14920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.445140496618095e-06,
+      "loss": 0.6164,
+      "step": 14930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.436061555222661e-06,
+      "loss": 0.6373,
+      "step": 14940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.4269826138272284e-06,
+      "loss": 0.707,
+      "step": 14950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.417903672431794e-06,
+      "loss": 0.6229,
+      "step": 14960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.408824731036362e-06,
+      "loss": 0.734,
+      "step": 14970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.3997457896409285e-06,
+      "loss": 0.6488,
+      "step": 14980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.390666848245494e-06,
+      "loss": 0.6781,
+      "step": 14990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.381587906850062e-06,
+      "loss": 0.616,
+      "step": 15000
+    },
+    {
+      "epoch": 0.68,
+      "eval_accuracy": 0.623128923225495,
+      "eval_loss": 0.6619547009468079,
+      "eval_runtime": 73.1853,
+      "eval_samples_per_second": 56.596,
+      "eval_steps_per_second": 14.156,
+      "step": 15000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.3725089654546295e-06,
+      "loss": 0.6363,
+      "step": 15010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.363430024059195e-06,
+      "loss": 0.7072,
+      "step": 15020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.354351082663762e-06,
+      "loss": 0.6193,
+      "step": 15030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.345272141268328e-06,
+      "loss": 0.6348,
+      "step": 15040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.336193199872895e-06,
+      "loss": 0.6953,
+      "step": 15050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.327114258477463e-06,
+      "loss": 0.5598,
+      "step": 15060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.318035317082029e-06,
+      "loss": 0.6547,
+      "step": 15070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.3089563756865955e-06,
+      "loss": 0.6287,
+      "step": 15080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.299877434291161e-06,
+      "loss": 0.5846,
+      "step": 15090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.290798492895729e-06,
+      "loss": 0.7217,
+      "step": 15100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.281719551500296e-06,
+      "loss": 0.7309,
+      "step": 15110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.272640610104862e-06,
+      "loss": 0.6762,
+      "step": 15120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.263561668709429e-06,
+      "loss": 0.7738,
+      "step": 15130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.254482727313995e-06,
+      "loss": 0.6906,
+      "step": 15140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.245403785918562e-06,
+      "loss": 0.6582,
+      "step": 15150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.236324844523129e-06,
+      "loss": 0.6228,
+      "step": 15160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.227245903127696e-06,
+      "loss": 0.7645,
+      "step": 15170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.2181669617322625e-06,
+      "loss": 0.6684,
+      "step": 15180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.20908802033683e-06,
+      "loss": 0.7598,
+      "step": 15190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.200009078941396e-06,
+      "loss": 0.6254,
+      "step": 15200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1909301375459625e-06,
+      "loss": 0.6281,
+      "step": 15210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.181851196150529e-06,
+      "loss": 0.6301,
+      "step": 15220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.172772254755096e-06,
+      "loss": 0.6805,
+      "step": 15230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1636933133596635e-06,
+      "loss": 0.7369,
+      "step": 15240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.154614371964229e-06,
+      "loss": 0.6443,
+      "step": 15250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.145535430568796e-06,
+      "loss": 0.6247,
+      "step": 15260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.136456489173363e-06,
+      "loss": 0.6518,
+      "step": 15270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.127377547777929e-06,
+      "loss": 0.6061,
+      "step": 15280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.118298606382497e-06,
+      "loss": 0.6303,
+      "step": 15290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.109219664987063e-06,
+      "loss": 0.5945,
+      "step": 15300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1001407235916295e-06,
+      "loss": 0.5559,
+      "step": 15310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.091061782196196e-06,
+      "loss": 0.6348,
+      "step": 15320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.081982840800763e-06,
+      "loss": 0.5775,
+      "step": 15330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.0729038994053304e-06,
+      "loss": 0.6215,
+      "step": 15340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.063824958009896e-06,
+      "loss": 0.641,
+      "step": 15350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.054746016614463e-06,
+      "loss": 0.6947,
+      "step": 15360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.04566707521903e-06,
+      "loss": 0.698,
+      "step": 15370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.036588133823596e-06,
+      "loss": 0.7543,
+      "step": 15380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.027509192428164e-06,
+      "loss": 0.7227,
+      "step": 15390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.01843025103273e-06,
+      "loss": 0.6172,
+      "step": 15400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.0093513096372965e-06,
+      "loss": 0.7646,
+      "step": 15410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.000272368241864e-06,
+      "loss": 0.7711,
+      "step": 15420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.99119342684643e-06,
+      "loss": 0.6197,
+      "step": 15430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.982114485450997e-06,
+      "loss": 0.6904,
+      "step": 15440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.973035544055563e-06,
+      "loss": 0.7703,
+      "step": 15450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.96395660266013e-06,
+      "loss": 0.607,
+      "step": 15460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9548776612646975e-06,
+      "loss": 0.877,
+      "step": 15470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.945798719869263e-06,
+      "loss": 0.6273,
+      "step": 15480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.936719778473831e-06,
+      "loss": 0.7051,
+      "step": 15490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.927640837078397e-06,
+      "loss": 0.7094,
+      "step": 15500
+    },
+    {
+      "epoch": 0.7,
+      "eval_accuracy": 0.6284403669724771,
+      "eval_loss": 0.6594442129135132,
+      "eval_runtime": 73.777,
+      "eval_samples_per_second": 56.142,
+      "eval_steps_per_second": 14.042,
+      "step": 15500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9185618956829634e-06,
+      "loss": 0.6646,
+      "step": 15510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.909482954287531e-06,
+      "loss": 0.6553,
+      "step": 15520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.900404012892097e-06,
+      "loss": 0.7227,
+      "step": 15530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.891325071496664e-06,
+      "loss": 0.6779,
+      "step": 15540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.88224613010123e-06,
+      "loss": 0.6443,
+      "step": 15550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.873167188705797e-06,
+      "loss": 0.7105,
+      "step": 15560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.8640882473103644e-06,
+      "loss": 0.6504,
+      "step": 15570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.85500930591493e-06,
+      "loss": 0.5859,
+      "step": 15580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.845930364519498e-06,
+      "loss": 0.5393,
+      "step": 15590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.8368514231240645e-06,
+      "loss": 0.6582,
+      "step": 15600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.82777248172863e-06,
+      "loss": 0.6428,
+      "step": 15610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.818693540333198e-06,
+      "loss": 0.6365,
+      "step": 15620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.809614598937764e-06,
+      "loss": 0.6689,
+      "step": 15630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.800535657542331e-06,
+      "loss": 0.6746,
+      "step": 15640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.791456716146898e-06,
+      "loss": 0.7037,
+      "step": 15650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.782377774751464e-06,
+      "loss": 0.6582,
+      "step": 15660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.773298833356031e-06,
+      "loss": 0.5789,
+      "step": 15670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.764219891960597e-06,
+      "loss": 0.6115,
+      "step": 15680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.755140950565165e-06,
+      "loss": 0.6639,
+      "step": 15690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.7460620091697315e-06,
+      "loss": 0.7234,
+      "step": 15700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.736983067774297e-06,
+      "loss": 0.6412,
+      "step": 15710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.727904126378865e-06,
+      "loss": 0.6016,
+      "step": 15720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.718825184983431e-06,
+      "loss": 0.6486,
+      "step": 15730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.709746243587998e-06,
+      "loss": 0.6322,
+      "step": 15740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.700667302192565e-06,
+      "loss": 0.649,
+      "step": 15750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.691588360797131e-06,
+      "loss": 0.65,
+      "step": 15760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.682509419401698e-06,
+      "loss": 0.6649,
+      "step": 15770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.673430478006265e-06,
+      "loss": 0.6457,
+      "step": 15780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.664351536610832e-06,
+      "loss": 0.6773,
+      "step": 15790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.6552725952153985e-06,
+      "loss": 0.5795,
+      "step": 15800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.646193653819964e-06,
+      "loss": 0.5668,
+      "step": 15810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.637114712424532e-06,
+      "loss": 0.5938,
+      "step": 15820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.6280357710290985e-06,
+      "loss": 0.7176,
+      "step": 15830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.618956829633665e-06,
+      "loss": 0.6992,
+      "step": 15840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.609877888238232e-06,
+      "loss": 0.6344,
+      "step": 15850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.600798946842798e-06,
+      "loss": 0.5469,
+      "step": 15860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.591720005447365e-06,
+      "loss": 0.6225,
+      "step": 15870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.582641064051932e-06,
+      "loss": 0.6426,
+      "step": 15880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.573562122656499e-06,
+      "loss": 0.6982,
+      "step": 15890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.564483181261065e-06,
+      "loss": 0.6342,
+      "step": 15900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.555404239865631e-06,
+      "loss": 0.6494,
+      "step": 15910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.546325298470199e-06,
+      "loss": 0.8711,
+      "step": 15920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5372463570747655e-06,
+      "loss": 0.6488,
+      "step": 15930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528167415679332e-06,
+      "loss": 0.6447,
+      "step": 15940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.519088474283899e-06,
+      "loss": 0.6604,
+      "step": 15950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5100095328884664e-06,
+      "loss": 0.6629,
+      "step": 15960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.500930591493032e-06,
+      "loss": 0.7656,
+      "step": 15970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.491851650097599e-06,
+      "loss": 0.6625,
+      "step": 15980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.482772708702166e-06,
+      "loss": 0.6834,
+      "step": 15990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.473693767306732e-06,
+      "loss": 0.5994,
+      "step": 16000
+    },
+    {
+      "epoch": 0.73,
+      "eval_accuracy": 0.6253017865765331,
+      "eval_loss": 0.6599798798561096,
+      "eval_runtime": 73.4934,
+      "eval_samples_per_second": 56.359,
+      "eval_steps_per_second": 14.097,
+      "step": 16000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4646148259113e-06,
+      "loss": 0.4986,
+      "step": 16010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.455535884515866e-06,
+      "loss": 0.6568,
+      "step": 16020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4464569431204325e-06,
+      "loss": 0.7852,
+      "step": 16030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.437378001724999e-06,
+      "loss": 0.6438,
+      "step": 16040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.428299060329566e-06,
+      "loss": 0.6207,
+      "step": 16050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.419220118934133e-06,
+      "loss": 0.7109,
+      "step": 16060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.410141177538699e-06,
+      "loss": 0.5861,
+      "step": 16070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.401062236143266e-06,
+      "loss": 0.698,
+      "step": 16080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.391983294747833e-06,
+      "loss": 0.6994,
+      "step": 16090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.382904353352399e-06,
+      "loss": 0.7209,
+      "step": 16100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.373825411956967e-06,
+      "loss": 0.6584,
+      "step": 16110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.364746470561533e-06,
+      "loss": 0.6723,
+      "step": 16120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.3556675291660994e-06,
+      "loss": 0.609,
+      "step": 16130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.346588587770667e-06,
+      "loss": 0.6607,
+      "step": 16140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.337509646375233e-06,
+      "loss": 0.6723,
+      "step": 16150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.3284307049798e-06,
+      "loss": 0.6447,
+      "step": 16160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.319351763584366e-06,
+      "loss": 0.732,
+      "step": 16170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.310272822188933e-06,
+      "loss": 0.6377,
+      "step": 16180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.3011938807935005e-06,
+      "loss": 0.6717,
+      "step": 16190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.292114939398066e-06,
+      "loss": 0.6777,
+      "step": 16200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.283035998002634e-06,
+      "loss": 0.7375,
+      "step": 16210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2739570566072e-06,
+      "loss": 0.6996,
+      "step": 16220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.264878115211766e-06,
+      "loss": 0.8133,
+      "step": 16230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.255799173816334e-06,
+      "loss": 0.6363,
+      "step": 16240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2467202324209e-06,
+      "loss": 0.7176,
+      "step": 16250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.237641291025467e-06,
+      "loss": 0.7248,
+      "step": 16260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.228562349630033e-06,
+      "loss": 0.7535,
+      "step": 16270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2194834082346e-06,
+      "loss": 0.6496,
+      "step": 16280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.210404466839167e-06,
+      "loss": 0.6551,
+      "step": 16290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.201325525443733e-06,
+      "loss": 0.7176,
+      "step": 16300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.192246584048301e-06,
+      "loss": 0.7488,
+      "step": 16310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1831676426528675e-06,
+      "loss": 0.5922,
+      "step": 16320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.174088701257433e-06,
+      "loss": 0.6285,
+      "step": 16330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.165009759862001e-06,
+      "loss": 0.7195,
+      "step": 16340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.155930818466567e-06,
+      "loss": 0.5939,
+      "step": 16350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.146851877071134e-06,
+      "loss": 0.6443,
+      "step": 16360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.137772935675701e-06,
+      "loss": 0.7504,
+      "step": 16370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.128693994280267e-06,
+      "loss": 0.5725,
+      "step": 16380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.119615052884834e-06,
+      "loss": 0.6207,
+      "step": 16390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1105361114894e-06,
+      "loss": 0.6525,
+      "step": 16400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.101457170093968e-06,
+      "loss": 0.6426,
+      "step": 16410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0923782286985345e-06,
+      "loss": 0.6492,
+      "step": 16420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0832992873031e-06,
+      "loss": 0.6195,
+      "step": 16430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.074220345907668e-06,
+      "loss": 0.7227,
+      "step": 16440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.065141404512234e-06,
+      "loss": 0.6459,
+      "step": 16450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.056062463116801e-06,
+      "loss": 0.7871,
+      "step": 16460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.046983521721368e-06,
+      "loss": 0.709,
+      "step": 16470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.037904580325934e-06,
+      "loss": 0.7148,
+      "step": 16480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.028825638930501e-06,
+      "loss": 0.7053,
+      "step": 16490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.019746697535067e-06,
+      "loss": 0.6969,
+      "step": 16500
+    },
+    {
+      "epoch": 0.75,
+      "eval_accuracy": 0.6332689521970063,
+      "eval_loss": 0.6564537286758423,
+      "eval_runtime": 73.2629,
+      "eval_samples_per_second": 56.536,
+      "eval_steps_per_second": 14.141,
+      "step": 16500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.010667756139635e-06,
+      "loss": 0.6316,
+      "step": 16510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0015888147442014e-06,
+      "loss": 0.649,
+      "step": 16520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.992509873348768e-06,
+      "loss": 0.6922,
+      "step": 16530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.983430931953335e-06,
+      "loss": 0.7539,
+      "step": 16540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9743519905579015e-06,
+      "loss": 0.7371,
+      "step": 16550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.965273049162468e-06,
+      "loss": 0.7238,
+      "step": 16560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.956194107767035e-06,
+      "loss": 0.6398,
+      "step": 16570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.947115166371602e-06,
+      "loss": 0.6449,
+      "step": 16580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.938036224976168e-06,
+      "loss": 0.6453,
+      "step": 16590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.928957283580735e-06,
+      "loss": 0.6664,
+      "step": 16600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.919878342185302e-06,
+      "loss": 0.6332,
+      "step": 16610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.910799400789868e-06,
+      "loss": 0.6787,
+      "step": 16620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.901720459394435e-06,
+      "loss": 0.6451,
+      "step": 16630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.892641517999002e-06,
+      "loss": 0.6668,
+      "step": 16640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8835625766035685e-06,
+      "loss": 0.7297,
+      "step": 16650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.874483635208135e-06,
+      "loss": 0.6457,
+      "step": 16660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.865404693812702e-06,
+      "loss": 0.709,
+      "step": 16670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8563257524172686e-06,
+      "loss": 0.6281,
+      "step": 16680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.847246811021835e-06,
+      "loss": 0.7352,
+      "step": 16690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.838167869626402e-06,
+      "loss": 0.6623,
+      "step": 16700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.829088928230969e-06,
+      "loss": 0.6873,
+      "step": 16710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.820009986835535e-06,
+      "loss": 0.6742,
+      "step": 16720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810931045440102e-06,
+      "loss": 0.5553,
+      "step": 16730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.801852104044669e-06,
+      "loss": 0.652,
+      "step": 16740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7927731626492354e-06,
+      "loss": 0.6018,
+      "step": 16750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.783694221253802e-06,
+      "loss": 0.6379,
+      "step": 16760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.774615279858369e-06,
+      "loss": 0.6402,
+      "step": 16770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7655363384629355e-06,
+      "loss": 0.6189,
+      "step": 16780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.756457397067502e-06,
+      "loss": 0.7432,
+      "step": 16790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.747378455672069e-06,
+      "loss": 0.7102,
+      "step": 16800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.738299514276636e-06,
+      "loss": 0.6535,
+      "step": 16810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.729220572881202e-06,
+      "loss": 0.6627,
+      "step": 16820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.720141631485769e-06,
+      "loss": 0.5818,
+      "step": 16830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.711062690090336e-06,
+      "loss": 0.635,
+      "step": 16840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.701983748694902e-06,
+      "loss": 0.7068,
+      "step": 16850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.692904807299469e-06,
+      "loss": 0.6361,
+      "step": 16860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.683825865904036e-06,
+      "loss": 0.5547,
+      "step": 16870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6747469245086025e-06,
+      "loss": 0.5598,
+      "step": 16880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.665667983113169e-06,
+      "loss": 0.6877,
+      "step": 16890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.656589041717736e-06,
+      "loss": 0.6459,
+      "step": 16900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.647510100322303e-06,
+      "loss": 0.575,
+      "step": 16910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.638431158926869e-06,
+      "loss": 0.6687,
+      "step": 16920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.629352217531437e-06,
+      "loss": 0.6676,
+      "step": 16930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.620273276136003e-06,
+      "loss": 0.7098,
+      "step": 16940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.611194334740569e-06,
+      "loss": 0.7119,
+      "step": 16950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.602115393345136e-06,
+      "loss": 0.5803,
+      "step": 16960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.593036451949703e-06,
+      "loss": 0.6615,
+      "step": 16970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.58395751055427e-06,
+      "loss": 0.6672,
+      "step": 16980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.574878569158836e-06,
+      "loss": 0.7334,
+      "step": 16990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.565799627763403e-06,
+      "loss": 0.7033,
+      "step": 17000
+    },
+    {
+      "epoch": 0.77,
+      "eval_accuracy": 0.6286817962337036,
+      "eval_loss": 0.6592499613761902,
+      "eval_runtime": 72.898,
+      "eval_samples_per_second": 56.819,
+      "eval_steps_per_second": 14.212,
+      "step": 17000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5567206863679695e-06,
+      "loss": 0.6502,
+      "step": 17010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.547641744972536e-06,
+      "loss": 0.7656,
+      "step": 17020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.538562803577104e-06,
+      "loss": 0.6209,
+      "step": 17030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.52948386218167e-06,
+      "loss": 0.624,
+      "step": 17040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.520404920786236e-06,
+      "loss": 0.6344,
+      "step": 17050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.511325979390803e-06,
+      "loss": 0.6299,
+      "step": 17060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5022470379953706e-06,
+      "loss": 0.6219,
+      "step": 17070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.493168096599937e-06,
+      "loss": 0.6348,
+      "step": 17080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.484089155204503e-06,
+      "loss": 0.7537,
+      "step": 17090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47501021380907e-06,
+      "loss": 0.5959,
+      "step": 17100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4659312724136365e-06,
+      "loss": 0.7732,
+      "step": 17110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.456852331018204e-06,
+      "loss": 0.6707,
+      "step": 17120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.447773389622771e-06,
+      "loss": 0.6881,
+      "step": 17130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.438694448227337e-06,
+      "loss": 0.6984,
+      "step": 17140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.429615506831903e-06,
+      "loss": 0.5365,
+      "step": 17150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.420536565436471e-06,
+      "loss": 0.6693,
+      "step": 17160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4114576240410375e-06,
+      "loss": 0.6051,
+      "step": 17170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.402378682645604e-06,
+      "loss": 0.6814,
+      "step": 17180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.39329974125017e-06,
+      "loss": 0.7059,
+      "step": 17190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.384220799854737e-06,
+      "loss": 0.6785,
+      "step": 17200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.375141858459304e-06,
+      "loss": 0.5293,
+      "step": 17210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.366062917063871e-06,
+      "loss": 0.7861,
+      "step": 17220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.356983975668438e-06,
+      "loss": 0.6998,
+      "step": 17230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3479050342730035e-06,
+      "loss": 0.6299,
+      "step": 17240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.338826092877571e-06,
+      "loss": 0.5969,
+      "step": 17250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.329747151482138e-06,
+      "loss": 0.727,
+      "step": 17260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3206682100867045e-06,
+      "loss": 0.5861,
+      "step": 17270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.311589268691271e-06,
+      "loss": 0.6994,
+      "step": 17280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.302510327295837e-06,
+      "loss": 0.6826,
+      "step": 17290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2934313859004046e-06,
+      "loss": 0.5617,
+      "step": 17300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.284352444504971e-06,
+      "loss": 0.6279,
+      "step": 17310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.275273503109538e-06,
+      "loss": 0.5873,
+      "step": 17320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.266194561714105e-06,
+      "loss": 0.742,
+      "step": 17330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.257115620318671e-06,
+      "loss": 0.7314,
+      "step": 17340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.248036678923238e-06,
+      "loss": 0.551,
+      "step": 17350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.238957737527805e-06,
+      "loss": 0.7438,
+      "step": 17360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2298787961323714e-06,
+      "loss": 0.6326,
+      "step": 17370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220799854736938e-06,
+      "loss": 0.6129,
+      "step": 17380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.211720913341505e-06,
+      "loss": 0.802,
+      "step": 17390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2026419719460715e-06,
+      "loss": 0.6312,
+      "step": 17400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.193563030550638e-06,
+      "loss": 0.6674,
+      "step": 17410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.184484089155205e-06,
+      "loss": 0.8195,
+      "step": 17420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.175405147759772e-06,
+      "loss": 0.6348,
+      "step": 17430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.166326206364338e-06,
+      "loss": 0.6832,
+      "step": 17440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.157247264968905e-06,
+      "loss": 0.5834,
+      "step": 17450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.148168323573472e-06,
+      "loss": 0.6127,
+      "step": 17460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.139089382178038e-06,
+      "loss": 0.6381,
+      "step": 17470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.130010440782605e-06,
+      "loss": 0.5922,
+      "step": 17480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.120931499387172e-06,
+      "loss": 0.6482,
+      "step": 17490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1118525579917385e-06,
+      "loss": 0.6711,
+      "step": 17500
+    },
+    {
+      "epoch": 0.79,
+      "eval_accuracy": 0.6301303718010622,
+      "eval_loss": 0.6614124178886414,
+      "eval_runtime": 72.5414,
+      "eval_samples_per_second": 57.098,
+      "eval_steps_per_second": 14.281,
+      "step": 17500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.102773616596305e-06,
+      "loss": 0.5473,
+      "step": 17510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.093694675200872e-06,
+      "loss": 0.6637,
+      "step": 17520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.084615733805439e-06,
+      "loss": 0.724,
+      "step": 17530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075536792410005e-06,
+      "loss": 0.615,
+      "step": 17540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.066457851014572e-06,
+      "loss": 0.6775,
+      "step": 17550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.057378909619139e-06,
+      "loss": 0.6363,
+      "step": 17560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.048299968223705e-06,
+      "loss": 0.7193,
+      "step": 17570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.039221026828272e-06,
+      "loss": 0.6891,
+      "step": 17580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.030142085432839e-06,
+      "loss": 0.7309,
+      "step": 17590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0210631440374055e-06,
+      "loss": 0.8832,
+      "step": 17600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.011984202641972e-06,
+      "loss": 0.651,
+      "step": 17610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.002905261246539e-06,
+      "loss": 0.7375,
+      "step": 17620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9938263198511055e-06,
+      "loss": 0.785,
+      "step": 17630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.984747378455672e-06,
+      "loss": 0.7391,
+      "step": 17640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.975668437060239e-06,
+      "loss": 0.6408,
+      "step": 17650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.966589495664806e-06,
+      "loss": 0.5838,
+      "step": 17660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.957510554269372e-06,
+      "loss": 0.5918,
+      "step": 17670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.948431612873939e-06,
+      "loss": 0.5525,
+      "step": 17680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.939352671478506e-06,
+      "loss": 0.7316,
+      "step": 17690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.930273730083073e-06,
+      "loss": 0.7426,
+      "step": 17700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.921194788687639e-06,
+      "loss": 0.6811,
+      "step": 17710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.912115847292206e-06,
+      "loss": 0.6797,
+      "step": 17720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9030369058967725e-06,
+      "loss": 0.5119,
+      "step": 17730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.893957964501339e-06,
+      "loss": 0.6432,
+      "step": 17740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.884879023105907e-06,
+      "loss": 0.6939,
+      "step": 17750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.875800081710473e-06,
+      "loss": 0.6508,
+      "step": 17760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.866721140315039e-06,
+      "loss": 0.5836,
+      "step": 17770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.857642198919606e-06,
+      "loss": 0.6607,
+      "step": 17780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8485632575241735e-06,
+      "loss": 0.7164,
+      "step": 17790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.83948431612874e-06,
+      "loss": 0.6176,
+      "step": 17800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.830405374733306e-06,
+      "loss": 0.6244,
+      "step": 17810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.821326433337873e-06,
+      "loss": 0.6908,
+      "step": 17820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8122474919424395e-06,
+      "loss": 0.7539,
+      "step": 17830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8031685505470066e-06,
+      "loss": 0.7873,
+      "step": 17840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7940896091515733e-06,
+      "loss": 0.6102,
+      "step": 17850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.78501066775614e-06,
+      "loss": 0.5908,
+      "step": 17860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7759317263607067e-06,
+      "loss": 0.5654,
+      "step": 17870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.766852784965274e-06,
+      "loss": 0.7463,
+      "step": 17880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75777384356984e-06,
+      "loss": 0.6754,
+      "step": 17890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7486949021744068e-06,
+      "loss": 0.7715,
+      "step": 17900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7396159607789735e-06,
+      "loss": 0.6137,
+      "step": 17910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.73053701938354e-06,
+      "loss": 0.593,
+      "step": 17920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7214580779881073e-06,
+      "loss": 0.6748,
+      "step": 17930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7123791365926735e-06,
+      "loss": 0.618,
+      "step": 17940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7033001951972402e-06,
+      "loss": 0.5645,
+      "step": 17950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.694221253801807e-06,
+      "loss": 0.684,
+      "step": 17960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.685142312406374e-06,
+      "loss": 0.634,
+      "step": 17970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6760633710109408e-06,
+      "loss": 0.6729,
+      "step": 17980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.666984429615507e-06,
+      "loss": 0.7057,
+      "step": 17990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6579054882200737e-06,
+      "loss": 0.6484,
+      "step": 18000
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.6308546595847416,
+      "eval_loss": 0.659805417060852,
+      "eval_runtime": 73.2829,
+      "eval_samples_per_second": 56.521,
+      "eval_steps_per_second": 14.137,
+      "step": 18000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6488265468246404e-06,
+      "loss": 0.6398,
+      "step": 18010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6397476054292075e-06,
+      "loss": 0.5816,
+      "step": 18020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6306686640337742e-06,
+      "loss": 0.6373,
+      "step": 18030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6215897226383405e-06,
+      "loss": 0.6348,
+      "step": 18040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.612510781242907e-06,
+      "loss": 0.6752,
+      "step": 18050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6034318398474743e-06,
+      "loss": 0.7084,
+      "step": 18060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.594352898452041e-06,
+      "loss": 0.5984,
+      "step": 18070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5852739570566077e-06,
+      "loss": 0.5811,
+      "step": 18080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.576195015661174e-06,
+      "loss": 0.7691,
+      "step": 18090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5671160742657407e-06,
+      "loss": 0.6145,
+      "step": 18100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.558037132870308e-06,
+      "loss": 0.6488,
+      "step": 18110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5489581914748745e-06,
+      "loss": 0.6141,
+      "step": 18120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.539879250079441e-06,
+      "loss": 0.8035,
+      "step": 18130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5308003086840075e-06,
+      "loss": 0.6059,
+      "step": 18140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.521721367288574e-06,
+      "loss": 0.6791,
+      "step": 18150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5126424258931413e-06,
+      "loss": 0.5494,
+      "step": 18160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.503563484497708e-06,
+      "loss": 0.6541,
+      "step": 18170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4944845431022747e-06,
+      "loss": 0.6811,
+      "step": 18180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.485405601706841e-06,
+      "loss": 0.6953,
+      "step": 18190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.476326660311408e-06,
+      "loss": 0.7562,
+      "step": 18200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4672477189159748e-06,
+      "loss": 0.649,
+      "step": 18210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4581687775205415e-06,
+      "loss": 0.692,
+      "step": 18220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.449089836125108e-06,
+      "loss": 0.633,
+      "step": 18230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4400108947296744e-06,
+      "loss": 0.6006,
+      "step": 18240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4309319533342416e-06,
+      "loss": 0.526,
+      "step": 18250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4218530119388082e-06,
+      "loss": 0.6,
+      "step": 18260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.412774070543375e-06,
+      "loss": 0.7064,
+      "step": 18270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4036951291479412e-06,
+      "loss": 0.7047,
+      "step": 18280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3946161877525088e-06,
+      "loss": 0.7266,
+      "step": 18290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.385537246357075e-06,
+      "loss": 0.6955,
+      "step": 18300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3764583049616417e-06,
+      "loss": 0.5686,
+      "step": 18310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3673793635662084e-06,
+      "loss": 0.5816,
+      "step": 18320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3583004221707747e-06,
+      "loss": 0.6555,
+      "step": 18330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3492214807753422e-06,
+      "loss": 0.5975,
+      "step": 18340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3401425393799085e-06,
+      "loss": 0.6977,
+      "step": 18350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.331063597984475e-06,
+      "loss": 0.6643,
+      "step": 18360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.321984656589042e-06,
+      "loss": 0.7504,
+      "step": 18370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.312905715193609e-06,
+      "loss": 0.6539,
+      "step": 18380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3038267737981757e-06,
+      "loss": 0.7016,
+      "step": 18390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.294747832402742e-06,
+      "loss": 0.5623,
+      "step": 18400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2856688910073087e-06,
+      "loss": 0.5807,
+      "step": 18410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2765899496118754e-06,
+      "loss": 0.701,
+      "step": 18420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2675110082164425e-06,
+      "loss": 0.6535,
+      "step": 18430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2584320668210088e-06,
+      "loss": 0.6252,
+      "step": 18440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2493531254255755e-06,
+      "loss": 0.576,
+      "step": 18450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.240274184030142e-06,
+      "loss": 0.7225,
+      "step": 18460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2311952426347093e-06,
+      "loss": 0.574,
+      "step": 18470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.222116301239276e-06,
+      "loss": 0.6803,
+      "step": 18480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2130373598438423e-06,
+      "loss": 0.6945,
+      "step": 18490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.203958418448409e-06,
+      "loss": 0.5729,
+      "step": 18500
+    },
+    {
+      "epoch": 0.84,
+      "eval_accuracy": 0.6269917914051183,
+      "eval_loss": 0.6618302464485168,
+      "eval_runtime": 74.0627,
+      "eval_samples_per_second": 55.926,
+      "eval_steps_per_second": 13.988,
+      "step": 18500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1948794770529757e-06,
+      "loss": 0.6121,
+      "step": 18510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1858005356575428e-06,
+      "loss": 0.632,
+      "step": 18520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1767215942621095e-06,
+      "loss": 0.5984,
+      "step": 18530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1676426528666757e-06,
+      "loss": 0.8053,
+      "step": 18540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1585637114712424e-06,
+      "loss": 0.7311,
+      "step": 18550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1494847700758096e-06,
+      "loss": 0.6496,
+      "step": 18560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1404058286803763e-06,
+      "loss": 0.7105,
+      "step": 18570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.131326887284943e-06,
+      "loss": 0.6768,
+      "step": 18580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1222479458895092e-06,
+      "loss": 0.7305,
+      "step": 18590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.113169004494076e-06,
+      "loss": 0.8035,
+      "step": 18600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.104090063098643e-06,
+      "loss": 0.6086,
+      "step": 18610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0950111217032097e-06,
+      "loss": 0.6119,
+      "step": 18620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0859321803077764e-06,
+      "loss": 0.7191,
+      "step": 18630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0768532389123427e-06,
+      "loss": 0.648,
+      "step": 18640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.06777429751691e-06,
+      "loss": 0.7826,
+      "step": 18650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0586953561214765e-06,
+      "loss": 0.7137,
+      "step": 18660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0496164147260432e-06,
+      "loss": 0.8223,
+      "step": 18670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.04053747333061e-06,
+      "loss": 0.7203,
+      "step": 18680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.031458531935176e-06,
+      "loss": 0.6809,
+      "step": 18690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0223795905397433e-06,
+      "loss": 0.6203,
+      "step": 18700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.01330064914431e-06,
+      "loss": 0.6621,
+      "step": 18710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0042217077488767e-06,
+      "loss": 0.6363,
+      "step": 18720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9951427663534434e-06,
+      "loss": 0.6408,
+      "step": 18730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9860638249580105e-06,
+      "loss": 0.5842,
+      "step": 18740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9769848835625768e-06,
+      "loss": 0.7553,
+      "step": 18750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9679059421671435e-06,
+      "loss": 0.7262,
+      "step": 18760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.95882700077171e-06,
+      "loss": 0.6549,
+      "step": 18770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.949748059376277e-06,
+      "loss": 0.7242,
+      "step": 18780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.940669117980844e-06,
+      "loss": 0.625,
+      "step": 18790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9315901765854103e-06,
+      "loss": 0.657,
+      "step": 18800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.922511235189977e-06,
+      "loss": 0.7918,
+      "step": 18810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9134322937945437e-06,
+      "loss": 0.6266,
+      "step": 18820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9043533523991108e-06,
+      "loss": 0.6428,
+      "step": 18830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8952744110036775e-06,
+      "loss": 0.7258,
+      "step": 18840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8861954696082437e-06,
+      "loss": 0.6338,
+      "step": 18850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8771165282128104e-06,
+      "loss": 0.5566,
+      "step": 18860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.868037586817377e-06,
+      "loss": 0.6768,
+      "step": 18870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8589586454219443e-06,
+      "loss": 0.7463,
+      "step": 18880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.849879704026511e-06,
+      "loss": 0.7348,
+      "step": 18890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8408007626310772e-06,
+      "loss": 0.741,
+      "step": 18900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.831721821235644e-06,
+      "loss": 0.7045,
+      "step": 18910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.822642879840211e-06,
+      "loss": 0.6104,
+      "step": 18920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8135639384447777e-06,
+      "loss": 0.5941,
+      "step": 18930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8044849970493444e-06,
+      "loss": 0.7625,
+      "step": 18940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7954060556539107e-06,
+      "loss": 0.6734,
+      "step": 18950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7863271142584774e-06,
+      "loss": 0.7344,
+      "step": 18960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7772481728630445e-06,
+      "loss": 0.6715,
+      "step": 18970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7681692314676112e-06,
+      "loss": 0.7699,
+      "step": 18980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.759090290072178e-06,
+      "loss": 0.6113,
+      "step": 18990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.750011348676744e-06,
+      "loss": 0.634,
+      "step": 19000
+    },
+    {
+      "epoch": 0.86,
+      "eval_accuracy": 0.629406084017383,
+      "eval_loss": 0.6582936644554138,
+      "eval_runtime": 74.1488,
+      "eval_samples_per_second": 55.861,
+      "eval_steps_per_second": 13.972,
+      "step": 19000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7409324072813113e-06,
+      "loss": 0.5986,
+      "step": 19010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.731853465885878e-06,
+      "loss": 0.7604,
+      "step": 19020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7227745244904447e-06,
+      "loss": 0.6496,
+      "step": 19030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7136955830950114e-06,
+      "loss": 0.6102,
+      "step": 19040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7046166416995777e-06,
+      "loss": 0.7336,
+      "step": 19050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6955377003041448e-06,
+      "loss": 0.7105,
+      "step": 19060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6864587589087115e-06,
+      "loss": 0.6516,
+      "step": 19070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.677379817513278e-06,
+      "loss": 0.6199,
+      "step": 19080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.668300876117845e-06,
+      "loss": 0.6559,
+      "step": 19090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.659221934722412e-06,
+      "loss": 0.6232,
+      "step": 19100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6501429933269783e-06,
+      "loss": 0.634,
+      "step": 19110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.641064051931545e-06,
+      "loss": 0.6158,
+      "step": 19120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6319851105361117e-06,
+      "loss": 0.685,
+      "step": 19130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6229061691406784e-06,
+      "loss": 0.6271,
+      "step": 19140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6138272277452455e-06,
+      "loss": 0.6379,
+      "step": 19150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6047482863498117e-06,
+      "loss": 0.6584,
+      "step": 19160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5956693449543784e-06,
+      "loss": 0.6168,
+      "step": 19170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.586590403558945e-06,
+      "loss": 0.5928,
+      "step": 19180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.577511462163512e-06,
+      "loss": 0.5943,
+      "step": 19190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.568432520768079e-06,
+      "loss": 0.7664,
+      "step": 19200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5593535793726452e-06,
+      "loss": 0.698,
+      "step": 19210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.550274637977212e-06,
+      "loss": 0.6602,
+      "step": 19220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5411956965817786e-06,
+      "loss": 0.6725,
+      "step": 19230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5321167551863457e-06,
+      "loss": 0.6062,
+      "step": 19240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5230378137909124e-06,
+      "loss": 0.6545,
+      "step": 19250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5139588723954787e-06,
+      "loss": 0.5836,
+      "step": 19260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5048799310000454e-06,
+      "loss": 0.6107,
+      "step": 19270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.495800989604612e-06,
+      "loss": 0.6293,
+      "step": 19280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.486722048209179e-06,
+      "loss": 0.6992,
+      "step": 19290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.477643106813746e-06,
+      "loss": 0.5889,
+      "step": 19300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.468564165418312e-06,
+      "loss": 0.6504,
+      "step": 19310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4594852240228793e-06,
+      "loss": 0.6891,
+      "step": 19320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4504062826274456e-06,
+      "loss": 0.6273,
+      "step": 19330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4413273412320127e-06,
+      "loss": 0.6385,
+      "step": 19340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4322483998365794e-06,
+      "loss": 0.6826,
+      "step": 19350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.423169458441146e-06,
+      "loss": 0.6477,
+      "step": 19360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.414090517045713e-06,
+      "loss": 0.6295,
+      "step": 19370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.405011575650279e-06,
+      "loss": 0.6473,
+      "step": 19380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.395932634254846e-06,
+      "loss": 0.7258,
+      "step": 19390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.386853692859413e-06,
+      "loss": 0.7238,
+      "step": 19400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3777747514639796e-06,
+      "loss": 0.602,
+      "step": 19410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3686958100685463e-06,
+      "loss": 0.6506,
+      "step": 19420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.359616868673113e-06,
+      "loss": 0.657,
+      "step": 19430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3505379272776797e-06,
+      "loss": 0.5479,
+      "step": 19440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3414589858822464e-06,
+      "loss": 0.7652,
+      "step": 19450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.332380044486813e-06,
+      "loss": 0.6119,
+      "step": 19460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3233011030913798e-06,
+      "loss": 0.625,
+      "step": 19470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3142221616959464e-06,
+      "loss": 0.634,
+      "step": 19480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.305143220300513e-06,
+      "loss": 0.6096,
+      "step": 19490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.29606427890508e-06,
+      "loss": 0.6139,
+      "step": 19500
+    },
+    {
+      "epoch": 0.89,
+      "eval_accuracy": 0.6296475132786093,
+      "eval_loss": 0.6594942212104797,
+      "eval_runtime": 73.3398,
+      "eval_samples_per_second": 56.477,
+      "eval_steps_per_second": 14.126,
+      "step": 19500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2869853375096465e-06,
+      "loss": 0.6381,
+      "step": 19510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2779063961142132e-06,
+      "loss": 0.6336,
+      "step": 19520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.26882745471878e-06,
+      "loss": 0.7096,
+      "step": 19530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2597485133233466e-06,
+      "loss": 0.6396,
+      "step": 19540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2506695719279133e-06,
+      "loss": 0.6719,
+      "step": 19550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.24159063053248e-06,
+      "loss": 0.6197,
+      "step": 19560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2325116891370467e-06,
+      "loss": 0.6291,
+      "step": 19570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2234327477416134e-06,
+      "loss": 0.6397,
+      "step": 19580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.21435380634618e-06,
+      "loss": 0.7871,
+      "step": 19590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.205274864950747e-06,
+      "loss": 0.7424,
+      "step": 19600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.196195923555314e-06,
+      "loss": 0.7363,
+      "step": 19610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.18711698215988e-06,
+      "loss": 0.7027,
+      "step": 19620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.178038040764447e-06,
+      "loss": 0.627,
+      "step": 19630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1689590993690136e-06,
+      "loss": 0.7043,
+      "step": 19640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1598801579735803e-06,
+      "loss": 0.5365,
+      "step": 19650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1508012165781474e-06,
+      "loss": 0.607,
+      "step": 19660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1417222751827137e-06,
+      "loss": 0.6062,
+      "step": 19670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.132643333787281e-06,
+      "loss": 0.7021,
+      "step": 19680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.123564392391847e-06,
+      "loss": 0.5836,
+      "step": 19690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.114485450996414e-06,
+      "loss": 0.6492,
+      "step": 19700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.105406509600981e-06,
+      "loss": 0.5834,
+      "step": 19710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.096327568205547e-06,
+      "loss": 0.5445,
+      "step": 19720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0872486268101143e-06,
+      "loss": 0.5725,
+      "step": 19730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0781696854146805e-06,
+      "loss": 0.6119,
+      "step": 19740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0690907440192477e-06,
+      "loss": 0.5551,
+      "step": 19750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0600118026238144e-06,
+      "loss": 0.6842,
+      "step": 19760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.050932861228381e-06,
+      "loss": 0.6477,
+      "step": 19770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0418539198329478e-06,
+      "loss": 0.6268,
+      "step": 19780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0327749784375144e-06,
+      "loss": 0.6814,
+      "step": 19790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.023696037042081e-06,
+      "loss": 0.6207,
+      "step": 19800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.014617095646648e-06,
+      "loss": 0.5721,
+      "step": 19810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0055381542512145e-06,
+      "loss": 0.6545,
+      "step": 19820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9964592128557812e-06,
+      "loss": 0.7377,
+      "step": 19830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.987380271460348e-06,
+      "loss": 0.6426,
+      "step": 19840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9783013300649146e-06,
+      "loss": 0.6916,
+      "step": 19850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9692223886694813e-06,
+      "loss": 0.609,
+      "step": 19860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.960143447274048e-06,
+      "loss": 0.7535,
+      "step": 19870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9510645058786147e-06,
+      "loss": 0.6238,
+      "step": 19880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9419855644831814e-06,
+      "loss": 0.7176,
+      "step": 19890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.932906623087748e-06,
+      "loss": 0.6641,
+      "step": 19900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.923827681692315e-06,
+      "loss": 0.5863,
+      "step": 19910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9147487402968815e-06,
+      "loss": 0.6531,
+      "step": 19920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9056697989014484e-06,
+      "loss": 0.6604,
+      "step": 19930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8965908575060149e-06,
+      "loss": 0.6605,
+      "step": 19940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8875119161105818e-06,
+      "loss": 0.7225,
+      "step": 19950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8784329747151483e-06,
+      "loss": 0.7346,
+      "step": 19960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8693540333197152e-06,
+      "loss": 0.6707,
+      "step": 19970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8602750919242819e-06,
+      "loss": 0.6119,
+      "step": 19980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8511961505288484e-06,
+      "loss": 0.6988,
+      "step": 19990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8421172091334153e-06,
+      "loss": 0.6289,
+      "step": 20000
+    },
+    {
+      "epoch": 0.91,
+      "eval_accuracy": 0.6286817962337036,
+      "eval_loss": 0.6617236733436584,
+      "eval_runtime": 74.5822,
+      "eval_samples_per_second": 55.536,
+      "eval_steps_per_second": 13.891,
+      "step": 20000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8330382677379818e-06,
+      "loss": 0.6711,
+      "step": 20010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8239593263425487e-06,
+      "loss": 0.5879,
+      "step": 20020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8148803849471154e-06,
+      "loss": 0.7063,
+      "step": 20030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.805801443551682e-06,
+      "loss": 0.5883,
+      "step": 20040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7967225021562488e-06,
+      "loss": 0.5684,
+      "step": 20050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7876435607608152e-06,
+      "loss": 0.6596,
+      "step": 20060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7785646193653822e-06,
+      "loss": 0.5432,
+      "step": 20070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7694856779699488e-06,
+      "loss": 0.5748,
+      "step": 20080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7604067365745155e-06,
+      "loss": 0.7727,
+      "step": 20090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7513277951790822e-06,
+      "loss": 0.5312,
+      "step": 20100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7422488537836491e-06,
+      "loss": 0.717,
+      "step": 20110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7331699123882156e-06,
+      "loss": 0.6273,
+      "step": 20120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7240909709927825e-06,
+      "loss": 0.627,
+      "step": 20130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.715012029597349e-06,
+      "loss": 0.8084,
+      "step": 20140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7059330882019157e-06,
+      "loss": 0.557,
+      "step": 20150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6968541468064826e-06,
+      "loss": 0.75,
+      "step": 20160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6877752054110491e-06,
+      "loss": 0.7277,
+      "step": 20170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.678696264015616e-06,
+      "loss": 0.6887,
+      "step": 20180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6696173226201825e-06,
+      "loss": 0.5959,
+      "step": 20190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6605383812247494e-06,
+      "loss": 0.6359,
+      "step": 20200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.651459439829316e-06,
+      "loss": 0.5998,
+      "step": 20210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6423804984338828e-06,
+      "loss": 0.585,
+      "step": 20220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6333015570384495e-06,
+      "loss": 0.6037,
+      "step": 20230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624222615643016e-06,
+      "loss": 0.7137,
+      "step": 20240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.615143674247583e-06,
+      "loss": 0.7148,
+      "step": 20250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6060647328521494e-06,
+      "loss": 0.6832,
+      "step": 20260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5969857914567163e-06,
+      "loss": 0.6246,
+      "step": 20270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.587906850061283e-06,
+      "loss": 0.6691,
+      "step": 20280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5788279086658497e-06,
+      "loss": 0.6387,
+      "step": 20290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5697489672704164e-06,
+      "loss": 0.6922,
+      "step": 20300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5606700258749833e-06,
+      "loss": 0.6711,
+      "step": 20310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5515910844795498e-06,
+      "loss": 0.693,
+      "step": 20320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5425121430841165e-06,
+      "loss": 0.618,
+      "step": 20330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5334332016886832e-06,
+      "loss": 0.6441,
+      "step": 20340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5243542602932499e-06,
+      "loss": 0.6623,
+      "step": 20350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5152753188978168e-06,
+      "loss": 0.6813,
+      "step": 20360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5061963775023832e-06,
+      "loss": 0.6551,
+      "step": 20370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4971174361069502e-06,
+      "loss": 0.6084,
+      "step": 20380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4880384947115166e-06,
+      "loss": 0.6887,
+      "step": 20390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4789595533160835e-06,
+      "loss": 0.7918,
+      "step": 20400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4698806119206502e-06,
+      "loss": 0.626,
+      "step": 20410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4608016705252167e-06,
+      "loss": 0.6773,
+      "step": 20420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4517227291297836e-06,
+      "loss": 0.6996,
+      "step": 20430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4426437877343501e-06,
+      "loss": 0.6697,
+      "step": 20440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.433564846338917e-06,
+      "loss": 0.5686,
+      "step": 20450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4244859049434837e-06,
+      "loss": 0.6312,
+      "step": 20460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4154069635480504e-06,
+      "loss": 0.6873,
+      "step": 20470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4063280221526171e-06,
+      "loss": 0.7268,
+      "step": 20480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.397249080757184e-06,
+      "loss": 0.6863,
+      "step": 20490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3881701393617505e-06,
+      "loss": 0.5889,
+      "step": 20500
+    },
+    {
+      "epoch": 0.93,
+      "eval_accuracy": 0.6301303718010622,
+      "eval_loss": 0.6595224738121033,
+      "eval_runtime": 73.6905,
+      "eval_samples_per_second": 56.208,
+      "eval_steps_per_second": 14.059,
+      "step": 20500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3790911979663172e-06,
+      "loss": 0.5518,
+      "step": 20510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.370012256570884e-06,
+      "loss": 0.7367,
+      "step": 20520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3609333151754506e-06,
+      "loss": 0.5818,
+      "step": 20530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3518543737800175e-06,
+      "loss": 0.7182,
+      "step": 20540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.342775432384584e-06,
+      "loss": 0.5711,
+      "step": 20550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.333696490989151e-06,
+      "loss": 0.6314,
+      "step": 20560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3246175495937174e-06,
+      "loss": 0.6828,
+      "step": 20570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.315538608198284e-06,
+      "loss": 0.6562,
+      "step": 20580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.306459666802851e-06,
+      "loss": 0.8039,
+      "step": 20590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2973807254074175e-06,
+      "loss": 0.6371,
+      "step": 20600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2883017840119844e-06,
+      "loss": 0.6227,
+      "step": 20610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2792228426165509e-06,
+      "loss": 0.6477,
+      "step": 20620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2701439012211178e-06,
+      "loss": 0.7176,
+      "step": 20630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2610649598256845e-06,
+      "loss": 0.6961,
+      "step": 20640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2519860184302512e-06,
+      "loss": 0.5751,
+      "step": 20650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2429070770348179e-06,
+      "loss": 0.6572,
+      "step": 20660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2338281356393846e-06,
+      "loss": 0.6223,
+      "step": 20670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2247491942439513e-06,
+      "loss": 0.768,
+      "step": 20680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.215670252848518e-06,
+      "loss": 0.5842,
+      "step": 20690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2065913114530846e-06,
+      "loss": 0.6072,
+      "step": 20700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1975123700576513e-06,
+      "loss": 0.6664,
+      "step": 20710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.188433428662218e-06,
+      "loss": 0.6254,
+      "step": 20720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1793544872667847e-06,
+      "loss": 0.6463,
+      "step": 20730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1702755458713514e-06,
+      "loss": 0.7381,
+      "step": 20740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1611966044759181e-06,
+      "loss": 0.6648,
+      "step": 20750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1521176630804848e-06,
+      "loss": 0.7143,
+      "step": 20760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1430387216850517e-06,
+      "loss": 0.6377,
+      "step": 20770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1339597802896184e-06,
+      "loss": 0.6785,
+      "step": 20780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1248808388941851e-06,
+      "loss": 0.7199,
+      "step": 20790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1158018974987516e-06,
+      "loss": 0.6695,
+      "step": 20800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1067229561033183e-06,
+      "loss": 0.6596,
+      "step": 20810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0976440147078852e-06,
+      "loss": 0.7838,
+      "step": 20820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.088565073312452e-06,
+      "loss": 0.6062,
+      "step": 20830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0794861319170186e-06,
+      "loss": 0.6529,
+      "step": 20840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0704071905215853e-06,
+      "loss": 0.7447,
+      "step": 20850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.061328249126152e-06,
+      "loss": 0.7277,
+      "step": 20860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0522493077307187e-06,
+      "loss": 0.6977,
+      "step": 20870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0431703663352854e-06,
+      "loss": 0.5682,
+      "step": 20880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.034091424939852e-06,
+      "loss": 0.6207,
+      "step": 20890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0250124835444188e-06,
+      "loss": 0.6986,
+      "step": 20900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0159335421489855e-06,
+      "loss": 0.6393,
+      "step": 20910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0068546007535522e-06,
+      "loss": 0.609,
+      "step": 20920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.977756593581189e-07,
+      "loss": 0.727,
+      "step": 20930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.886967179626856e-07,
+      "loss": 0.6385,
+      "step": 20940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.796177765672525e-07,
+      "loss": 0.7049,
+      "step": 20950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.705388351718192e-07,
+      "loss": 0.6059,
+      "step": 20960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.614598937763857e-07,
+      "loss": 0.6453,
+      "step": 20970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 0.61,
+      "step": 20980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.433020109855191e-07,
+      "loss": 0.7059,
+      "step": 20990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.342230695900858e-07,
+      "loss": 0.6539,
+      "step": 21000
+    },
+    {
+      "epoch": 0.95,
+      "eval_accuracy": 0.6279575084500242,
+      "eval_loss": 0.6598912477493286,
+      "eval_runtime": 73.9974,
+      "eval_samples_per_second": 55.975,
+      "eval_steps_per_second": 14.0,
+      "step": 21000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.251441281946525e-07,
+      "loss": 0.7365,
+      "step": 21010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.160651867992193e-07,
+      "loss": 0.7266,
+      "step": 21020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.06986245403786e-07,
+      "loss": 0.7639,
+      "step": 21030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.979073040083527e-07,
+      "loss": 0.6879,
+      "step": 21040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.888283626129194e-07,
+      "loss": 0.7623,
+      "step": 21050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.79749421217486e-07,
+      "loss": 0.6229,
+      "step": 21060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.706704798220528e-07,
+      "loss": 0.5939,
+      "step": 21070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.615915384266195e-07,
+      "loss": 0.6016,
+      "step": 21080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.525125970311862e-07,
+      "loss": 0.5709,
+      "step": 21090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.434336556357529e-07,
+      "loss": 0.6391,
+      "step": 21100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.343547142403197e-07,
+      "loss": 0.5955,
+      "step": 21110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.252757728448864e-07,
+      "loss": 0.6453,
+      "step": 21120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.161968314494531e-07,
+      "loss": 0.5959,
+      "step": 21130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.071178900540198e-07,
+      "loss": 0.7219,
+      "step": 21140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.980389486585864e-07,
+      "loss": 0.7617,
+      "step": 21150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.889600072631531e-07,
+      "loss": 0.6307,
+      "step": 21160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.798810658677199e-07,
+      "loss": 0.6355,
+      "step": 21170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.708021244722866e-07,
+      "loss": 0.6818,
+      "step": 21180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.617231830768533e-07,
+      "loss": 0.7445,
+      "step": 21190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.5264424168142e-07,
+      "loss": 0.6539,
+      "step": 21200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.435653002859868e-07,
+      "loss": 0.6465,
+      "step": 21210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.344863588905535e-07,
+      "loss": 0.6133,
+      "step": 21220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.254074174951202e-07,
+      "loss": 0.6699,
+      "step": 21230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.163284760996868e-07,
+      "loss": 0.6191,
+      "step": 21240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.072495347042535e-07,
+      "loss": 0.5492,
+      "step": 21250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.981705933088203e-07,
+      "loss": 0.8207,
+      "step": 21260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.89091651913387e-07,
+      "loss": 0.6326,
+      "step": 21270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.800127105179537e-07,
+      "loss": 0.7082,
+      "step": 21280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.709337691225204e-07,
+      "loss": 0.7359,
+      "step": 21290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.618548277270872e-07,
+      "loss": 0.7242,
+      "step": 21300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.527758863316538e-07,
+      "loss": 0.7369,
+      "step": 21310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.436969449362204e-07,
+      "loss": 0.6687,
+      "step": 21320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.346180035407871e-07,
+      "loss": 0.7807,
+      "step": 21330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.255390621453538e-07,
+      "loss": 0.6871,
+      "step": 21340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.164601207499206e-07,
+      "loss": 0.6143,
+      "step": 21350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.073811793544873e-07,
+      "loss": 0.6752,
+      "step": 21360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.98302237959054e-07,
+      "loss": 0.6838,
+      "step": 21370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.892232965636207e-07,
+      "loss": 0.5486,
+      "step": 21380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.801443551681874e-07,
+      "loss": 0.6605,
+      "step": 21390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.710654137727541e-07,
+      "loss": 0.6533,
+      "step": 21400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.619864723773208e-07,
+      "loss": 0.616,
+      "step": 21410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.529075309818876e-07,
+      "loss": 0.6836,
+      "step": 21420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.438285895864542e-07,
+      "loss": 0.6934,
+      "step": 21430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.34749648191021e-07,
+      "loss": 0.768,
+      "step": 21440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.256707067955877e-07,
+      "loss": 0.6928,
+      "step": 21450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165917654001544e-07,
+      "loss": 0.6156,
+      "step": 21460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.075128240047211e-07,
+      "loss": 0.6945,
+      "step": 21470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.984338826092878e-07,
+      "loss": 0.6664,
+      "step": 21480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.893549412138545e-07,
+      "loss": 0.6572,
+      "step": 21490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.802759998184212e-07,
+      "loss": 0.6396,
+      "step": 21500
+    },
+    {
+      "epoch": 0.98,
+      "eval_accuracy": 0.6253017865765331,
+      "eval_loss": 0.6602552533149719,
+      "eval_runtime": 73.7722,
+      "eval_samples_per_second": 56.146,
+      "eval_steps_per_second": 14.043,
+      "step": 21500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7119705842298793e-07,
+      "loss": 0.6551,
+      "step": 21510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6211811702755463e-07,
+      "loss": 0.6627,
+      "step": 21520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.530391756321213e-07,
+      "loss": 0.7145,
+      "step": 21530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4396023423668807e-07,
+      "loss": 0.6893,
+      "step": 21540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.3488129284125477e-07,
+      "loss": 0.6969,
+      "step": 21550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.258023514458214e-07,
+      "loss": 0.7125,
+      "step": 21560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1672341005038816e-07,
+      "loss": 0.6242,
+      "step": 21570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0764446865495486e-07,
+      "loss": 0.635,
+      "step": 21580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.985655272595216e-07,
+      "loss": 0.7063,
+      "step": 21590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8948658586408825e-07,
+      "loss": 0.775,
+      "step": 21600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8040764446865495e-07,
+      "loss": 0.5451,
+      "step": 21610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.713287030732217e-07,
+      "loss": 0.5568,
+      "step": 21620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.622497616777884e-07,
+      "loss": 0.7299,
+      "step": 21630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5317082028235514e-07,
+      "loss": 0.6406,
+      "step": 21640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.440918788869218e-07,
+      "loss": 0.6998,
+      "step": 21650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3501293749148853e-07,
+      "loss": 0.6822,
+      "step": 21660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2593399609605523e-07,
+      "loss": 0.6109,
+      "step": 21670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.16855054700622e-07,
+      "loss": 0.6658,
+      "step": 21680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.077761133051886e-07,
+      "loss": 0.6207,
+      "step": 21690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.986971719097553e-07,
+      "loss": 0.7873,
+      "step": 21700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8961823051432207e-07,
+      "loss": 0.6695,
+      "step": 21710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8053928911888876e-07,
+      "loss": 0.6246,
+      "step": 21720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7146034772345546e-07,
+      "loss": 0.6021,
+      "step": 21730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.623814063280222e-07,
+      "loss": 0.6416,
+      "step": 21740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.533024649325889e-07,
+      "loss": 0.6463,
+      "step": 21750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.442235235371556e-07,
+      "loss": 0.6547,
+      "step": 21760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.351445821417223e-07,
+      "loss": 0.6244,
+      "step": 21770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2606564074628902e-07,
+      "loss": 0.6291,
+      "step": 21780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1698669935085571e-07,
+      "loss": 0.7139,
+      "step": 21790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0790775795542244e-07,
+      "loss": 0.6533,
+      "step": 21800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.988288165599891e-07,
+      "loss": 0.7258,
+      "step": 21810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8974987516455583e-07,
+      "loss": 0.6953,
+      "step": 21820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8067093376912252e-07,
+      "loss": 0.6939,
+      "step": 21830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7159199237368925e-07,
+      "loss": 0.7047,
+      "step": 21840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6251305097825594e-07,
+      "loss": 0.6329,
+      "step": 21850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5343410958282267e-07,
+      "loss": 0.6727,
+      "step": 21860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4435516818738936e-07,
+      "loss": 0.7387,
+      "step": 21870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3527622679195606e-07,
+      "loss": 0.7191,
+      "step": 21880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2619728539652278e-07,
+      "loss": 0.7266,
+      "step": 21890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1711834400108948e-07,
+      "loss": 0.6119,
+      "step": 21900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0803940260565619e-07,
+      "loss": 0.6201,
+      "step": 21910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.896046121022288e-08,
+      "loss": 0.7092,
+      "step": 21920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.98815198147896e-08,
+      "loss": 0.6348,
+      "step": 21930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.080257841935631e-08,
+      "loss": 0.7164,
+      "step": 21940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.172363702392301e-08,
+      "loss": 0.6273,
+      "step": 21950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.264469562848973e-08,
+      "loss": 0.6154,
+      "step": 21960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3565754233056436e-08,
+      "loss": 0.7488,
+      "step": 21970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.448681283762314e-08,
+      "loss": 0.6797,
+      "step": 21980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.540787144218984e-08,
+      "loss": 0.6977,
+      "step": 21990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6328930046756547e-08,
+      "loss": 0.632,
+      "step": 22000
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.6269917914051183,
+      "eval_loss": 0.6599072813987732,
+      "eval_runtime": 73.7894,
+      "eval_samples_per_second": 56.133,
+      "eval_steps_per_second": 14.04,
+      "step": 22000
+    }
+  ],
+  "max_steps": 22029,
+  "num_train_epochs": 1,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}