diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29276 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9984012789768184,
+  "global_step": 47500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 20.3727,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 22.0558,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 21.7109,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 20.6611,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 22.4204,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 19.938,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 17.4213,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 20.1383,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 19.8127,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 23.1777,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-06,
+      "loss": 23.3547,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 16.6238,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-06,
+      "loss": 17.8069,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 20.2351,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 21.9262,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 18.2762,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 20.3464,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 19.9402,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 18.0535,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 24.1097,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 22.9605,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-06,
+      "loss": 18.8428,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 13.8694,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 17.6931,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 20.0871,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-05,
+      "loss": 15.7877,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 13.5578,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 17.1961,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.16e-05,
+      "loss": 14.2598,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 12.4614,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 11.931,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 13.7064,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 14.322,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 12.1773,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-05,
+      "loss": 15.734,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 15.678,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.48e-05,
+      "loss": 13.6815,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 13.4937,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 11.4397,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 13.4706,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.64e-05,
+      "loss": 12.5004,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 11.8415,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.72e-05,
+      "loss": 16.3443,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76e-05,
+      "loss": 12.4893,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8e-05,
+      "loss": 8.1513,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 11.0122,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.88e-05,
+      "loss": 9.984,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 9.837,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 8.5917,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 8.5767,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 3.7170910835266113,
+      "eval_runtime": 244.7625,
+      "eval_samples_per_second": 10.226,
+      "eval_steps_per_second": 5.115,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995748118542458e-05,
+      "loss": 7.3945,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999149623708491e-05,
+      "loss": 7.7318,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987244355627367e-05,
+      "loss": 6.6012,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982992474169823e-05,
+      "loss": 7.4308,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978740592712276e-05,
+      "loss": 6.8742,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974488711254732e-05,
+      "loss": 6.6656,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970236829797188e-05,
+      "loss": 6.815,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9965984948339644e-05,
+      "loss": 6.5222,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9961733066882097e-05,
+      "loss": 7.4163,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957481185424553e-05,
+      "loss": 6.5741,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995322930396701e-05,
+      "loss": 7.1728,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948977422509462e-05,
+      "loss": 5.768,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944725541051918e-05,
+      "loss": 6.0008,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940473659594374e-05,
+      "loss": 5.5226,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9936221778136827e-05,
+      "loss": 5.7387,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9931969896679283e-05,
+      "loss": 5.1679,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9927718015221736e-05,
+      "loss": 5.0973,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9923466133764192e-05,
+      "loss": 5.1589,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9919214252306648e-05,
+      "loss": 5.1914,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99149623708491e-05,
+      "loss": 4.5417,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9910710489391557e-05,
+      "loss": 5.7502,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9906458607934013e-05,
+      "loss": 4.8608,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9902206726476466e-05,
+      "loss": 5.4553,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9897954845018922e-05,
+      "loss": 4.9113,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9893702963561375e-05,
+      "loss": 4.6249,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988945108210383e-05,
+      "loss": 4.3366,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9885199200646287e-05,
+      "loss": 4.7683,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9880947319188743e-05,
+      "loss": 4.2559,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9876695437731196e-05,
+      "loss": 4.6332,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9872443556273652e-05,
+      "loss": 4.6606,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986819167481611e-05,
+      "loss": 4.4119,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986393979335856e-05,
+      "loss": 5.0149,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9859687911901017e-05,
+      "loss": 4.6104,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9855436030443474e-05,
+      "loss": 4.1719,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9851184148985926e-05,
+      "loss": 4.5308,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9846932267528382e-05,
+      "loss": 4.1654,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.984268038607084e-05,
+      "loss": 5.0468,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9838428504613295e-05,
+      "loss": 4.0722,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9834176623155748e-05,
+      "loss": 3.9921,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9829924741698204e-05,
+      "loss": 4.235,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982567286024066e-05,
+      "loss": 4.0093,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9821420978783113e-05,
+      "loss": 4.496,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.981716909732557e-05,
+      "loss": 4.3413,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9812917215868025e-05,
+      "loss": 4.7558,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9808665334410478e-05,
+      "loss": 4.7494,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9804413452952934e-05,
+      "loss": 4.213,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.980016157149539e-05,
+      "loss": 4.8979,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9795909690037846e-05,
+      "loss": 4.2093,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.97916578085803e-05,
+      "loss": 4.4593,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9787405927122755e-05,
+      "loss": 3.8047,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.6360342502593994,
+      "eval_runtime": 306.5779,
+      "eval_samples_per_second": 8.164,
+      "eval_steps_per_second": 4.084,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9783154045665208e-05,
+      "loss": 3.9264,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9778902164207664e-05,
+      "loss": 4.5501,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.977465028275012e-05,
+      "loss": 4.2464,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9770398401292573e-05,
+      "loss": 3.8889,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.976614651983503e-05,
+      "loss": 4.1145,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9761894638377482e-05,
+      "loss": 3.9731,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9757642756919938e-05,
+      "loss": 3.9231,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9753390875462394e-05,
+      "loss": 5.0785,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9749138994004847e-05,
+      "loss": 4.017,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9744887112547303e-05,
+      "loss": 3.7215,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.974063523108976e-05,
+      "loss": 3.976,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9736383349632212e-05,
+      "loss": 3.7487,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9732131468174668e-05,
+      "loss": 4.1519,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9727879586717124e-05,
+      "loss": 3.9152,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9723627705259577e-05,
+      "loss": 3.4579,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9719375823802033e-05,
+      "loss": 3.4721,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.971512394234449e-05,
+      "loss": 3.7831,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9710872060886942e-05,
+      "loss": 3.5758,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9706620179429398e-05,
+      "loss": 4.974,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9702368297971854e-05,
+      "loss": 3.5293,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.969811641651431e-05,
+      "loss": 4.3557,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9693864535056763e-05,
+      "loss": 3.6766,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968961265359922e-05,
+      "loss": 3.6694,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9685360772141676e-05,
+      "loss": 3.9718,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.968110889068413e-05,
+      "loss": 3.8476,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9676857009226585e-05,
+      "loss": 3.825,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.967260512776904e-05,
+      "loss": 3.9017,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9668353246311493e-05,
+      "loss": 3.8978,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.966410136485395e-05,
+      "loss": 3.8765,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9659849483396406e-05,
+      "loss": 3.7227,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9655597601938862e-05,
+      "loss": 3.3455,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9651345720481315e-05,
+      "loss": 3.4121,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.964709383902377e-05,
+      "loss": 3.3375,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9642841957566227e-05,
+      "loss": 3.8211,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.963859007610868e-05,
+      "loss": 3.2857,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9634338194651136e-05,
+      "loss": 3.7018,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9630086313193592e-05,
+      "loss": 3.5457,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9625834431736045e-05,
+      "loss": 3.8864,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.96215825502785e-05,
+      "loss": 3.8284,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9617330668820954e-05,
+      "loss": 3.3557,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.961307878736341e-05,
+      "loss": 3.2405,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9608826905905866e-05,
+      "loss": 3.9348,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.960457502444832e-05,
+      "loss": 3.3789,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9600323142990775e-05,
+      "loss": 3.2882,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9596071261533228e-05,
+      "loss": 3.4028,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9591819380075684e-05,
+      "loss": 3.5537,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.958756749861814e-05,
+      "loss": 3.3862,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9583315617160593e-05,
+      "loss": 3.3672,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.957906373570305e-05,
+      "loss": 3.0539,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9574811854245505e-05,
+      "loss": 3.2658,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.444038152694702,
+      "eval_runtime": 325.0737,
+      "eval_samples_per_second": 7.7,
+      "eval_steps_per_second": 3.851,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.957055997278796e-05,
+      "loss": 3.7892,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9566308091330414e-05,
+      "loss": 3.0661,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.956205620987287e-05,
+      "loss": 3.7278,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9557804328415326e-05,
+      "loss": 3.6798,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.955355244695778e-05,
+      "loss": 3.8645,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9549300565500235e-05,
+      "loss": 3.3384,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.954504868404269e-05,
+      "loss": 3.3859,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9540796802585144e-05,
+      "loss": 3.1837,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.95365449211276e-05,
+      "loss": 3.6801,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9532293039670056e-05,
+      "loss": 3.3226,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.952804115821251e-05,
+      "loss": 3.9958,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9523789276754965e-05,
+      "loss": 3.3704,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.951953739529742e-05,
+      "loss": 3.1825,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9515285513839878e-05,
+      "loss": 3.3603,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.951103363238233e-05,
+      "loss": 3.3135,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9506781750924787e-05,
+      "loss": 3.2238,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9502529869467243e-05,
+      "loss": 3.2581,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9498277988009695e-05,
+      "loss": 3.1484,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.949402610655215e-05,
+      "loss": 3.2461,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9489774225094608e-05,
+      "loss": 3.4481,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.948552234363706e-05,
+      "loss": 3.5146,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9481270462179517e-05,
+      "loss": 3.23,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9477018580721973e-05,
+      "loss": 3.7308,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9472766699264426e-05,
+      "loss": 3.8861,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9468514817806882e-05,
+      "loss": 3.1529,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9464262936349338e-05,
+      "loss": 3.2984,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.946001105489179e-05,
+      "loss": 3.5839,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9455759173434247e-05,
+      "loss": 3.6126,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.94515072919767e-05,
+      "loss": 3.2796,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9447255410519156e-05,
+      "loss": 3.3423,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9443003529061612e-05,
+      "loss": 3.3523,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9438751647604065e-05,
+      "loss": 3.2899,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.943449976614652e-05,
+      "loss": 3.4348,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9430247884688977e-05,
+      "loss": 3.2762,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.942599600323143e-05,
+      "loss": 3.2729,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9421744121773886e-05,
+      "loss": 3.0372,
+      "step": 1860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9417492240316342e-05,
+      "loss": 3.2345,
+      "step": 1870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9413240358858795e-05,
+      "loss": 3.4763,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.940898847740125e-05,
+      "loss": 3.0048,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9404736595943707e-05,
+      "loss": 3.0538,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.940048471448616e-05,
+      "loss": 2.8958,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9396232833028616e-05,
+      "loss": 3.5233,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9391980951571072e-05,
+      "loss": 3.3356,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.938772907011353e-05,
+      "loss": 3.295,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.938347718865598e-05,
+      "loss": 3.837,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9379225307198437e-05,
+      "loss": 3.4193,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9374973425740893e-05,
+      "loss": 3.148,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9370721544283346e-05,
+      "loss": 2.9353,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9366469662825802e-05,
+      "loss": 3.6072,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.936221778136826e-05,
+      "loss": 3.3311,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.323817729949951,
+      "eval_runtime": 333.611,
+      "eval_samples_per_second": 7.503,
+      "eval_steps_per_second": 3.753,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.935796589991071e-05,
+      "loss": 3.2877,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9353714018453167e-05,
+      "loss": 3.2228,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9349462136995624e-05,
+      "loss": 3.3077,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9345210255538076e-05,
+      "loss": 3.5034,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9340958374080532e-05,
+      "loss": 3.548,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.933670649262299e-05,
+      "loss": 3.4721,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9332454611165445e-05,
+      "loss": 3.6428,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9328202729707898e-05,
+      "loss": 3.2712,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9323950848250354e-05,
+      "loss": 3.0507,
+      "step": 2090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9319698966792806e-05,
+      "loss": 2.9622,
+      "step": 2100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9315447085335263e-05,
+      "loss": 3.0459,
+      "step": 2110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.931119520387772e-05,
+      "loss": 3.1112,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.930694332242017e-05,
+      "loss": 2.7483,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9302691440962628e-05,
+      "loss": 2.857,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9298439559505084e-05,
+      "loss": 3.1156,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9294187678047537e-05,
+      "loss": 3.0197,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9289935796589993e-05,
+      "loss": 3.4195,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9285683915132446e-05,
+      "loss": 3.0336,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.92814320336749e-05,
+      "loss": 3.3716,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9277180152217358e-05,
+      "loss": 3.0866,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.927292827075981e-05,
+      "loss": 3.2992,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9268676389302267e-05,
+      "loss": 3.4868,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9264424507844723e-05,
+      "loss": 3.4946,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9260172626387176e-05,
+      "loss": 3.2567,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9255920744929632e-05,
+      "loss": 3.1213,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9251668863472088e-05,
+      "loss": 3.3634,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9247416982014544e-05,
+      "loss": 3.3944,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9243165100556997e-05,
+      "loss": 3.3117,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9238913219099453e-05,
+      "loss": 3.4853,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.923466133764191e-05,
+      "loss": 2.935,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9230409456184362e-05,
+      "loss": 2.8551,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9226157574726818e-05,
+      "loss": 3.8506,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9221905693269274e-05,
+      "loss": 2.854,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9217653811811727e-05,
+      "loss": 3.1607,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9213401930354183e-05,
+      "loss": 3.4625,
+      "step": 2350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.920915004889664e-05,
+      "loss": 2.5954,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9204898167439095e-05,
+      "loss": 3.1057,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9200646285981548e-05,
+      "loss": 2.9618,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9196394404524004e-05,
+      "loss": 2.9407,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.919214252306646e-05,
+      "loss": 2.7634,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9187890641608913e-05,
+      "loss": 2.6354,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.918363876015137e-05,
+      "loss": 3.3772,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9179386878693826e-05,
+      "loss": 3.1235,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.917513499723628e-05,
+      "loss": 2.9871,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9170883115778735e-05,
+      "loss": 3.3303,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.916663123432119e-05,
+      "loss": 3.1218,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9162379352863643e-05,
+      "loss": 3.142,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.91581274714061e-05,
+      "loss": 3.0239,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9153875589948552e-05,
+      "loss": 2.8272,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.914962370849101e-05,
+      "loss": 3.5088,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.284396171569824,
+      "eval_runtime": 180.2968,
+      "eval_samples_per_second": 13.883,
+      "eval_steps_per_second": 6.944,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9145371827033465e-05,
+      "loss": 3.3707,
+      "step": 2510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9141119945575917e-05,
+      "loss": 3.2181,
+      "step": 2520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9136868064118374e-05,
+      "loss": 3.0043,
+      "step": 2530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.913261618266083e-05,
+      "loss": 2.6138,
+      "step": 2540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9128364301203282e-05,
+      "loss": 3.474,
+      "step": 2550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.912411241974574e-05,
+      "loss": 2.9508,
+      "step": 2560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.911986053828819e-05,
+      "loss": 3.0993,
+      "step": 2570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9115608656830648e-05,
+      "loss": 2.964,
+      "step": 2580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9111356775373104e-05,
+      "loss": 3.2431,
+      "step": 2590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.910710489391556e-05,
+      "loss": 3.5088,
+      "step": 2600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9102853012458013e-05,
+      "loss": 2.9043,
+      "step": 2610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.909860113100047e-05,
+      "loss": 2.7854,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9094349249542925e-05,
+      "loss": 2.6917,
+      "step": 2630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9090097368085378e-05,
+      "loss": 2.8437,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9085845486627834e-05,
+      "loss": 2.9471,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.908159360517029e-05,
+      "loss": 3.3191,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9077341723712743e-05,
+      "loss": 3.2082,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.90730898422552e-05,
+      "loss": 2.9716,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9068837960797655e-05,
+      "loss": 2.7102,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.906458607934011e-05,
+      "loss": 3.2996,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9060334197882564e-05,
+      "loss": 3.1837,
+      "step": 2710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.905608231642502e-05,
+      "loss": 3.2329,
+      "step": 2720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9051830434967476e-05,
+      "loss": 2.9437,
+      "step": 2730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.904757855350993e-05,
+      "loss": 2.7768,
+      "step": 2740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9043326672052385e-05,
+      "loss": 3.008,
+      "step": 2750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.903907479059484e-05,
+      "loss": 3.2435,
+      "step": 2760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9034822909137294e-05,
+      "loss": 3.1503,
+      "step": 2770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.903057102767975e-05,
+      "loss": 2.9688,
+      "step": 2780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9026319146222206e-05,
+      "loss": 2.7462,
+      "step": 2790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9022067264764663e-05,
+      "loss": 3.0445,
+      "step": 2800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9017815383307115e-05,
+      "loss": 2.9424,
+      "step": 2810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.901356350184957e-05,
+      "loss": 3.3523,
+      "step": 2820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9009311620392024e-05,
+      "loss": 2.9877,
+      "step": 2830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.900505973893448e-05,
+      "loss": 2.8386,
+      "step": 2840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9000807857476937e-05,
+      "loss": 2.9439,
+      "step": 2850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.899655597601939e-05,
+      "loss": 3.1248,
+      "step": 2860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8992304094561845e-05,
+      "loss": 2.6066,
+      "step": 2870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8988052213104298e-05,
+      "loss": 2.7073,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8983800331646754e-05,
+      "loss": 2.8504,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.897954845018921e-05,
+      "loss": 2.951,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8975296568731663e-05,
+      "loss": 3.1729,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.897104468727412e-05,
+      "loss": 2.8276,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8966792805816576e-05,
+      "loss": 3.0242,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.896254092435903e-05,
+      "loss": 2.8324,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8958289042901485e-05,
+      "loss": 3.336,
+      "step": 2950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.895403716144394e-05,
+      "loss": 2.609,
+      "step": 2960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8949785279986393e-05,
+      "loss": 3.0654,
+      "step": 2970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.894553339852885e-05,
+      "loss": 2.9748,
+      "step": 2980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8941281517071306e-05,
+      "loss": 3.0587,
+      "step": 2990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8937029635613762e-05,
+      "loss": 2.8801,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.220201015472412,
+      "eval_runtime": 157.7813,
+      "eval_samples_per_second": 15.864,
+      "eval_steps_per_second": 7.935,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8932777754156215e-05,
+      "loss": 3.1024,
+      "step": 3010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.892852587269867e-05,
+      "loss": 2.9528,
+      "step": 3020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8924273991241127e-05,
+      "loss": 3.0732,
+      "step": 3030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.892002210978358e-05,
+      "loss": 2.8469,
+      "step": 3040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8915770228326036e-05,
+      "loss": 2.6897,
+      "step": 3050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8911518346868492e-05,
+      "loss": 3.291,
+      "step": 3060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8907266465410945e-05,
+      "loss": 2.6824,
+      "step": 3070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.89030145839534e-05,
+      "loss": 3.202,
+      "step": 3080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8898762702495857e-05,
+      "loss": 2.5288,
+      "step": 3090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.889451082103831e-05,
+      "loss": 3.4173,
+      "step": 3100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8890258939580766e-05,
+      "loss": 3.4438,
+      "step": 3110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8886007058123222e-05,
+      "loss": 2.9537,
+      "step": 3120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.888175517666568e-05,
+      "loss": 2.8619,
+      "step": 3130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.887750329520813e-05,
+      "loss": 3.2763,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8873251413750587e-05,
+      "loss": 3.1419,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8868999532293043e-05,
+      "loss": 3.1411,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8864747650835496e-05,
+      "loss": 2.9325,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8860495769377952e-05,
+      "loss": 3.1641,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.885624388792041e-05,
+      "loss": 2.8487,
+      "step": 3190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.885199200646286e-05,
+      "loss": 2.8405,
+      "step": 3200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8847740125005317e-05,
+      "loss": 2.6751,
+      "step": 3210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.884348824354777e-05,
+      "loss": 3.2453,
+      "step": 3220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8839236362090226e-05,
+      "loss": 3.1109,
+      "step": 3230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8834984480632682e-05,
+      "loss": 3.1686,
+      "step": 3240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8830732599175135e-05,
+      "loss": 3.1336,
+      "step": 3250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.882648071771759e-05,
+      "loss": 3.0892,
+      "step": 3260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8822228836260044e-05,
+      "loss": 2.9393,
+      "step": 3270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.88179769548025e-05,
+      "loss": 2.5524,
+      "step": 3280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8813725073344956e-05,
+      "loss": 3.1743,
+      "step": 3290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.880947319188741e-05,
+      "loss": 2.9226,
+      "step": 3300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8805221310429865e-05,
+      "loss": 2.862,
+      "step": 3310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.880096942897232e-05,
+      "loss": 2.6228,
+      "step": 3320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8796717547514778e-05,
+      "loss": 3.1708,
+      "step": 3330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.879246566605723e-05,
+      "loss": 2.6939,
+      "step": 3340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8788213784599687e-05,
+      "loss": 3.0765,
+      "step": 3350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8783961903142143e-05,
+      "loss": 2.9789,
+      "step": 3360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8779710021684596e-05,
+      "loss": 2.7487,
+      "step": 3370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.877545814022705e-05,
+      "loss": 2.4453,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8771206258769508e-05,
+      "loss": 3.0888,
+      "step": 3390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.876695437731196e-05,
+      "loss": 2.6603,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8762702495854417e-05,
+      "loss": 2.9912,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8758450614396873e-05,
+      "loss": 3.1868,
+      "step": 3420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.875419873293933e-05,
+      "loss": 3.0776,
+      "step": 3430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8749946851481782e-05,
+      "loss": 2.9323,
+      "step": 3440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8745694970024238e-05,
+      "loss": 3.0786,
+      "step": 3450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8741443088566694e-05,
+      "loss": 2.7427,
+      "step": 3460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8737191207109147e-05,
+      "loss": 2.6738,
+      "step": 3470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8732939325651603e-05,
+      "loss": 2.7113,
+      "step": 3480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.872868744419406e-05,
+      "loss": 2.9205,
+      "step": 3490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8724435562736512e-05,
+      "loss": 3.002,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.175995349884033,
+      "eval_runtime": 157.4547,
+      "eval_samples_per_second": 15.897,
+      "eval_steps_per_second": 7.951,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8720183681278968e-05,
+      "loss": 2.3474,
+      "step": 3510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8715931799821424e-05,
+      "loss": 2.8694,
+      "step": 3520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8711679918363877e-05,
+      "loss": 2.6717,
+      "step": 3530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8707428036906333e-05,
+      "loss": 2.9325,
+      "step": 3540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.870317615544879e-05,
+      "loss": 3.0843,
+      "step": 3550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8698924273991242e-05,
+      "loss": 3.1572,
+      "step": 3560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8694672392533698e-05,
+      "loss": 2.3833,
+      "step": 3570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8690420511076154e-05,
+      "loss": 3.0576,
+      "step": 3580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8686168629618607e-05,
+      "loss": 2.7256,
+      "step": 3590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8681916748161063e-05,
+      "loss": 2.8938,
+      "step": 3600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8677664866703516e-05,
+      "loss": 3.5121,
+      "step": 3610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8673412985245972e-05,
+      "loss": 3.0321,
+      "step": 3620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.866916110378843e-05,
+      "loss": 2.7772,
+      "step": 3630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.866490922233088e-05,
+      "loss": 2.4553,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8660657340873337e-05,
+      "loss": 2.3899,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8656405459415793e-05,
+      "loss": 2.8928,
+      "step": 3660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8652153577958246e-05,
+      "loss": 2.5838,
+      "step": 3670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8647901696500702e-05,
+      "loss": 2.8532,
+      "step": 3680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.864364981504316e-05,
+      "loss": 2.6418,
+      "step": 3690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.863939793358561e-05,
+      "loss": 2.7067,
+      "step": 3700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8635146052128067e-05,
+      "loss": 2.8495,
+      "step": 3710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8630894170670524e-05,
+      "loss": 2.57,
+      "step": 3720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8626642289212976e-05,
+      "loss": 2.8358,
+      "step": 3730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8622390407755433e-05,
+      "loss": 2.5213,
+      "step": 3740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.861813852629789e-05,
+      "loss": 2.6923,
+      "step": 3750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8613886644840345e-05,
+      "loss": 2.7872,
+      "step": 3760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8609634763382798e-05,
+      "loss": 3.0025,
+      "step": 3770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8605382881925254e-05,
+      "loss": 3.0608,
+      "step": 3780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.860113100046771e-05,
+      "loss": 2.9327,
+      "step": 3790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8596879119010163e-05,
+      "loss": 2.6853,
+      "step": 3800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.859262723755262e-05,
+      "loss": 3.0209,
+      "step": 3810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8588375356095075e-05,
+      "loss": 3.1764,
+      "step": 3820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8584123474637528e-05,
+      "loss": 2.6595,
+      "step": 3830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8579871593179984e-05,
+      "loss": 3.1195,
+      "step": 3840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.857561971172244e-05,
+      "loss": 2.4837,
+      "step": 3850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8571367830264896e-05,
+      "loss": 2.9376,
+      "step": 3860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.856711594880735e-05,
+      "loss": 2.7025,
+      "step": 3870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8562864067349805e-05,
+      "loss": 2.7312,
+      "step": 3880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.855861218589226e-05,
+      "loss": 2.7804,
+      "step": 3890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8554360304434714e-05,
+      "loss": 2.6139,
+      "step": 3900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.855010842297717e-05,
+      "loss": 2.4627,
+      "step": 3910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8545856541519623e-05,
+      "loss": 3.0103,
+      "step": 3920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.854160466006208e-05,
+      "loss": 3.1759,
+      "step": 3930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8537352778604535e-05,
+      "loss": 2.3429,
+      "step": 3940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8533100897146988e-05,
+      "loss": 2.5575,
+      "step": 3950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8528849015689444e-05,
+      "loss": 2.6588,
+      "step": 3960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85245971342319e-05,
+      "loss": 2.3116,
+      "step": 3970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8520345252774353e-05,
+      "loss": 2.6692,
+      "step": 3980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.851609337131681e-05,
+      "loss": 2.4895,
+      "step": 3990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8511841489859262e-05,
+      "loss": 2.5881,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.1984293460845947,
+      "eval_runtime": 157.7819,
+      "eval_samples_per_second": 15.864,
+      "eval_steps_per_second": 7.935,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8507589608401718e-05,
+      "loss": 3.1104,
+      "step": 4010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8503337726944174e-05,
+      "loss": 2.9546,
+      "step": 4020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8499085845486627e-05,
+      "loss": 2.5194,
+      "step": 4030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8494833964029083e-05,
+      "loss": 2.501,
+      "step": 4040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.849058208257154e-05,
+      "loss": 2.9073,
+      "step": 4050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8486330201113992e-05,
+      "loss": 2.5472,
+      "step": 4060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8482078319656448e-05,
+      "loss": 3.0469,
+      "step": 4070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8477826438198904e-05,
+      "loss": 2.54,
+      "step": 4080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.847357455674136e-05,
+      "loss": 2.5786,
+      "step": 4090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8469322675283813e-05,
+      "loss": 3.0802,
+      "step": 4100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.846507079382627e-05,
+      "loss": 2.5621,
+      "step": 4110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8460818912368726e-05,
+      "loss": 2.7328,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.845656703091118e-05,
+      "loss": 3.0167,
+      "step": 4130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8452315149453635e-05,
+      "loss": 2.6309,
+      "step": 4140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.844806326799609e-05,
+      "loss": 2.6624,
+      "step": 4150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8443811386538543e-05,
+      "loss": 2.439,
+      "step": 4160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8439559505081e-05,
+      "loss": 3.1749,
+      "step": 4170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8435307623623456e-05,
+      "loss": 3.0303,
+      "step": 4180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8431055742165912e-05,
+      "loss": 2.8716,
+      "step": 4190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8426803860708365e-05,
+      "loss": 2.6538,
+      "step": 4200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842255197925082e-05,
+      "loss": 2.8078,
+      "step": 4210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8418300097793277e-05,
+      "loss": 2.5088,
+      "step": 4220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841404821633573e-05,
+      "loss": 2.9373,
+      "step": 4230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8409796334878186e-05,
+      "loss": 3.0147,
+      "step": 4240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8405544453420642e-05,
+      "loss": 3.4054,
+      "step": 4250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8401292571963095e-05,
+      "loss": 3.2158,
+      "step": 4260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839704069050555e-05,
+      "loss": 2.3821,
+      "step": 4270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8392788809048007e-05,
+      "loss": 2.7904,
+      "step": 4280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838853692759046e-05,
+      "loss": 2.9567,
+      "step": 4290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8384285046132916e-05,
+      "loss": 2.2942,
+      "step": 4300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838003316467537e-05,
+      "loss": 2.965,
+      "step": 4310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8375781283217825e-05,
+      "loss": 2.6916,
+      "step": 4320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837152940176028e-05,
+      "loss": 2.5174,
+      "step": 4330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8367277520302734e-05,
+      "loss": 2.5812,
+      "step": 4340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836302563884519e-05,
+      "loss": 2.8423,
+      "step": 4350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8358773757387646e-05,
+      "loss": 2.5655,
+      "step": 4360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.83545218759301e-05,
+      "loss": 2.3715,
+      "step": 4370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8350269994472555e-05,
+      "loss": 2.8404,
+      "step": 4380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.834601811301501e-05,
+      "loss": 2.9515,
+      "step": 4390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8341766231557464e-05,
+      "loss": 2.5119,
+      "step": 4400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833751435009992e-05,
+      "loss": 2.5983,
+      "step": 4410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8333262468642376e-05,
+      "loss": 2.7768,
+      "step": 4420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832901058718483e-05,
+      "loss": 2.9894,
+      "step": 4430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8324758705727285e-05,
+      "loss": 2.5563,
+      "step": 4440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832050682426974e-05,
+      "loss": 2.8879,
+      "step": 4450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8316254942812194e-05,
+      "loss": 2.6981,
+      "step": 4460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831200306135465e-05,
+      "loss": 3.0622,
+      "step": 4470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8307751179897106e-05,
+      "loss": 2.883,
+      "step": 4480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8303499298439563e-05,
+      "loss": 2.9416,
+      "step": 4490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8299247416982015e-05,
+      "loss": 2.8995,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 2.1155664920806885,
+      "eval_runtime": 157.6562,
+      "eval_samples_per_second": 15.876,
+      "eval_steps_per_second": 7.941,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829499553552447e-05,
+      "loss": 3.3596,
+      "step": 4510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8290743654066928e-05,
+      "loss": 2.5666,
+      "step": 4520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828649177260938e-05,
+      "loss": 2.7884,
+      "step": 4530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8282239891151837e-05,
+      "loss": 2.5327,
+      "step": 4540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8277988009694293e-05,
+      "loss": 2.4704,
+      "step": 4550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8273736128236746e-05,
+      "loss": 3.302,
+      "step": 4560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82694842467792e-05,
+      "loss": 2.9727,
+      "step": 4570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8265232365321658e-05,
+      "loss": 2.4977,
+      "step": 4580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.826098048386411e-05,
+      "loss": 2.5478,
+      "step": 4590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8256728602406567e-05,
+      "loss": 3.0426,
+      "step": 4600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8252476720949023e-05,
+      "loss": 2.6851,
+      "step": 4610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824822483949148e-05,
+      "loss": 2.7621,
+      "step": 4620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8243972958033932e-05,
+      "loss": 2.8677,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8239721076576388e-05,
+      "loss": 2.7745,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823546919511884e-05,
+      "loss": 2.7964,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8231217313661297e-05,
+      "loss": 2.7733,
+      "step": 4660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8226965432203753e-05,
+      "loss": 2.7332,
+      "step": 4670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8222713550746206e-05,
+      "loss": 2.8351,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8218461669288662e-05,
+      "loss": 2.9184,
+      "step": 4690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8214209787831115e-05,
+      "loss": 2.775,
+      "step": 4700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820995790637357e-05,
+      "loss": 2.6668,
+      "step": 4710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8205706024916027e-05,
+      "loss": 2.5477,
+      "step": 4720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820145414345848e-05,
+      "loss": 2.5597,
+      "step": 4730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8197202262000936e-05,
+      "loss": 2.6637,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8192950380543392e-05,
+      "loss": 2.5096,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8188698499085845e-05,
+      "loss": 2.9088,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81844466176283e-05,
+      "loss": 3.2694,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8180194736170757e-05,
+      "loss": 2.3419,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817594285471321e-05,
+      "loss": 2.3274,
+      "step": 4790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8171690973255666e-05,
+      "loss": 2.5307,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8167439091798122e-05,
+      "loss": 2.7317,
+      "step": 4810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816318721034058e-05,
+      "loss": 2.4159,
+      "step": 4820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815893532888303e-05,
+      "loss": 3.0825,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8154683447425487e-05,
+      "loss": 2.4835,
+      "step": 4840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8150431565967943e-05,
+      "loss": 2.3891,
+      "step": 4850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8146179684510396e-05,
+      "loss": 2.445,
+      "step": 4860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8141927803052852e-05,
+      "loss": 2.457,
+      "step": 4870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.813767592159531e-05,
+      "loss": 2.534,
+      "step": 4880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.813342404013776e-05,
+      "loss": 2.7876,
+      "step": 4890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8129172158680217e-05,
+      "loss": 2.847,
+      "step": 4900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8124920277222674e-05,
+      "loss": 2.7351,
+      "step": 4910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.812066839576513e-05,
+      "loss": 2.5956,
+      "step": 4920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8116416514307583e-05,
+      "loss": 2.9122,
+      "step": 4930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.811216463285004e-05,
+      "loss": 2.2897,
+      "step": 4940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8107912751392495e-05,
+      "loss": 2.7125,
+      "step": 4950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8103660869934948e-05,
+      "loss": 2.6486,
+      "step": 4960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8099408988477404e-05,
+      "loss": 2.5903,
+      "step": 4970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809515710701986e-05,
+      "loss": 2.3824,
+      "step": 4980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8090905225562313e-05,
+      "loss": 2.87,
+      "step": 4990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808665334410477e-05,
+      "loss": 2.8354,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 2.103600025177002,
+      "eval_runtime": 158.2313,
+      "eval_samples_per_second": 15.819,
+      "eval_steps_per_second": 7.912,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8082401462647225e-05,
+      "loss": 2.6096,
+      "step": 5010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8078149581189678e-05,
+      "loss": 2.8438,
+      "step": 5020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8073897699732134e-05,
+      "loss": 2.5016,
+      "step": 5030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8069645818274587e-05,
+      "loss": 2.7681,
+      "step": 5040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8065393936817043e-05,
+      "loss": 2.9139,
+      "step": 5050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.80611420553595e-05,
+      "loss": 2.7467,
+      "step": 5060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8056890173901952e-05,
+      "loss": 2.5214,
+      "step": 5070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8052638292444408e-05,
+      "loss": 2.5733,
+      "step": 5080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804838641098686e-05,
+      "loss": 2.6025,
+      "step": 5090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8044134529529317e-05,
+      "loss": 3.0755,
+      "step": 5100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8039882648071773e-05,
+      "loss": 2.4955,
+      "step": 5110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8035630766614226e-05,
+      "loss": 2.347,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8031378885156682e-05,
+      "loss": 2.984,
+      "step": 5130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8027127003699138e-05,
+      "loss": 2.4085,
+      "step": 5140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8022875122241594e-05,
+      "loss": 2.6538,
+      "step": 5150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8018623240784047e-05,
+      "loss": 2.5329,
+      "step": 5160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8014371359326503e-05,
+      "loss": 2.6974,
+      "step": 5170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.801011947786896e-05,
+      "loss": 2.6907,
+      "step": 5180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8005867596411412e-05,
+      "loss": 3.0843,
+      "step": 5190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8001615714953868e-05,
+      "loss": 2.7841,
+      "step": 5200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7997363833496324e-05,
+      "loss": 2.7677,
+      "step": 5210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7993111952038777e-05,
+      "loss": 2.4248,
+      "step": 5220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7988860070581233e-05,
+      "loss": 2.6094,
+      "step": 5230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.798460818912369e-05,
+      "loss": 2.8525,
+      "step": 5240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7980356307666146e-05,
+      "loss": 3.099,
+      "step": 5250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7976104426208598e-05,
+      "loss": 2.2656,
+      "step": 5260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7971852544751054e-05,
+      "loss": 2.7402,
+      "step": 5270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.796760066329351e-05,
+      "loss": 2.9918,
+      "step": 5280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7963348781835963e-05,
+      "loss": 2.81,
+      "step": 5290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.795909690037842e-05,
+      "loss": 2.5187,
+      "step": 5300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7954845018920876e-05,
+      "loss": 2.6014,
+      "step": 5310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.795059313746333e-05,
+      "loss": 3.0447,
+      "step": 5320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7946341256005785e-05,
+      "loss": 3.066,
+      "step": 5330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.794208937454824e-05,
+      "loss": 2.7345,
+      "step": 5340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937837493090693e-05,
+      "loss": 2.7044,
+      "step": 5350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793358561163315e-05,
+      "loss": 2.748,
+      "step": 5360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929333730175606e-05,
+      "loss": 2.5782,
+      "step": 5370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792508184871806e-05,
+      "loss": 2.9447,
+      "step": 5380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920829967260515e-05,
+      "loss": 2.8424,
+      "step": 5390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791657808580297e-05,
+      "loss": 2.8229,
+      "step": 5400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912326204345424e-05,
+      "loss": 2.6956,
+      "step": 5410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790807432288788e-05,
+      "loss": 2.7828,
+      "step": 5420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903822441430333e-05,
+      "loss": 2.9823,
+      "step": 5430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789957055997279e-05,
+      "loss": 3.0025,
+      "step": 5440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895318678515245e-05,
+      "loss": 2.6832,
+      "step": 5450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891066797057698e-05,
+      "loss": 2.6522,
+      "step": 5460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886814915600154e-05,
+      "loss": 2.4215,
+      "step": 5470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788256303414261e-05,
+      "loss": 2.8462,
+      "step": 5480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878311152685063e-05,
+      "loss": 2.2992,
+      "step": 5490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787405927122752e-05,
+      "loss": 2.7561,
+      "step": 5500
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 2.066274404525757,
+      "eval_runtime": 175.5176,
+      "eval_samples_per_second": 14.261,
+      "eval_steps_per_second": 7.133,
+      "step": 5500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7869807389769975e-05,
+      "loss": 2.6396,
+      "step": 5510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7865555508312428e-05,
+      "loss": 2.6427,
+      "step": 5520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7861303626854884e-05,
+      "loss": 3.0599,
+      "step": 5530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.785705174539734e-05,
+      "loss": 2.7358,
+      "step": 5540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7852799863939796e-05,
+      "loss": 2.8105,
+      "step": 5550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.784854798248225e-05,
+      "loss": 2.6185,
+      "step": 5560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7844296101024705e-05,
+      "loss": 2.4536,
+      "step": 5570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.784004421956716e-05,
+      "loss": 2.9464,
+      "step": 5580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835792338109614e-05,
+      "loss": 2.4645,
+      "step": 5590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783154045665207e-05,
+      "loss": 2.913,
+      "step": 5600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827288575194526e-05,
+      "loss": 2.6963,
+      "step": 5610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782303669373698e-05,
+      "loss": 2.5597,
+      "step": 5620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818784812279435e-05,
+      "loss": 2.9756,
+      "step": 5630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781453293082189e-05,
+      "loss": 2.9978,
+      "step": 5640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810281049364344e-05,
+      "loss": 2.6279,
+      "step": 5650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78060291679068e-05,
+      "loss": 2.6256,
+      "step": 5660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801777286449256e-05,
+      "loss": 2.6303,
+      "step": 5670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797525404991713e-05,
+      "loss": 2.6303,
+      "step": 5680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793273523534165e-05,
+      "loss": 3.2514,
+      "step": 5690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778902164207662e-05,
+      "loss": 2.5286,
+      "step": 5700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784769760619078e-05,
+      "loss": 2.751,
+      "step": 5710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778051787916153e-05,
+      "loss": 2.4097,
+      "step": 5720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776265997703987e-05,
+      "loss": 2.733,
+      "step": 5730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777201411624644e-05,
+      "loss": 2.4291,
+      "step": 5740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767762234788896e-05,
+      "loss": 2.1838,
+      "step": 5750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776351035333135e-05,
+      "loss": 2.5028,
+      "step": 5760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759258471873804e-05,
+      "loss": 2.741,
+      "step": 5770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775500659041626e-05,
+      "loss": 2.5766,
+      "step": 5780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750754708958717e-05,
+      "loss": 3.1094,
+      "step": 5790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774650282750117e-05,
+      "loss": 2.4569,
+      "step": 5800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742250946043626e-05,
+      "loss": 2.9686,
+      "step": 5810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773799906458608e-05,
+      "loss": 2.9987,
+      "step": 5820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7733747183128535e-05,
+      "loss": 2.411,
+      "step": 5830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.772949530167099e-05,
+      "loss": 2.8807,
+      "step": 5840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7725243420213444e-05,
+      "loss": 2.774,
+      "step": 5850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.77209915387559e-05,
+      "loss": 2.6391,
+      "step": 5860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7716739657298356e-05,
+      "loss": 2.375,
+      "step": 5870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7712487775840812e-05,
+      "loss": 2.4736,
+      "step": 5880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7708235894383265e-05,
+      "loss": 2.9305,
+      "step": 5890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.770398401292572e-05,
+      "loss": 2.6085,
+      "step": 5900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7699732131468177e-05,
+      "loss": 2.6392,
+      "step": 5910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.769548025001063e-05,
+      "loss": 2.0514,
+      "step": 5920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7691228368553086e-05,
+      "loss": 2.3497,
+      "step": 5930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7686976487095542e-05,
+      "loss": 2.774,
+      "step": 5940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7682724605637995e-05,
+      "loss": 3.2576,
+      "step": 5950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767847272418045e-05,
+      "loss": 2.362,
+      "step": 5960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7674220842722907e-05,
+      "loss": 2.0393,
+      "step": 5970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669968961265363e-05,
+      "loss": 2.8308,
+      "step": 5980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665717079807816e-05,
+      "loss": 2.5397,
+      "step": 5990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661465198350272e-05,
+      "loss": 2.685,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 2.057528257369995,
+      "eval_runtime": 175.9189,
+      "eval_samples_per_second": 14.228,
+      "eval_steps_per_second": 7.117,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765721331689273e-05,
+      "loss": 2.747,
+      "step": 6010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765296143543518e-05,
+      "loss": 2.8322,
+      "step": 6020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648709553977637e-05,
+      "loss": 2.3471,
+      "step": 6030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644457672520093e-05,
+      "loss": 2.8333,
+      "step": 6040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640205791062546e-05,
+      "loss": 3.0349,
+      "step": 6050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635953909605002e-05,
+      "loss": 2.3392,
+      "step": 6060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.763170202814746e-05,
+      "loss": 3.0696,
+      "step": 6070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.762745014668991e-05,
+      "loss": 2.8775,
+      "step": 6080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7623198265232367e-05,
+      "loss": 3.0682,
+      "step": 6090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7618946383774824e-05,
+      "loss": 2.7299,
+      "step": 6100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7614694502317276e-05,
+      "loss": 2.4431,
+      "step": 6110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7610442620859733e-05,
+      "loss": 2.4789,
+      "step": 6120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7606190739402185e-05,
+      "loss": 3.0541,
+      "step": 6130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.760193885794464e-05,
+      "loss": 2.7687,
+      "step": 6140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7597686976487098e-05,
+      "loss": 2.5623,
+      "step": 6150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.759343509502955e-05,
+      "loss": 2.5698,
+      "step": 6160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7589183213572007e-05,
+      "loss": 3.0957,
+      "step": 6170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7584931332114463e-05,
+      "loss": 3.0593,
+      "step": 6180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7580679450656915e-05,
+      "loss": 2.4348,
+      "step": 6190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.757642756919937e-05,
+      "loss": 2.7781,
+      "step": 6200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7572175687741828e-05,
+      "loss": 2.6811,
+      "step": 6210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.756792380628428e-05,
+      "loss": 2.0182,
+      "step": 6220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7563671924826737e-05,
+      "loss": 3.3047,
+      "step": 6230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7559420043369193e-05,
+      "loss": 3.0397,
+      "step": 6240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7555168161911646e-05,
+      "loss": 2.6576,
+      "step": 6250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7550916280454102e-05,
+      "loss": 2.9801,
+      "step": 6260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7546664398996558e-05,
+      "loss": 2.2294,
+      "step": 6270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.754241251753901e-05,
+      "loss": 2.9889,
+      "step": 6280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7538160636081467e-05,
+      "loss": 2.264,
+      "step": 6290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7533908754623923e-05,
+      "loss": 2.8822,
+      "step": 6300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.752965687316638e-05,
+      "loss": 3.1016,
+      "step": 6310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7525404991708832e-05,
+      "loss": 2.3495,
+      "step": 6320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7521153110251288e-05,
+      "loss": 2.54,
+      "step": 6330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7516901228793744e-05,
+      "loss": 2.4479,
+      "step": 6340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7512649347336197e-05,
+      "loss": 2.154,
+      "step": 6350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7508397465878653e-05,
+      "loss": 2.6394,
+      "step": 6360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.750414558442111e-05,
+      "loss": 3.0602,
+      "step": 6370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7499893702963562e-05,
+      "loss": 2.1946,
+      "step": 6380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7495641821506018e-05,
+      "loss": 2.7575,
+      "step": 6390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7491389940048474e-05,
+      "loss": 2.4044,
+      "step": 6400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.748713805859093e-05,
+      "loss": 2.5772,
+      "step": 6410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7482886177133383e-05,
+      "loss": 2.7697,
+      "step": 6420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.747863429567584e-05,
+      "loss": 2.5757,
+      "step": 6430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7474382414218296e-05,
+      "loss": 2.6727,
+      "step": 6440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7470130532760748e-05,
+      "loss": 2.8515,
+      "step": 6450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7465878651303204e-05,
+      "loss": 2.0677,
+      "step": 6460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7461626769845657e-05,
+      "loss": 2.3963,
+      "step": 6470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7457374888388113e-05,
+      "loss": 2.6843,
+      "step": 6480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.745312300693057e-05,
+      "loss": 2.6221,
+      "step": 6490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7448871125473022e-05,
+      "loss": 2.4269,
+      "step": 6500
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 2.064359664916992,
+      "eval_runtime": 174.2548,
+      "eval_samples_per_second": 14.364,
+      "eval_steps_per_second": 7.185,
+      "step": 6500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.744461924401548e-05,
+      "loss": 2.4325,
+      "step": 6510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.744036736255793e-05,
+      "loss": 2.9225,
+      "step": 6520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7436115481100387e-05,
+      "loss": 2.3408,
+      "step": 6530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7431863599642843e-05,
+      "loss": 2.4789,
+      "step": 6540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7427611718185296e-05,
+      "loss": 2.8185,
+      "step": 6550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7423359836727752e-05,
+      "loss": 2.8567,
+      "step": 6560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.741910795527021e-05,
+      "loss": 5.9013,
+      "step": 6570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.741485607381266e-05,
+      "loss": 2.8576,
+      "step": 6580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7410604192355117e-05,
+      "loss": 2.1402,
+      "step": 6590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7406352310897574e-05,
+      "loss": 2.3285,
+      "step": 6600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7402100429440026e-05,
+      "loss": 2.6474,
+      "step": 6610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7397848547982483e-05,
+      "loss": 2.8244,
+      "step": 6620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.739359666652494e-05,
+      "loss": 2.8179,
+      "step": 6630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7389344785067395e-05,
+      "loss": 2.7042,
+      "step": 6640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7385092903609848e-05,
+      "loss": 2.3928,
+      "step": 6650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7380841022152304e-05,
+      "loss": 2.4957,
+      "step": 6660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.737658914069476e-05,
+      "loss": 2.7624,
+      "step": 6670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7372337259237213e-05,
+      "loss": 2.8772,
+      "step": 6680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.736808537777967e-05,
+      "loss": 2.9064,
+      "step": 6690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7363833496322125e-05,
+      "loss": 2.9803,
+      "step": 6700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7359581614864578e-05,
+      "loss": 2.3177,
+      "step": 6710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7355329733407034e-05,
+      "loss": 2.6294,
+      "step": 6720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.735107785194949e-05,
+      "loss": 2.4746,
+      "step": 6730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7346825970491946e-05,
+      "loss": 1.9463,
+      "step": 6740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.73425740890344e-05,
+      "loss": 2.6038,
+      "step": 6750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7338322207576855e-05,
+      "loss": 2.3591,
+      "step": 6760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.733407032611931e-05,
+      "loss": 2.5867,
+      "step": 6770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7329818444661764e-05,
+      "loss": 2.3575,
+      "step": 6780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.732556656320422e-05,
+      "loss": 2.5036,
+      "step": 6790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7321314681746676e-05,
+      "loss": 2.3801,
+      "step": 6800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.731706280028913e-05,
+      "loss": 2.2803,
+      "step": 6810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7312810918831585e-05,
+      "loss": 2.5604,
+      "step": 6820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.730855903737404e-05,
+      "loss": 2.3178,
+      "step": 6830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7304307155916494e-05,
+      "loss": 2.466,
+      "step": 6840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.730005527445895e-05,
+      "loss": 2.7137,
+      "step": 6850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7295803393001403e-05,
+      "loss": 2.4662,
+      "step": 6860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.729155151154386e-05,
+      "loss": 2.5953,
+      "step": 6870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7287299630086315e-05,
+      "loss": 2.8784,
+      "step": 6880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7283047748628768e-05,
+      "loss": 2.4881,
+      "step": 6890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7278795867171224e-05,
+      "loss": 2.57,
+      "step": 6900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7274543985713677e-05,
+      "loss": 3.0467,
+      "step": 6910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7270292104256133e-05,
+      "loss": 2.1957,
+      "step": 6920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.726604022279859e-05,
+      "loss": 2.5759,
+      "step": 6930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7261788341341046e-05,
+      "loss": 2.3922,
+      "step": 6940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7257536459883498e-05,
+      "loss": 2.7761,
+      "step": 6950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7253284578425954e-05,
+      "loss": 2.3974,
+      "step": 6960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.724903269696841e-05,
+      "loss": 2.8827,
+      "step": 6970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7244780815510863e-05,
+      "loss": 2.5402,
+      "step": 6980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.724052893405332e-05,
+      "loss": 2.4587,
+      "step": 6990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7236277052595776e-05,
+      "loss": 2.3543,
+      "step": 7000
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 2.0410194396972656,
+      "eval_runtime": 208.3814,
+      "eval_samples_per_second": 12.012,
+      "eval_steps_per_second": 6.008,
+      "step": 7000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.723202517113823e-05,
+      "loss": 2.6713,
+      "step": 7010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7227773289680685e-05,
+      "loss": 2.7262,
+      "step": 7020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.722352140822314e-05,
+      "loss": 2.8134,
+      "step": 7030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7219269526765597e-05,
+      "loss": 2.3668,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.721501764530805e-05,
+      "loss": 2.6703,
+      "step": 7050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7210765763850506e-05,
+      "loss": 2.7111,
+      "step": 7060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7206513882392962e-05,
+      "loss": 2.4478,
+      "step": 7070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7202262000935415e-05,
+      "loss": 2.1211,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.719801011947787e-05,
+      "loss": 2.6789,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7193758238020327e-05,
+      "loss": 2.5602,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.718950635656278e-05,
+      "loss": 3.0027,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7185254475105236e-05,
+      "loss": 2.3461,
+      "step": 7120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7181002593647692e-05,
+      "loss": 2.7754,
+      "step": 7130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7176750712190145e-05,
+      "loss": 2.5173,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.71724988307326e-05,
+      "loss": 2.3433,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7168246949275057e-05,
+      "loss": 2.2877,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.716399506781751e-05,
+      "loss": 2.5632,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7159743186359966e-05,
+      "loss": 2.6671,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7155491304902422e-05,
+      "loss": 2.5462,
+      "step": 7190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7151239423444875e-05,
+      "loss": 2.7029,
+      "step": 7200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.714698754198733e-05,
+      "loss": 2.1722,
+      "step": 7210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7142735660529787e-05,
+      "loss": 2.6146,
+      "step": 7220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.713848377907224e-05,
+      "loss": 2.695,
+      "step": 7230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.7134231897614696e-05,
+      "loss": 2.782,
+      "step": 7240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.712998001615715e-05,
+      "loss": 2.7765,
+      "step": 7250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7125728134699605e-05,
+      "loss": 2.8255,
+      "step": 7260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.712147625324206e-05,
+      "loss": 2.662,
+      "step": 7270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7117224371784514e-05,
+      "loss": 2.9787,
+      "step": 7280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.711297249032697e-05,
+      "loss": 2.7982,
+      "step": 7290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7108720608869426e-05,
+      "loss": 2.363,
+      "step": 7300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.710446872741188e-05,
+      "loss": 2.5674,
+      "step": 7310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7100216845954335e-05,
+      "loss": 2.5751,
+      "step": 7320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.709596496449679e-05,
+      "loss": 2.8134,
+      "step": 7330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7091713083039244e-05,
+      "loss": 2.2201,
+      "step": 7340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.70874612015817e-05,
+      "loss": 2.2872,
+      "step": 7350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7083209320124157e-05,
+      "loss": 2.5633,
+      "step": 7360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7078957438666613e-05,
+      "loss": 2.7733,
+      "step": 7370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7074705557209065e-05,
+      "loss": 2.8384,
+      "step": 7380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.707045367575152e-05,
+      "loss": 2.5711,
+      "step": 7390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7066201794293978e-05,
+      "loss": 2.4719,
+      "step": 7400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.706194991283643e-05,
+      "loss": 2.2026,
+      "step": 7410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7057698031378887e-05,
+      "loss": 2.845,
+      "step": 7420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7053446149921343e-05,
+      "loss": 2.4645,
+      "step": 7430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7049194268463796e-05,
+      "loss": 2.6133,
+      "step": 7440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7044942387006252e-05,
+      "loss": 2.2499,
+      "step": 7450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7040690505548708e-05,
+      "loss": 2.8703,
+      "step": 7460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7036438624091164e-05,
+      "loss": 2.6793,
+      "step": 7470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.7032186742633617e-05,
+      "loss": 2.7365,
+      "step": 7480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7027934861176073e-05,
+      "loss": 2.6857,
+      "step": 7490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.702368297971853e-05,
+      "loss": 2.5896,
+      "step": 7500
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 2.008735179901123,
+      "eval_runtime": 320.0336,
+      "eval_samples_per_second": 7.821,
+      "eval_steps_per_second": 3.912,
+      "step": 7500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7019431098260982e-05,
+      "loss": 2.4888,
+      "step": 7510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7015179216803438e-05,
+      "loss": 2.4542,
+      "step": 7520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7010927335345894e-05,
+      "loss": 2.6477,
+      "step": 7530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7006675453888347e-05,
+      "loss": 2.9978,
+      "step": 7540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7002423572430803e-05,
+      "loss": 2.6587,
+      "step": 7550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6998171690973256e-05,
+      "loss": 2.8396,
+      "step": 7560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6993919809515712e-05,
+      "loss": 2.4553,
+      "step": 7570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6989667928058168e-05,
+      "loss": 2.534,
+      "step": 7580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.698541604660062e-05,
+      "loss": 2.2884,
+      "step": 7590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6981164165143077e-05,
+      "loss": 2.4381,
+      "step": 7600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6976912283685533e-05,
+      "loss": 2.7952,
+      "step": 7610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6972660402227986e-05,
+      "loss": 2.6889,
+      "step": 7620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6968408520770442e-05,
+      "loss": 2.1633,
+      "step": 7630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6964156639312895e-05,
+      "loss": 2.5299,
+      "step": 7640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.695990475785535e-05,
+      "loss": 2.7168,
+      "step": 7650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6955652876397807e-05,
+      "loss": 2.0173,
+      "step": 7660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.695140099494026e-05,
+      "loss": 2.6909,
+      "step": 7670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6947149113482716e-05,
+      "loss": 2.5593,
+      "step": 7680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6942897232025172e-05,
+      "loss": 2.5208,
+      "step": 7690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.693864535056763e-05,
+      "loss": 2.2565,
+      "step": 7700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.693439346911008e-05,
+      "loss": 2.5508,
+      "step": 7710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6930141587652537e-05,
+      "loss": 2.5953,
+      "step": 7720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6925889706194993e-05,
+      "loss": 2.8496,
+      "step": 7730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6921637824737446e-05,
+      "loss": 2.5317,
+      "step": 7740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6917385943279902e-05,
+      "loss": 2.7538,
+      "step": 7750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.691313406182236e-05,
+      "loss": 2.4385,
+      "step": 7760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.690888218036481e-05,
+      "loss": 2.5723,
+      "step": 7770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6904630298907267e-05,
+      "loss": 2.6366,
+      "step": 7780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6900378417449724e-05,
+      "loss": 2.937,
+      "step": 7790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.689612653599218e-05,
+      "loss": 2.8732,
+      "step": 7800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6891874654534633e-05,
+      "loss": 2.6569,
+      "step": 7810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.688762277307709e-05,
+      "loss": 2.6205,
+      "step": 7820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6883370891619545e-05,
+      "loss": 2.6153,
+      "step": 7830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6879119010161998e-05,
+      "loss": 2.3836,
+      "step": 7840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6874867128704454e-05,
+      "loss": 2.1343,
+      "step": 7850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.687061524724691e-05,
+      "loss": 2.1818,
+      "step": 7860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6866363365789363e-05,
+      "loss": 2.4748,
+      "step": 7870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.686211148433182e-05,
+      "loss": 2.3274,
+      "step": 7880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6857859602874275e-05,
+      "loss": 3.206,
+      "step": 7890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6853607721416728e-05,
+      "loss": 2.2356,
+      "step": 7900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6849355839959184e-05,
+      "loss": 2.7072,
+      "step": 7910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.684510395850164e-05,
+      "loss": 2.6011,
+      "step": 7920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6840852077044093e-05,
+      "loss": 2.3916,
+      "step": 7930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.683660019558655e-05,
+      "loss": 2.1616,
+      "step": 7940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6832348314129002e-05,
+      "loss": 2.4811,
+      "step": 7950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6828096432671458e-05,
+      "loss": 2.5085,
+      "step": 7960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6823844551213914e-05,
+      "loss": 2.4976,
+      "step": 7970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6819592669756367e-05,
+      "loss": 2.4867,
+      "step": 7980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6815340788298823e-05,
+      "loss": 2.5794,
+      "step": 7990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.681108890684128e-05,
+      "loss": 2.6095,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 2.00496768951416,
+      "eval_runtime": 337.7379,
+      "eval_samples_per_second": 7.411,
+      "eval_steps_per_second": 3.707,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6806837025383732e-05,
+      "loss": 2.1561,
+      "step": 8010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6802585143926188e-05,
+      "loss": 2.4532,
+      "step": 8020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6798333262468644e-05,
+      "loss": 2.2522,
+      "step": 8030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6794081381011097e-05,
+      "loss": 2.3841,
+      "step": 8040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6789829499553553e-05,
+      "loss": 2.3188,
+      "step": 8050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.678557761809601e-05,
+      "loss": 2.5529,
+      "step": 8060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6781325736638462e-05,
+      "loss": 3.0713,
+      "step": 8070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6777073855180918e-05,
+      "loss": 2.6333,
+      "step": 8080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6772821973723374e-05,
+      "loss": 2.4813,
+      "step": 8090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.676857009226583e-05,
+      "loss": 2.3983,
+      "step": 8100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6764318210808283e-05,
+      "loss": 2.2578,
+      "step": 8110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.676006632935074e-05,
+      "loss": 2.7892,
+      "step": 8120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6755814447893196e-05,
+      "loss": 2.6201,
+      "step": 8130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.675156256643565e-05,
+      "loss": 2.3545,
+      "step": 8140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6747310684978104e-05,
+      "loss": 2.475,
+      "step": 8150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.674305880352056e-05,
+      "loss": 2.7158,
+      "step": 8160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6738806922063013e-05,
+      "loss": 2.1479,
+      "step": 8170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.673455504060547e-05,
+      "loss": 2.6363,
+      "step": 8180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.6730303159147926e-05,
+      "loss": 2.0257,
+      "step": 8190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.672605127769038e-05,
+      "loss": 2.2188,
+      "step": 8200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6721799396232835e-05,
+      "loss": 2.8021,
+      "step": 8210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.671754751477529e-05,
+      "loss": 2.7151,
+      "step": 8220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6713295633317747e-05,
+      "loss": 2.3472,
+      "step": 8230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.67090437518602e-05,
+      "loss": 2.5458,
+      "step": 8240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6704791870402656e-05,
+      "loss": 2.3873,
+      "step": 8250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6700539988945112e-05,
+      "loss": 2.4865,
+      "step": 8260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6696288107487565e-05,
+      "loss": 2.7881,
+      "step": 8270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.669203622603002e-05,
+      "loss": 2.3015,
+      "step": 8280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6687784344572474e-05,
+      "loss": 2.5846,
+      "step": 8290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.668353246311493e-05,
+      "loss": 2.5504,
+      "step": 8300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6679280581657386e-05,
+      "loss": 2.3062,
+      "step": 8310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.667502870019984e-05,
+      "loss": 2.7249,
+      "step": 8320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6670776818742295e-05,
+      "loss": 2.4261,
+      "step": 8330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6666524937284748e-05,
+      "loss": 2.3477,
+      "step": 8340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6662273055827204e-05,
+      "loss": 2.2926,
+      "step": 8350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.665802117436966e-05,
+      "loss": 2.4989,
+      "step": 8360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6653769292912113e-05,
+      "loss": 2.6045,
+      "step": 8370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.664951741145457e-05,
+      "loss": 2.5208,
+      "step": 8380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6645265529997025e-05,
+      "loss": 2.3486,
+      "step": 8390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6641013648539478e-05,
+      "loss": 2.5779,
+      "step": 8400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6636761767081934e-05,
+      "loss": 2.395,
+      "step": 8410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.663250988562439e-05,
+      "loss": 2.2631,
+      "step": 8420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6628258004166846e-05,
+      "loss": 2.5388,
+      "step": 8430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.66240061227093e-05,
+      "loss": 2.5898,
+      "step": 8440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6619754241251755e-05,
+      "loss": 2.5621,
+      "step": 8450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.661550235979421e-05,
+      "loss": 2.1611,
+      "step": 8460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6611250478336664e-05,
+      "loss": 2.3162,
+      "step": 8470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.660699859687912e-05,
+      "loss": 2.9968,
+      "step": 8480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6602746715421576e-05,
+      "loss": 2.318,
+      "step": 8490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.659849483396403e-05,
+      "loss": 3.1466,
+      "step": 8500
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9874720573425293,
+      "eval_runtime": 335.1236,
+      "eval_samples_per_second": 7.469,
+      "eval_steps_per_second": 3.736,
+      "step": 8500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6594242952506485e-05,
+      "loss": 2.4721,
+      "step": 8510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.658999107104894e-05,
+      "loss": 2.2058,
+      "step": 8520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6585739189591398e-05,
+      "loss": 2.9404,
+      "step": 8530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.658148730813385e-05,
+      "loss": 2.6556,
+      "step": 8540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6577235426676307e-05,
+      "loss": 2.3895,
+      "step": 8550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6572983545218763e-05,
+      "loss": 2.4395,
+      "step": 8560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6568731663761215e-05,
+      "loss": 2.4652,
+      "step": 8570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.656447978230367e-05,
+      "loss": 2.3942,
+      "step": 8580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6560227900846128e-05,
+      "loss": 2.5459,
+      "step": 8590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.655597601938858e-05,
+      "loss": 2.2158,
+      "step": 8600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6551724137931037e-05,
+      "loss": 2.7693,
+      "step": 8610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6547472256473493e-05,
+      "loss": 2.6083,
+      "step": 8620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6543220375015946e-05,
+      "loss": 2.6171,
+      "step": 8630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6538968493558402e-05,
+      "loss": 1.9984,
+      "step": 8640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6534716612100858e-05,
+      "loss": 2.5385,
+      "step": 8650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.653046473064331e-05,
+      "loss": 2.9385,
+      "step": 8660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6526212849185767e-05,
+      "loss": 2.5199,
+      "step": 8670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.652196096772822e-05,
+      "loss": 2.1399,
+      "step": 8680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6517709086270676e-05,
+      "loss": 2.5872,
+      "step": 8690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6513457204813132e-05,
+      "loss": 2.5678,
+      "step": 8700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6509205323355585e-05,
+      "loss": 2.1971,
+      "step": 8710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650495344189804e-05,
+      "loss": 2.379,
+      "step": 8720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6500701560440494e-05,
+      "loss": 2.2224,
+      "step": 8730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.649644967898295e-05,
+      "loss": 2.7368,
+      "step": 8740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6492197797525406e-05,
+      "loss": 2.9536,
+      "step": 8750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6487945916067862e-05,
+      "loss": 2.1969,
+      "step": 8760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6483694034610315e-05,
+      "loss": 2.2591,
+      "step": 8770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647944215315277e-05,
+      "loss": 2.3555,
+      "step": 8780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6475190271695227e-05,
+      "loss": 2.1909,
+      "step": 8790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647093839023768e-05,
+      "loss": 2.6798,
+      "step": 8800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6466686508780136e-05,
+      "loss": 2.8922,
+      "step": 8810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6462434627322592e-05,
+      "loss": 2.4089,
+      "step": 8820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6458182745865045e-05,
+      "loss": 2.3367,
+      "step": 8830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64539308644075e-05,
+      "loss": 2.1407,
+      "step": 8840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6449678982949957e-05,
+      "loss": 2.7404,
+      "step": 8850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6445427101492413e-05,
+      "loss": 2.4875,
+      "step": 8860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6441175220034866e-05,
+      "loss": 3.2661,
+      "step": 8870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6436923338577322e-05,
+      "loss": 2.4685,
+      "step": 8880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643267145711978e-05,
+      "loss": 2.9441,
+      "step": 8890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642841957566223e-05,
+      "loss": 2.4203,
+      "step": 8900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6424167694204687e-05,
+      "loss": 2.5944,
+      "step": 8910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6419915812747143e-05,
+      "loss": 2.1158,
+      "step": 8920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6415663931289596e-05,
+      "loss": 2.8647,
+      "step": 8930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6411412049832052e-05,
+      "loss": 2.5355,
+      "step": 8940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.640716016837451e-05,
+      "loss": 2.6662,
+      "step": 8950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6402908286916965e-05,
+      "loss": 2.2165,
+      "step": 8960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6398656405459417e-05,
+      "loss": 2.1348,
+      "step": 8970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6394404524001874e-05,
+      "loss": 2.1628,
+      "step": 8980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6390152642544326e-05,
+      "loss": 2.3188,
+      "step": 8990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6385900761086783e-05,
+      "loss": 2.8703,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9859079122543335,
+      "eval_runtime": 337.3358,
+      "eval_samples_per_second": 7.42,
+      "eval_steps_per_second": 3.711,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.638164887962924e-05,
+      "loss": 2.5798,
+      "step": 9010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637739699817169e-05,
+      "loss": 2.5057,
+      "step": 9020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6373145116714148e-05,
+      "loss": 2.5603,
+      "step": 9030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6368893235256604e-05,
+      "loss": 2.5391,
+      "step": 9040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6364641353799057e-05,
+      "loss": 1.8705,
+      "step": 9050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6360389472341513e-05,
+      "loss": 2.1202,
+      "step": 9060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6356137590883965e-05,
+      "loss": 2.8342,
+      "step": 9070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.635188570942642e-05,
+      "loss": 2.7267,
+      "step": 9080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6347633827968878e-05,
+      "loss": 2.2638,
+      "step": 9090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.634338194651133e-05,
+      "loss": 2.9132,
+      "step": 9100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6339130065053787e-05,
+      "loss": 2.5094,
+      "step": 9110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6334878183596243e-05,
+      "loss": 2.6133,
+      "step": 9120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6330626302138696e-05,
+      "loss": 2.5961,
+      "step": 9130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6326374420681152e-05,
+      "loss": 2.2942,
+      "step": 9140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6322122539223608e-05,
+      "loss": 2.6119,
+      "step": 9150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.631787065776606e-05,
+      "loss": 2.2025,
+      "step": 9160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6313618776308517e-05,
+      "loss": 2.6507,
+      "step": 9170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6309366894850973e-05,
+      "loss": 2.395,
+      "step": 9180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.630511501339343e-05,
+      "loss": 2.738,
+      "step": 9190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6300863131935882e-05,
+      "loss": 2.2349,
+      "step": 9200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6296611250478338e-05,
+      "loss": 2.5223,
+      "step": 9210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6292359369020794e-05,
+      "loss": 2.4301,
+      "step": 9220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6288107487563247e-05,
+      "loss": 2.5581,
+      "step": 9230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6283855606105703e-05,
+      "loss": 2.7029,
+      "step": 9240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.627960372464816e-05,
+      "loss": 2.4213,
+      "step": 9250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6275351843190612e-05,
+      "loss": 2.2374,
+      "step": 9260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6271099961733068e-05,
+      "loss": 2.6529,
+      "step": 9270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6266848080275524e-05,
+      "loss": 2.0457,
+      "step": 9280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.626259619881798e-05,
+      "loss": 2.2516,
+      "step": 9290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6258344317360433e-05,
+      "loss": 2.4958,
+      "step": 9300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.625409243590289e-05,
+      "loss": 2.3028,
+      "step": 9310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6249840554445346e-05,
+      "loss": 2.623,
+      "step": 9320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.62455886729878e-05,
+      "loss": 2.1896,
+      "step": 9330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6241336791530254e-05,
+      "loss": 3.0736,
+      "step": 9340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.623708491007271e-05,
+      "loss": 2.4666,
+      "step": 9350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6232833028615163e-05,
+      "loss": 2.7424,
+      "step": 9360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.622858114715762e-05,
+      "loss": 1.9853,
+      "step": 9370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6224329265700072e-05,
+      "loss": 2.8972,
+      "step": 9380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.622007738424253e-05,
+      "loss": 2.6863,
+      "step": 9390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6215825502784985e-05,
+      "loss": 1.75,
+      "step": 9400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6211573621327437e-05,
+      "loss": 2.4478,
+      "step": 9410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6207321739869894e-05,
+      "loss": 3.1368,
+      "step": 9420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.620306985841235e-05,
+      "loss": 2.3129,
+      "step": 9430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6198817976954802e-05,
+      "loss": 2.7234,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.619456609549726e-05,
+      "loss": 2.6612,
+      "step": 9450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.619031421403971e-05,
+      "loss": 2.3061,
+      "step": 9460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6186062332582168e-05,
+      "loss": 2.1242,
+      "step": 9470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6181810451124624e-05,
+      "loss": 2.5048,
+      "step": 9480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.617755856966708e-05,
+      "loss": 2.4544,
+      "step": 9490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6173306688209533e-05,
+      "loss": 2.3058,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9763813018798828,
+      "eval_runtime": 157.87,
+      "eval_samples_per_second": 15.855,
+      "eval_steps_per_second": 7.931,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.616905480675199e-05,
+      "loss": 2.5569,
+      "step": 9510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6164802925294445e-05,
+      "loss": 1.7934,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6160551043836898e-05,
+      "loss": 2.4943,
+      "step": 9530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6156299162379354e-05,
+      "loss": 2.9005,
+      "step": 9540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.615204728092181e-05,
+      "loss": 2.1819,
+      "step": 9550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6147795399464263e-05,
+      "loss": 2.3916,
+      "step": 9560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.614354351800672e-05,
+      "loss": 2.6763,
+      "step": 9570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6139291636549175e-05,
+      "loss": 2.6101,
+      "step": 9580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.613503975509163e-05,
+      "loss": 2.5131,
+      "step": 9590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6130787873634084e-05,
+      "loss": 2.9305,
+      "step": 9600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.612653599217654e-05,
+      "loss": 2.1208,
+      "step": 9610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.6122284110718996e-05,
+      "loss": 2.4165,
+      "step": 9620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.611803222926145e-05,
+      "loss": 2.5368,
+      "step": 9630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6113780347803905e-05,
+      "loss": 2.8788,
+      "step": 9640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.610952846634636e-05,
+      "loss": 2.4815,
+      "step": 9650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6105276584888814e-05,
+      "loss": 2.8152,
+      "step": 9660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.610102470343127e-05,
+      "loss": 2.0329,
+      "step": 9670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6096772821973726e-05,
+      "loss": 2.9021,
+      "step": 9680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.609252094051618e-05,
+      "loss": 2.8002,
+      "step": 9690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6088269059058635e-05,
+      "loss": 2.4334,
+      "step": 9700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.608401717760109e-05,
+      "loss": 2.4941,
+      "step": 9710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6079765296143544e-05,
+      "loss": 2.2755,
+      "step": 9720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6075513414686e-05,
+      "loss": 2.052,
+      "step": 9730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6071261533228457e-05,
+      "loss": 2.4636,
+      "step": 9740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.606700965177091e-05,
+      "loss": 2.3624,
+      "step": 9750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6062757770313365e-05,
+      "loss": 2.3953,
+      "step": 9760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6058505888855818e-05,
+      "loss": 2.2087,
+      "step": 9770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6054254007398274e-05,
+      "loss": 1.9505,
+      "step": 9780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.605000212594073e-05,
+      "loss": 2.5386,
+      "step": 9790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6045750244483183e-05,
+      "loss": 2.4996,
+      "step": 9800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.604149836302564e-05,
+      "loss": 2.4536,
+      "step": 9810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6037246481568096e-05,
+      "loss": 1.847,
+      "step": 9820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.603299460011055e-05,
+      "loss": 2.4618,
+      "step": 9830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6028742718653004e-05,
+      "loss": 2.4738,
+      "step": 9840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.602449083719546e-05,
+      "loss": 2.1588,
+      "step": 9850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.6020238955737913e-05,
+      "loss": 2.3344,
+      "step": 9860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.601598707428037e-05,
+      "loss": 2.5213,
+      "step": 9870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6011735192822826e-05,
+      "loss": 2.374,
+      "step": 9880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.600748331136528e-05,
+      "loss": 2.6436,
+      "step": 9890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.6003231429907735e-05,
+      "loss": 2.1786,
+      "step": 9900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.599897954845019e-05,
+      "loss": 2.7025,
+      "step": 9910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5994727666992647e-05,
+      "loss": 2.4774,
+      "step": 9920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.59904757855351e-05,
+      "loss": 2.2819,
+      "step": 9930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5986223904077556e-05,
+      "loss": 2.1652,
+      "step": 9940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5981972022620012e-05,
+      "loss": 2.084,
+      "step": 9950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5977720141162465e-05,
+      "loss": 2.4612,
+      "step": 9960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.597346825970492e-05,
+      "loss": 2.0647,
+      "step": 9970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5969216378247377e-05,
+      "loss": 2.8856,
+      "step": 9980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.596496449678983e-05,
+      "loss": 2.4716,
+      "step": 9990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5960712615332286e-05,
+      "loss": 2.6674,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9638748168945312,
+      "eval_runtime": 157.7497,
+      "eval_samples_per_second": 15.867,
+      "eval_steps_per_second": 7.937,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5956460733874742e-05,
+      "loss": 2.8031,
+      "step": 10010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5952208852417198e-05,
+      "loss": 2.4077,
+      "step": 10020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.594795697095965e-05,
+      "loss": 2.1363,
+      "step": 10030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5943705089502107e-05,
+      "loss": 2.5921,
+      "step": 10040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5939453208044563e-05,
+      "loss": 2.0634,
+      "step": 10050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5935201326587016e-05,
+      "loss": 2.7191,
+      "step": 10060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5930949445129472e-05,
+      "loss": 2.2552,
+      "step": 10070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.592669756367193e-05,
+      "loss": 2.503,
+      "step": 10080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.592244568221438e-05,
+      "loss": 2.2036,
+      "step": 10090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5918193800756837e-05,
+      "loss": 2.637,
+      "step": 10100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.591394191929929e-05,
+      "loss": 2.2863,
+      "step": 10110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5909690037841746e-05,
+      "loss": 2.3317,
+      "step": 10120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5905438156384202e-05,
+      "loss": 2.2109,
+      "step": 10130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5901186274926655e-05,
+      "loss": 2.5821,
+      "step": 10140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.589693439346911e-05,
+      "loss": 2.5003,
+      "step": 10150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5892682512011564e-05,
+      "loss": 2.5828,
+      "step": 10160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.588843063055402e-05,
+      "loss": 2.2188,
+      "step": 10170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5884178749096476e-05,
+      "loss": 2.1588,
+      "step": 10180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.587992686763893e-05,
+      "loss": 2.7165,
+      "step": 10190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5875674986181385e-05,
+      "loss": 2.4099,
+      "step": 10200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.587142310472384e-05,
+      "loss": 2.1347,
+      "step": 10210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5867171223266294e-05,
+      "loss": 2.3912,
+      "step": 10220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.586291934180875e-05,
+      "loss": 2.3647,
+      "step": 10230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5858667460351207e-05,
+      "loss": 2.6857,
+      "step": 10240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5854415578893663e-05,
+      "loss": 2.5615,
+      "step": 10250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5850163697436115e-05,
+      "loss": 2.8766,
+      "step": 10260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.584591181597857e-05,
+      "loss": 2.1335,
+      "step": 10270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5841659934521028e-05,
+      "loss": 2.3573,
+      "step": 10280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.583740805306348e-05,
+      "loss": 2.3425,
+      "step": 10290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5833156171605937e-05,
+      "loss": 2.3004,
+      "step": 10300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5828904290148393e-05,
+      "loss": 2.429,
+      "step": 10310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5824652408690846e-05,
+      "loss": 2.3504,
+      "step": 10320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5820400527233302e-05,
+      "loss": 2.4716,
+      "step": 10330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5816148645775758e-05,
+      "loss": 2.3105,
+      "step": 10340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5811896764318214e-05,
+      "loss": 2.1203,
+      "step": 10350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5807644882860667e-05,
+      "loss": 2.7303,
+      "step": 10360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5803393001403123e-05,
+      "loss": 2.4012,
+      "step": 10370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.579914111994558e-05,
+      "loss": 2.5295,
+      "step": 10380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5794889238488032e-05,
+      "loss": 2.82,
+      "step": 10390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5790637357030488e-05,
+      "loss": 2.279,
+      "step": 10400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5786385475572944e-05,
+      "loss": 1.7126,
+      "step": 10410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5782133594115397e-05,
+      "loss": 2.3496,
+      "step": 10420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5777881712657853e-05,
+      "loss": 2.4562,
+      "step": 10430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.577362983120031e-05,
+      "loss": 2.485,
+      "step": 10440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5769377949742762e-05,
+      "loss": 2.0319,
+      "step": 10450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5765126068285218e-05,
+      "loss": 2.8354,
+      "step": 10460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5760874186827674e-05,
+      "loss": 1.943,
+      "step": 10470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5756622305370127e-05,
+      "loss": 2.2612,
+      "step": 10480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5752370423912583e-05,
+      "loss": 2.523,
+      "step": 10490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5748118542455036e-05,
+      "loss": 2.8775,
+      "step": 10500
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.934218406677246,
+      "eval_runtime": 157.7398,
+      "eval_samples_per_second": 15.868,
+      "eval_steps_per_second": 7.937,
+      "step": 10500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5743866660997492e-05,
+      "loss": 2.356,
+      "step": 10510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.573961477953995e-05,
+      "loss": 2.6021,
+      "step": 10520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.57353628980824e-05,
+      "loss": 2.2362,
+      "step": 10530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5731111016624857e-05,
+      "loss": 2.6383,
+      "step": 10540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5726859135167313e-05,
+      "loss": 2.407,
+      "step": 10550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5722607253709766e-05,
+      "loss": 2.8033,
+      "step": 10560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5718355372252222e-05,
+      "loss": 2.2644,
+      "step": 10570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.571410349079468e-05,
+      "loss": 3.0515,
+      "step": 10580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.570985160933713e-05,
+      "loss": 2.0456,
+      "step": 10590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5705599727879587e-05,
+      "loss": 2.1008,
+      "step": 10600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5701347846422044e-05,
+      "loss": 2.6307,
+      "step": 10610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5697095964964496e-05,
+      "loss": 2.4413,
+      "step": 10620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5692844083506952e-05,
+      "loss": 2.7403,
+      "step": 10630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.568859220204941e-05,
+      "loss": 2.5489,
+      "step": 10640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5684340320591865e-05,
+      "loss": 2.6111,
+      "step": 10650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5680088439134318e-05,
+      "loss": 2.4437,
+      "step": 10660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5675836557676774e-05,
+      "loss": 2.334,
+      "step": 10670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.567158467621923e-05,
+      "loss": 2.208,
+      "step": 10680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5667332794761683e-05,
+      "loss": 2.502,
+      "step": 10690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.566308091330414e-05,
+      "loss": 2.6235,
+      "step": 10700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5658829031846595e-05,
+      "loss": 2.4196,
+      "step": 10710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5654577150389048e-05,
+      "loss": 2.6142,
+      "step": 10720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5650325268931504e-05,
+      "loss": 2.6184,
+      "step": 10730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.564607338747396e-05,
+      "loss": 2.2517,
+      "step": 10740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5641821506016413e-05,
+      "loss": 2.8515,
+      "step": 10750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.563756962455887e-05,
+      "loss": 2.4117,
+      "step": 10760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5633317743101325e-05,
+      "loss": 2.2778,
+      "step": 10770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.562906586164378e-05,
+      "loss": 2.7605,
+      "step": 10780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5624813980186234e-05,
+      "loss": 2.4827,
+      "step": 10790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.562056209872869e-05,
+      "loss": 2.3697,
+      "step": 10800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.5616310217271143e-05,
+      "loss": 2.8352,
+      "step": 10810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.56120583358136e-05,
+      "loss": 2.1289,
+      "step": 10820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5607806454356055e-05,
+      "loss": 2.8038,
+      "step": 10830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5603554572898508e-05,
+      "loss": 2.6752,
+      "step": 10840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5599302691440964e-05,
+      "loss": 2.4184,
+      "step": 10850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.559505080998342e-05,
+      "loss": 2.7613,
+      "step": 10860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5590798928525873e-05,
+      "loss": 1.7401,
+      "step": 10870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.558654704706833e-05,
+      "loss": 2.1031,
+      "step": 10880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5582295165610782e-05,
+      "loss": 2.1724,
+      "step": 10890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5578043284153238e-05,
+      "loss": 2.3654,
+      "step": 10900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5573791402695694e-05,
+      "loss": 1.8411,
+      "step": 10910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5569539521238147e-05,
+      "loss": 2.3196,
+      "step": 10920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5565287639780603e-05,
+      "loss": 2.9633,
+      "step": 10930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.556103575832306e-05,
+      "loss": 2.9868,
+      "step": 10940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5556783876865512e-05,
+      "loss": 2.6125,
+      "step": 10950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5552531995407968e-05,
+      "loss": 2.3021,
+      "step": 10960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5548280113950424e-05,
+      "loss": 2.2658,
+      "step": 10970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.554402823249288e-05,
+      "loss": 2.5044,
+      "step": 10980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5539776351035333e-05,
+      "loss": 2.8643,
+      "step": 10990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.553552446957779e-05,
+      "loss": 2.5047,
+      "step": 11000
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.9329371452331543,
+      "eval_runtime": 157.6887,
+      "eval_samples_per_second": 15.873,
+      "eval_steps_per_second": 7.94,
+      "step": 11000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5531272588120246e-05,
+      "loss": 2.3317,
+      "step": 11010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.55270207066627e-05,
+      "loss": 2.6039,
+      "step": 11020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5522768825205155e-05,
+      "loss": 2.8393,
+      "step": 11030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.551851694374761e-05,
+      "loss": 2.6933,
+      "step": 11040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5514265062290063e-05,
+      "loss": 2.1474,
+      "step": 11050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.551001318083252e-05,
+      "loss": 2.6101,
+      "step": 11060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5505761299374976e-05,
+      "loss": 2.3679,
+      "step": 11070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5501509417917432e-05,
+      "loss": 1.9772,
+      "step": 11080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5497257536459885e-05,
+      "loss": 2.5859,
+      "step": 11090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.549300565500234e-05,
+      "loss": 1.9622,
+      "step": 11100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5488753773544797e-05,
+      "loss": 2.3536,
+      "step": 11110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.548450189208725e-05,
+      "loss": 2.4914,
+      "step": 11120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5480250010629706e-05,
+      "loss": 2.5306,
+      "step": 11130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5475998129172162e-05,
+      "loss": 2.2826,
+      "step": 11140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5471746247714615e-05,
+      "loss": 2.4863,
+      "step": 11150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.546749436625707e-05,
+      "loss": 2.7049,
+      "step": 11160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5463242484799527e-05,
+      "loss": 1.9233,
+      "step": 11170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.545899060334198e-05,
+      "loss": 1.9265,
+      "step": 11180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5454738721884436e-05,
+      "loss": 2.3964,
+      "step": 11190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.545048684042689e-05,
+      "loss": 2.4938,
+      "step": 11200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5446234958969345e-05,
+      "loss": 2.3325,
+      "step": 11210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.54419830775118e-05,
+      "loss": 2.0072,
+      "step": 11220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5437731196054254e-05,
+      "loss": 2.2178,
+      "step": 11230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.543347931459671e-05,
+      "loss": 2.2869,
+      "step": 11240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5429227433139166e-05,
+      "loss": 2.2809,
+      "step": 11250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.542497555168162e-05,
+      "loss": 2.0549,
+      "step": 11260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5420723670224075e-05,
+      "loss": 2.5829,
+      "step": 11270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5416471788766528e-05,
+      "loss": 2.6291,
+      "step": 11280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5412219907308984e-05,
+      "loss": 2.5527,
+      "step": 11290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.540796802585144e-05,
+      "loss": 2.4998,
+      "step": 11300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5403716144393896e-05,
+      "loss": 2.2437,
+      "step": 11310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.539946426293635e-05,
+      "loss": 2.8028,
+      "step": 11320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5395212381478805e-05,
+      "loss": 2.146,
+      "step": 11330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.539096050002126e-05,
+      "loss": 2.3451,
+      "step": 11340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5386708618563714e-05,
+      "loss": 2.5306,
+      "step": 11350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.538245673710617e-05,
+      "loss": 2.472,
+      "step": 11360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5378204855648626e-05,
+      "loss": 2.3911,
+      "step": 11370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.537395297419108e-05,
+      "loss": 2.4329,
+      "step": 11380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5369701092733535e-05,
+      "loss": 2.5001,
+      "step": 11390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.536544921127599e-05,
+      "loss": 2.8697,
+      "step": 11400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5361197329818448e-05,
+      "loss": 2.2511,
+      "step": 11410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.53569454483609e-05,
+      "loss": 2.6034,
+      "step": 11420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5352693566903357e-05,
+      "loss": 2.6464,
+      "step": 11430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5348441685445813e-05,
+      "loss": 2.543,
+      "step": 11440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5344189803988265e-05,
+      "loss": 2.3194,
+      "step": 11450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.533993792253072e-05,
+      "loss": 2.2008,
+      "step": 11460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5335686041073178e-05,
+      "loss": 2.3158,
+      "step": 11470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.533143415961563e-05,
+      "loss": 2.4198,
+      "step": 11480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5327182278158087e-05,
+      "loss": 2.2958,
+      "step": 11490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5322930396700543e-05,
+      "loss": 2.5799,
+      "step": 11500
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 1.9254064559936523,
+      "eval_runtime": 162.9964,
+      "eval_samples_per_second": 15.356,
+      "eval_steps_per_second": 7.681,
+      "step": 11500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5318678515243e-05,
+      "loss": 2.7565,
+      "step": 11510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5314426633785452e-05,
+      "loss": 2.1682,
+      "step": 11520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5310174752327908e-05,
+      "loss": 2.5621,
+      "step": 11530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.530592287087036e-05,
+      "loss": 2.3396,
+      "step": 11540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5301670989412817e-05,
+      "loss": 2.2505,
+      "step": 11550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5297419107955273e-05,
+      "loss": 2.0222,
+      "step": 11560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5293167226497726e-05,
+      "loss": 2.9315,
+      "step": 11570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5288915345040182e-05,
+      "loss": 2.4993,
+      "step": 11580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5284663463582635e-05,
+      "loss": 3.0336,
+      "step": 11590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.528041158212509e-05,
+      "loss": 2.3143,
+      "step": 11600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5276159700667547e-05,
+      "loss": 2.3467,
+      "step": 11610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.527190781921e-05,
+      "loss": 2.0148,
+      "step": 11620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5267655937752456e-05,
+      "loss": 2.5407,
+      "step": 11630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5263404056294912e-05,
+      "loss": 2.3283,
+      "step": 11640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5259152174837365e-05,
+      "loss": 2.1069,
+      "step": 11650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5254900293379823e-05,
+      "loss": 2.3977,
+      "step": 11660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5250648411922277e-05,
+      "loss": 2.3978,
+      "step": 11670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5246396530464732e-05,
+      "loss": 2.2315,
+      "step": 11680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5242144649007186e-05,
+      "loss": 2.5454,
+      "step": 11690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5237892767549642e-05,
+      "loss": 2.266,
+      "step": 11700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5233640886092097e-05,
+      "loss": 2.2491,
+      "step": 11710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5229389004634551e-05,
+      "loss": 2.0303,
+      "step": 11720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5225137123177007e-05,
+      "loss": 2.3941,
+      "step": 11730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5220885241719463e-05,
+      "loss": 2.4527,
+      "step": 11740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5216633360261916e-05,
+      "loss": 2.1319,
+      "step": 11750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5212381478804372e-05,
+      "loss": 1.9728,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5208129597346828e-05,
+      "loss": 2.3851,
+      "step": 11770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5203877715889281e-05,
+      "loss": 2.5249,
+      "step": 11780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5199625834431737e-05,
+      "loss": 2.318,
+      "step": 11790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5195373952974194e-05,
+      "loss": 2.3987,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5191122071516646e-05,
+      "loss": 2.2828,
+      "step": 11810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5186870190059102e-05,
+      "loss": 2.2173,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5182618308601557e-05,
+      "loss": 2.3421,
+      "step": 11830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5178366427144013e-05,
+      "loss": 2.3685,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5174114545686468e-05,
+      "loss": 2.3534,
+      "step": 11850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5169862664228922e-05,
+      "loss": 2.2374,
+      "step": 11860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5165610782771378e-05,
+      "loss": 2.2038,
+      "step": 11870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5161358901313833e-05,
+      "loss": 2.2161,
+      "step": 11880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5157107019856287e-05,
+      "loss": 2.6806,
+      "step": 11890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5152855138398743e-05,
+      "loss": 2.3153,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5148603256941196e-05,
+      "loss": 2.0646,
+      "step": 11910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5144351375483652e-05,
+      "loss": 2.3844,
+      "step": 11920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5140099494026108e-05,
+      "loss": 2.5267,
+      "step": 11930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5135847612568564e-05,
+      "loss": 2.3379,
+      "step": 11940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5131595731111017e-05,
+      "loss": 2.2537,
+      "step": 11950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5127343849653473e-05,
+      "loss": 2.0048,
+      "step": 11960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.512309196819593e-05,
+      "loss": 2.5609,
+      "step": 11970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5118840086738382e-05,
+      "loss": 2.4389,
+      "step": 11980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5114588205280838e-05,
+      "loss": 2.3636,
+      "step": 11990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5110336323823293e-05,
+      "loss": 2.4979,
+      "step": 12000
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.908052921295166,
+      "eval_runtime": 175.5855,
+      "eval_samples_per_second": 14.255,
+      "eval_steps_per_second": 7.13,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5106084442365747e-05,
+      "loss": 2.5816,
+      "step": 12010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5101832560908203e-05,
+      "loss": 2.7254,
+      "step": 12020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5097580679450658e-05,
+      "loss": 2.1259,
+      "step": 12030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5093328797993114e-05,
+      "loss": 2.6026,
+      "step": 12040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5089076916535569e-05,
+      "loss": 2.2787,
+      "step": 12050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5084825035078023e-05,
+      "loss": 2.3396,
+      "step": 12060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.508057315362048e-05,
+      "loss": 2.4013,
+      "step": 12070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5076321272162932e-05,
+      "loss": 2.5999,
+      "step": 12080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5072069390705388e-05,
+      "loss": 1.7368,
+      "step": 12090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5067817509247844e-05,
+      "loss": 2.3508,
+      "step": 12100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5063565627790297e-05,
+      "loss": 2.0358,
+      "step": 12110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5059313746332753e-05,
+      "loss": 2.4323,
+      "step": 12120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.505506186487521e-05,
+      "loss": 2.298,
+      "step": 12130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5050809983417665e-05,
+      "loss": 2.1689,
+      "step": 12140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5046558101960118e-05,
+      "loss": 2.3162,
+      "step": 12150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5042306220502574e-05,
+      "loss": 2.467,
+      "step": 12160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5038054339045029e-05,
+      "loss": 2.0638,
+      "step": 12170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5033802457587483e-05,
+      "loss": 2.399,
+      "step": 12180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.502955057612994e-05,
+      "loss": 2.4608,
+      "step": 12190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5025298694672394e-05,
+      "loss": 2.2544,
+      "step": 12200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5021046813214848e-05,
+      "loss": 2.4616,
+      "step": 12210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5016794931757303e-05,
+      "loss": 2.8345,
+      "step": 12220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5012543050299759e-05,
+      "loss": 1.9567,
+      "step": 12230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5008291168842213e-05,
+      "loss": 2.3264,
+      "step": 12240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.5004039287384668e-05,
+      "loss": 2.3828,
+      "step": 12250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4999787405927124e-05,
+      "loss": 2.2728,
+      "step": 12260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.499553552446958e-05,
+      "loss": 2.6114,
+      "step": 12270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4991283643012033e-05,
+      "loss": 2.6562,
+      "step": 12280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4987031761554489e-05,
+      "loss": 2.5292,
+      "step": 12290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4982779880096945e-05,
+      "loss": 1.8669,
+      "step": 12300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4978527998639398e-05,
+      "loss": 2.4771,
+      "step": 12310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4974276117181854e-05,
+      "loss": 1.9845,
+      "step": 12320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.497002423572431e-05,
+      "loss": 2.1029,
+      "step": 12330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4965772354266763e-05,
+      "loss": 2.3275,
+      "step": 12340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.496152047280922e-05,
+      "loss": 2.2364,
+      "step": 12350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4957268591351675e-05,
+      "loss": 2.2305,
+      "step": 12360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.495301670989413e-05,
+      "loss": 1.9404,
+      "step": 12370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4948764828436584e-05,
+      "loss": 2.4992,
+      "step": 12380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4944512946979039e-05,
+      "loss": 2.3739,
+      "step": 12390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4940261065521495e-05,
+      "loss": 2.1145,
+      "step": 12400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.493600918406395e-05,
+      "loss": 2.4534,
+      "step": 12410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4931757302606404e-05,
+      "loss": 2.2821,
+      "step": 12420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.492750542114886e-05,
+      "loss": 2.3779,
+      "step": 12430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4923253539691314e-05,
+      "loss": 2.7704,
+      "step": 12440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4919001658233769e-05,
+      "loss": 2.2016,
+      "step": 12450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4914749776776225e-05,
+      "loss": 2.4465,
+      "step": 12460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4910497895318681e-05,
+      "loss": 2.2369,
+      "step": 12470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4906246013861134e-05,
+      "loss": 2.1973,
+      "step": 12480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.490199413240359e-05,
+      "loss": 2.2326,
+      "step": 12490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4897742250946046e-05,
+      "loss": 2.8084,
+      "step": 12500
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 1.8825452327728271,
+      "eval_runtime": 175.6636,
+      "eval_samples_per_second": 14.249,
+      "eval_steps_per_second": 7.127,
+      "step": 12500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4893490369488499e-05,
+      "loss": 2.4276,
+      "step": 12510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4889238488030955e-05,
+      "loss": 2.6502,
+      "step": 12520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4884986606573411e-05,
+      "loss": 2.1759,
+      "step": 12530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4880734725115864e-05,
+      "loss": 2.8552,
+      "step": 12540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.487648284365832e-05,
+      "loss": 1.9784,
+      "step": 12550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4872230962200775e-05,
+      "loss": 2.3719,
+      "step": 12560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4867979080743231e-05,
+      "loss": 2.5856,
+      "step": 12570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4863727199285685e-05,
+      "loss": 1.9766,
+      "step": 12580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.485947531782814e-05,
+      "loss": 2.7214,
+      "step": 12590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4855223436370596e-05,
+      "loss": 2.6672,
+      "step": 12600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4850971554913049e-05,
+      "loss": 2.4406,
+      "step": 12610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4846719673455505e-05,
+      "loss": 2.1624,
+      "step": 12620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4842467791997961e-05,
+      "loss": 2.4437,
+      "step": 12630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4838215910540414e-05,
+      "loss": 2.6915,
+      "step": 12640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.483396402908287e-05,
+      "loss": 2.0104,
+      "step": 12650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4829712147625326e-05,
+      "loss": 2.6162,
+      "step": 12660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4825460266167782e-05,
+      "loss": 2.4617,
+      "step": 12670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4821208384710235e-05,
+      "loss": 2.0574,
+      "step": 12680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4816956503252691e-05,
+      "loss": 2.5159,
+      "step": 12690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4812704621795146e-05,
+      "loss": 2.6508,
+      "step": 12700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.48084527403376e-05,
+      "loss": 2.7016,
+      "step": 12710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4804200858880056e-05,
+      "loss": 2.598,
+      "step": 12720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.479994897742251e-05,
+      "loss": 2.2087,
+      "step": 12730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4795697095964965e-05,
+      "loss": 1.8749,
+      "step": 12740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4791445214507421e-05,
+      "loss": 2.2771,
+      "step": 12750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4787193333049876e-05,
+      "loss": 2.6843,
+      "step": 12760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.478294145159233e-05,
+      "loss": 2.4945,
+      "step": 12770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4778689570134785e-05,
+      "loss": 2.8105,
+      "step": 12780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.477443768867724e-05,
+      "loss": 2.5913,
+      "step": 12790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4770185807219697e-05,
+      "loss": 2.0956,
+      "step": 12800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.476593392576215e-05,
+      "loss": 2.7534,
+      "step": 12810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4761682044304606e-05,
+      "loss": 2.4345,
+      "step": 12820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4757430162847062e-05,
+      "loss": 2.3806,
+      "step": 12830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4753178281389515e-05,
+      "loss": 1.8217,
+      "step": 12840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4748926399931971e-05,
+      "loss": 2.4636,
+      "step": 12850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4744674518474427e-05,
+      "loss": 2.8308,
+      "step": 12860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.474042263701688e-05,
+      "loss": 2.1157,
+      "step": 12870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4736170755559336e-05,
+      "loss": 2.5942,
+      "step": 12880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4731918874101792e-05,
+      "loss": 2.4219,
+      "step": 12890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4727666992644247e-05,
+      "loss": 2.3702,
+      "step": 12900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4723415111186701e-05,
+      "loss": 2.4293,
+      "step": 12910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4719163229729157e-05,
+      "loss": 2.1792,
+      "step": 12920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4714911348271612e-05,
+      "loss": 2.4994,
+      "step": 12930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4710659466814066e-05,
+      "loss": 2.3339,
+      "step": 12940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.470640758535652e-05,
+      "loss": 2.5192,
+      "step": 12950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4702155703898977e-05,
+      "loss": 2.268,
+      "step": 12960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4697903822441431e-05,
+      "loss": 2.4811,
+      "step": 12970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4693651940983886e-05,
+      "loss": 2.2386,
+      "step": 12980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4689400059526342e-05,
+      "loss": 2.6465,
+      "step": 12990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4685148178068798e-05,
+      "loss": 2.6341,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 1.882176160812378,
+      "eval_runtime": 175.5149,
+      "eval_samples_per_second": 14.261,
+      "eval_steps_per_second": 7.133,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.468089629661125e-05,
+      "loss": 2.2379,
+      "step": 13010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4676644415153707e-05,
+      "loss": 2.446,
+      "step": 13020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4672392533696163e-05,
+      "loss": 2.3322,
+      "step": 13030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4668140652238616e-05,
+      "loss": 2.205,
+      "step": 13040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4663888770781072e-05,
+      "loss": 2.3751,
+      "step": 13050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4659636889323528e-05,
+      "loss": 2.529,
+      "step": 13060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4655385007865981e-05,
+      "loss": 2.4666,
+      "step": 13070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4651133126408437e-05,
+      "loss": 2.6353,
+      "step": 13080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4646881244950893e-05,
+      "loss": 2.2018,
+      "step": 13090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4642629363493348e-05,
+      "loss": 2.2065,
+      "step": 13100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4638377482035802e-05,
+      "loss": 2.1817,
+      "step": 13110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4634125600578257e-05,
+      "loss": 2.5871,
+      "step": 13120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4629873719120713e-05,
+      "loss": 2.6006,
+      "step": 13130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4625621837663167e-05,
+      "loss": 2.4155,
+      "step": 13140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4621369956205622e-05,
+      "loss": 2.7159,
+      "step": 13150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4617118074748078e-05,
+      "loss": 1.9283,
+      "step": 13160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.461286619329053e-05,
+      "loss": 2.2639,
+      "step": 13170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4608614311832987e-05,
+      "loss": 2.7927,
+      "step": 13180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4604362430375443e-05,
+      "loss": 2.4052,
+      "step": 13190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4600110548917896e-05,
+      "loss": 2.4822,
+      "step": 13200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4595858667460352e-05,
+      "loss": 2.7162,
+      "step": 13210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4591606786002808e-05,
+      "loss": 2.9996,
+      "step": 13220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4587354904545264e-05,
+      "loss": 1.9623,
+      "step": 13230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4583103023087717e-05,
+      "loss": 2.4969,
+      "step": 13240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4578851141630173e-05,
+      "loss": 2.3253,
+      "step": 13250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4574599260172627e-05,
+      "loss": 2.2383,
+      "step": 13260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4570347378715082e-05,
+      "loss": 2.1804,
+      "step": 13270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4566095497257538e-05,
+      "loss": 2.4405,
+      "step": 13280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4561843615799993e-05,
+      "loss": 2.0605,
+      "step": 13290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4557591734342447e-05,
+      "loss": 2.4252,
+      "step": 13300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4553339852884903e-05,
+      "loss": 2.3363,
+      "step": 13310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4549087971427358e-05,
+      "loss": 2.5494,
+      "step": 13320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4544836089969814e-05,
+      "loss": 2.2146,
+      "step": 13330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4540584208512267e-05,
+      "loss": 2.1387,
+      "step": 13340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4536332327054723e-05,
+      "loss": 2.5365,
+      "step": 13350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4532080445597179e-05,
+      "loss": 2.2978,
+      "step": 13360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4527828564139632e-05,
+      "loss": 2.3413,
+      "step": 13370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4523576682682088e-05,
+      "loss": 1.9884,
+      "step": 13380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4519324801224544e-05,
+      "loss": 2.0978,
+      "step": 13390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4515072919766997e-05,
+      "loss": 1.9556,
+      "step": 13400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4510821038309453e-05,
+      "loss": 2.4953,
+      "step": 13410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4506569156851909e-05,
+      "loss": 2.1745,
+      "step": 13420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4502317275394363e-05,
+      "loss": 2.0909,
+      "step": 13430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4498065393936818e-05,
+      "loss": 2.2396,
+      "step": 13440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4493813512479274e-05,
+      "loss": 2.2915,
+      "step": 13450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4489561631021728e-05,
+      "loss": 2.2874,
+      "step": 13460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4485309749564183e-05,
+      "loss": 2.7083,
+      "step": 13470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4481057868106639e-05,
+      "loss": 2.1006,
+      "step": 13480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4476805986649094e-05,
+      "loss": 1.7699,
+      "step": 13490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4472554105191548e-05,
+      "loss": 2.0629,
+      "step": 13500
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 1.8929394483566284,
+      "eval_runtime": 175.8173,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 7.121,
+      "step": 13500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4468302223734002e-05,
+      "loss": 2.5065,
+      "step": 13510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4464050342276459e-05,
+      "loss": 2.4348,
+      "step": 13520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4459798460818915e-05,
+      "loss": 2.2692,
+      "step": 13530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4455546579361368e-05,
+      "loss": 2.1182,
+      "step": 13540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4451294697903824e-05,
+      "loss": 2.4457,
+      "step": 13550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.444704281644628e-05,
+      "loss": 2.9266,
+      "step": 13560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4442790934988733e-05,
+      "loss": 2.1608,
+      "step": 13570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4438539053531189e-05,
+      "loss": 2.1326,
+      "step": 13580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4434287172073645e-05,
+      "loss": 1.7725,
+      "step": 13590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4430035290616098e-05,
+      "loss": 2.1611,
+      "step": 13600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4425783409158554e-05,
+      "loss": 2.2394,
+      "step": 13610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.442153152770101e-05,
+      "loss": 2.3544,
+      "step": 13620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4417279646243464e-05,
+      "loss": 2.0564,
+      "step": 13630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4413027764785919e-05,
+      "loss": 2.8817,
+      "step": 13640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.4408775883328373e-05,
+      "loss": 2.2314,
+      "step": 13650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.440452400187083e-05,
+      "loss": 2.2732,
+      "step": 13660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4400272120413284e-05,
+      "loss": 2.1687,
+      "step": 13670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4396020238955738e-05,
+      "loss": 2.3604,
+      "step": 13680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4391768357498195e-05,
+      "loss": 2.3441,
+      "step": 13690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4387516476040649e-05,
+      "loss": 2.2658,
+      "step": 13700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4383264594583104e-05,
+      "loss": 2.542,
+      "step": 13710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.437901271312556e-05,
+      "loss": 2.4687,
+      "step": 13720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4374760831668012e-05,
+      "loss": 2.4955,
+      "step": 13730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4370508950210469e-05,
+      "loss": 2.3333,
+      "step": 13740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4366257068752925e-05,
+      "loss": 2.249,
+      "step": 13750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4362005187295381e-05,
+      "loss": 2.4311,
+      "step": 13760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4357753305837834e-05,
+      "loss": 2.3719,
+      "step": 13770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.435350142438029e-05,
+      "loss": 2.1542,
+      "step": 13780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4349249542922746e-05,
+      "loss": 2.3959,
+      "step": 13790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4344997661465199e-05,
+      "loss": 2.0495,
+      "step": 13800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4340745780007655e-05,
+      "loss": 2.0001,
+      "step": 13810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.433649389855011e-05,
+      "loss": 2.6268,
+      "step": 13820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4332242017092564e-05,
+      "loss": 2.2777,
+      "step": 13830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.432799013563502e-05,
+      "loss": 2.0227,
+      "step": 13840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4323738254177474e-05,
+      "loss": 1.8101,
+      "step": 13850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.431948637271993e-05,
+      "loss": 1.9943,
+      "step": 13860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4315234491262385e-05,
+      "loss": 2.2595,
+      "step": 13870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.431098260980484e-05,
+      "loss": 2.376,
+      "step": 13880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4306730728347296e-05,
+      "loss": 2.3874,
+      "step": 13890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4302478846889748e-05,
+      "loss": 2.1329,
+      "step": 13900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4298226965432205e-05,
+      "loss": 2.3208,
+      "step": 13910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.429397508397466e-05,
+      "loss": 2.4588,
+      "step": 13920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4289723202517113e-05,
+      "loss": 2.8068,
+      "step": 13930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.428547132105957e-05,
+      "loss": 2.4027,
+      "step": 13940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4281219439602026e-05,
+      "loss": 2.4779,
+      "step": 13950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4276967558144482e-05,
+      "loss": 2.403,
+      "step": 13960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4272715676686935e-05,
+      "loss": 2.1801,
+      "step": 13970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.426846379522939e-05,
+      "loss": 1.9641,
+      "step": 13980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4264211913771845e-05,
+      "loss": 2.4286,
+      "step": 13990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.42599600323143e-05,
+      "loss": 2.701,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 1.856201410293579,
+      "eval_runtime": 175.7531,
+      "eval_samples_per_second": 14.242,
+      "eval_steps_per_second": 7.124,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4255708150856756e-05,
+      "loss": 2.5715,
+      "step": 14010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.425145626939921e-05,
+      "loss": 2.6098,
+      "step": 14020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4247204387941665e-05,
+      "loss": 2.6931,
+      "step": 14030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.424295250648412e-05,
+      "loss": 2.5217,
+      "step": 14040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4238700625026575e-05,
+      "loss": 2.2983,
+      "step": 14050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4234448743569032e-05,
+      "loss": 1.8287,
+      "step": 14060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4230196862111484e-05,
+      "loss": 2.5353,
+      "step": 14070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.422594498065394e-05,
+      "loss": 1.932,
+      "step": 14080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4221693099196397e-05,
+      "loss": 2.2853,
+      "step": 14090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.421744121773885e-05,
+      "loss": 1.8722,
+      "step": 14100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4213189336281306e-05,
+      "loss": 2.2378,
+      "step": 14110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4208937454823762e-05,
+      "loss": 2.3306,
+      "step": 14120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.4204685573366214e-05,
+      "loss": 2.4661,
+      "step": 14130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.420043369190867e-05,
+      "loss": 2.5235,
+      "step": 14140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4196181810451127e-05,
+      "loss": 2.7158,
+      "step": 14150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4191929928993581e-05,
+      "loss": 2.104,
+      "step": 14160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4187678047536036e-05,
+      "loss": 2.3975,
+      "step": 14170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4183426166078492e-05,
+      "loss": 1.9535,
+      "step": 14180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4179174284620946e-05,
+      "loss": 2.3345,
+      "step": 14190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.41749224031634e-05,
+      "loss": 2.3982,
+      "step": 14200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4170670521705855e-05,
+      "loss": 2.6212,
+      "step": 14210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4166418640248311e-05,
+      "loss": 2.0249,
+      "step": 14220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4162166758790766e-05,
+      "loss": 2.366,
+      "step": 14230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.415791487733322e-05,
+      "loss": 2.2266,
+      "step": 14240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4153662995875676e-05,
+      "loss": 2.6209,
+      "step": 14250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4149411114418131e-05,
+      "loss": 2.3738,
+      "step": 14260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4145159232960585e-05,
+      "loss": 2.2413,
+      "step": 14270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4140907351503042e-05,
+      "loss": 2.3953,
+      "step": 14280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4136655470045498e-05,
+      "loss": 1.8551,
+      "step": 14290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.413240358858795e-05,
+      "loss": 2.2475,
+      "step": 14300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4128151707130407e-05,
+      "loss": 2.5065,
+      "step": 14310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4123899825672863e-05,
+      "loss": 2.3926,
+      "step": 14320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4119647944215316e-05,
+      "loss": 2.8505,
+      "step": 14330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4115396062757772e-05,
+      "loss": 2.351,
+      "step": 14340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4111144181300228e-05,
+      "loss": 2.5845,
+      "step": 14350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.410689229984268e-05,
+      "loss": 2.4993,
+      "step": 14360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4102640418385137e-05,
+      "loss": 2.5347,
+      "step": 14370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4098388536927591e-05,
+      "loss": 2.4627,
+      "step": 14380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4094136655470047e-05,
+      "loss": 2.4872,
+      "step": 14390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4089884774012502e-05,
+      "loss": 2.0532,
+      "step": 14400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4085632892554956e-05,
+      "loss": 2.5615,
+      "step": 14410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4081381011097412e-05,
+      "loss": 2.0314,
+      "step": 14420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4077129129639865e-05,
+      "loss": 2.3463,
+      "step": 14430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4072877248182321e-05,
+      "loss": 2.3824,
+      "step": 14440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4068625366724777e-05,
+      "loss": 2.4601,
+      "step": 14450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.406437348526723e-05,
+      "loss": 2.4652,
+      "step": 14460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4060121603809686e-05,
+      "loss": 2.0304,
+      "step": 14470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4055869722352143e-05,
+      "loss": 2.5522,
+      "step": 14480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4051617840894599e-05,
+      "loss": 2.0558,
+      "step": 14490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4047365959437051e-05,
+      "loss": 2.4705,
+      "step": 14500
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 1.872928261756897,
+      "eval_runtime": 175.8168,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 7.121,
+      "step": 14500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4043114077979508e-05,
+      "loss": 2.9471,
+      "step": 14510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4038862196521962e-05,
+      "loss": 2.2038,
+      "step": 14520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4034610315064417e-05,
+      "loss": 2.2991,
+      "step": 14530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4030358433606873e-05,
+      "loss": 2.4386,
+      "step": 14540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4026106552149327e-05,
+      "loss": 2.4721,
+      "step": 14550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4021854670691782e-05,
+      "loss": 2.31,
+      "step": 14560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4017602789234238e-05,
+      "loss": 2.3514,
+      "step": 14570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4013350907776692e-05,
+      "loss": 2.4382,
+      "step": 14580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4009099026319148e-05,
+      "loss": 1.9629,
+      "step": 14590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4004847144861601e-05,
+      "loss": 2.5185,
+      "step": 14600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.4000595263404057e-05,
+      "loss": 2.4013,
+      "step": 14610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3996343381946513e-05,
+      "loss": 2.2089,
+      "step": 14620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3992091500488966e-05,
+      "loss": 2.7305,
+      "step": 14630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3987839619031422e-05,
+      "loss": 2.0156,
+      "step": 14640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3983587737573879e-05,
+      "loss": 2.512,
+      "step": 14650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3979335856116331e-05,
+      "loss": 2.8554,
+      "step": 14660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3975083974658787e-05,
+      "loss": 2.9559,
+      "step": 14670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3970832093201244e-05,
+      "loss": 1.9561,
+      "step": 14680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3966580211743698e-05,
+      "loss": 2.2716,
+      "step": 14690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3962328330286152e-05,
+      "loss": 1.9613,
+      "step": 14700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3958076448828609e-05,
+      "loss": 2.7032,
+      "step": 14710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3953824567371063e-05,
+      "loss": 2.0244,
+      "step": 14720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3949572685913518e-05,
+      "loss": 2.2966,
+      "step": 14730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3945320804455974e-05,
+      "loss": 2.0398,
+      "step": 14740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3941068922998428e-05,
+      "loss": 2.3359,
+      "step": 14750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3936817041540883e-05,
+      "loss": 2.3542,
+      "step": 14760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3932565160083337e-05,
+      "loss": 2.6888,
+      "step": 14770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3928313278625793e-05,
+      "loss": 2.1765,
+      "step": 14780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3924061397168248e-05,
+      "loss": 2.5522,
+      "step": 14790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3919809515710702e-05,
+      "loss": 1.9513,
+      "step": 14800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3915557634253158e-05,
+      "loss": 1.9658,
+      "step": 14810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3911305752795614e-05,
+      "loss": 1.8836,
+      "step": 14820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3907053871338067e-05,
+      "loss": 2.4389,
+      "step": 14830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.3902801989880523e-05,
+      "loss": 2.0884,
+      "step": 14840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.389855010842298e-05,
+      "loss": 2.33,
+      "step": 14850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3894298226965432e-05,
+      "loss": 2.5639,
+      "step": 14860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3890046345507888e-05,
+      "loss": 2.5959,
+      "step": 14870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3885794464050345e-05,
+      "loss": 2.5403,
+      "step": 14880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3881542582592797e-05,
+      "loss": 2.5303,
+      "step": 14890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3877290701135254e-05,
+      "loss": 1.7927,
+      "step": 14900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.387303881967771e-05,
+      "loss": 1.9845,
+      "step": 14910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3868786938220164e-05,
+      "loss": 2.5558,
+      "step": 14920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3864535056762619e-05,
+      "loss": 2.8331,
+      "step": 14930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3860283175305073e-05,
+      "loss": 2.6268,
+      "step": 14940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.385603129384753e-05,
+      "loss": 2.0546,
+      "step": 14950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3851779412389984e-05,
+      "loss": 2.0332,
+      "step": 14960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3847527530932438e-05,
+      "loss": 1.941,
+      "step": 14970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3843275649474894e-05,
+      "loss": 2.4304,
+      "step": 14980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3839023768017347e-05,
+      "loss": 2.3973,
+      "step": 14990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3834771886559803e-05,
+      "loss": 1.8707,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.883968710899353,
+      "eval_runtime": 174.809,
+      "eval_samples_per_second": 14.318,
+      "eval_steps_per_second": 7.162,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.383052000510226e-05,
+      "loss": 2.941,
+      "step": 15010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3826268123644715e-05,
+      "loss": 2.4149,
+      "step": 15020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3822016242187168e-05,
+      "loss": 2.3082,
+      "step": 15030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3817764360729624e-05,
+      "loss": 2.2591,
+      "step": 15040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.381351247927208e-05,
+      "loss": 2.2677,
+      "step": 15050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3809260597814533e-05,
+      "loss": 2.216,
+      "step": 15060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.380500871635699e-05,
+      "loss": 2.0113,
+      "step": 15070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3800756834899444e-05,
+      "loss": 2.2585,
+      "step": 15080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.3796504953441898e-05,
+      "loss": 2.2651,
+      "step": 15090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3792253071984355e-05,
+      "loss": 2.4062,
+      "step": 15100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3788001190526809e-05,
+      "loss": 2.7675,
+      "step": 15110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3783749309069265e-05,
+      "loss": 2.1533,
+      "step": 15120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.377949742761172e-05,
+      "loss": 2.1812,
+      "step": 15130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3775245546154174e-05,
+      "loss": 2.3384,
+      "step": 15140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.377099366469663e-05,
+      "loss": 2.4825,
+      "step": 15150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3766741783239083e-05,
+      "loss": 2.3008,
+      "step": 15160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3762489901781539e-05,
+      "loss": 2.5847,
+      "step": 15170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3758238020323995e-05,
+      "loss": 2.0014,
+      "step": 15180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3753986138866448e-05,
+      "loss": 2.3731,
+      "step": 15190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3749734257408904e-05,
+      "loss": 2.5206,
+      "step": 15200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.374548237595136e-05,
+      "loss": 2.1148,
+      "step": 15210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3741230494493817e-05,
+      "loss": 2.2672,
+      "step": 15220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.373697861303627e-05,
+      "loss": 2.6911,
+      "step": 15230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3732726731578725e-05,
+      "loss": 2.2124,
+      "step": 15240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.372847485012118e-05,
+      "loss": 2.2615,
+      "step": 15250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3724222968663634e-05,
+      "loss": 1.7735,
+      "step": 15260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.371997108720609e-05,
+      "loss": 2.3346,
+      "step": 15270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3715719205748545e-05,
+      "loss": 2.4171,
+      "step": 15280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3711467324291e-05,
+      "loss": 2.5776,
+      "step": 15290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3707215442833456e-05,
+      "loss": 2.3896,
+      "step": 15300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.370296356137591e-05,
+      "loss": 2.488,
+      "step": 15310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3698711679918364e-05,
+      "loss": 2.0988,
+      "step": 15320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.3694459798460819e-05,
+      "loss": 2.3117,
+      "step": 15330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3690207917003275e-05,
+      "loss": 2.2531,
+      "step": 15340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3685956035545731e-05,
+      "loss": 2.2214,
+      "step": 15350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3681704154088184e-05,
+      "loss": 2.3255,
+      "step": 15360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.367745227263064e-05,
+      "loss": 2.5794,
+      "step": 15370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3673200391173096e-05,
+      "loss": 2.2217,
+      "step": 15380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3668948509715549e-05,
+      "loss": 2.5223,
+      "step": 15390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3664696628258005e-05,
+      "loss": 2.2301,
+      "step": 15400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3660444746800461e-05,
+      "loss": 2.6519,
+      "step": 15410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3656192865342914e-05,
+      "loss": 2.2755,
+      "step": 15420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.365194098388537e-05,
+      "loss": 2.2666,
+      "step": 15430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3647689102427826e-05,
+      "loss": 2.3709,
+      "step": 15440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3643437220970281e-05,
+      "loss": 2.6039,
+      "step": 15450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3639185339512735e-05,
+      "loss": 2.5073,
+      "step": 15460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.363493345805519e-05,
+      "loss": 2.1521,
+      "step": 15470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3630681576597646e-05,
+      "loss": 2.7038,
+      "step": 15480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.36264296951401e-05,
+      "loss": 2.6435,
+      "step": 15490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3622177813682555e-05,
+      "loss": 1.9067,
+      "step": 15500
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 1.8592156171798706,
+      "eval_runtime": 175.7012,
+      "eval_samples_per_second": 14.246,
+      "eval_steps_per_second": 7.126,
+      "step": 15500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3617925932225011e-05,
+      "loss": 2.1339,
+      "step": 15510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3613674050767466e-05,
+      "loss": 2.3421,
+      "step": 15520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.360942216930992e-05,
+      "loss": 1.9724,
+      "step": 15530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3605170287852376e-05,
+      "loss": 2.4298,
+      "step": 15540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3600918406394832e-05,
+      "loss": 2.6905,
+      "step": 15550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3596666524937285e-05,
+      "loss": 2.1393,
+      "step": 15560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3592414643479741e-05,
+      "loss": 2.4432,
+      "step": 15570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3588162762022197e-05,
+      "loss": 2.36,
+      "step": 15580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.358391088056465e-05,
+      "loss": 2.3451,
+      "step": 15590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3579658999107106e-05,
+      "loss": 2.3377,
+      "step": 15600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3575407117649562e-05,
+      "loss": 2.1833,
+      "step": 15610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3571155236192015e-05,
+      "loss": 2.0854,
+      "step": 15620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3566903354734471e-05,
+      "loss": 2.6914,
+      "step": 15630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3562651473276926e-05,
+      "loss": 2.4138,
+      "step": 15640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3558399591819382e-05,
+      "loss": 2.6276,
+      "step": 15650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3554147710361836e-05,
+      "loss": 2.2723,
+      "step": 15660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3549895828904291e-05,
+      "loss": 1.9168,
+      "step": 15670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3545643947446747e-05,
+      "loss": 2.207,
+      "step": 15680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3541392065989201e-05,
+      "loss": 2.1042,
+      "step": 15690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3537140184531656e-05,
+      "loss": 2.593,
+      "step": 15700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3532888303074112e-05,
+      "loss": 1.8281,
+      "step": 15710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3528636421616565e-05,
+      "loss": 2.3192,
+      "step": 15720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3524384540159021e-05,
+      "loss": 2.4385,
+      "step": 15730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3520132658701477e-05,
+      "loss": 2.2035,
+      "step": 15740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.351588077724393e-05,
+      "loss": 2.3731,
+      "step": 15750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3511628895786386e-05,
+      "loss": 2.3403,
+      "step": 15760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3507377014328842e-05,
+      "loss": 1.8895,
+      "step": 15770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3503125132871298e-05,
+      "loss": 1.9967,
+      "step": 15780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3498873251413751e-05,
+      "loss": 2.3283,
+      "step": 15790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3494621369956207e-05,
+      "loss": 2.2754,
+      "step": 15800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3490369488498662e-05,
+      "loss": 2.3961,
+      "step": 15810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3486117607041116e-05,
+      "loss": 2.1002,
+      "step": 15820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3481865725583572e-05,
+      "loss": 2.339,
+      "step": 15830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3477613844126027e-05,
+      "loss": 2.7073,
+      "step": 15840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3473361962668481e-05,
+      "loss": 2.6434,
+      "step": 15850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3469110081210936e-05,
+      "loss": 2.0592,
+      "step": 15860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3464858199753392e-05,
+      "loss": 2.3421,
+      "step": 15870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3460606318295848e-05,
+      "loss": 2.53,
+      "step": 15880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.34563544368383e-05,
+      "loss": 1.9826,
+      "step": 15890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3452102555380757e-05,
+      "loss": 2.6527,
+      "step": 15900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3447850673923213e-05,
+      "loss": 2.0403,
+      "step": 15910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3443598792465666e-05,
+      "loss": 2.3132,
+      "step": 15920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3439346911008122e-05,
+      "loss": 2.2089,
+      "step": 15930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3435095029550578e-05,
+      "loss": 2.6101,
+      "step": 15940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3430843148093031e-05,
+      "loss": 2.5606,
+      "step": 15950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3426591266635487e-05,
+      "loss": 2.819,
+      "step": 15960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3422339385177943e-05,
+      "loss": 2.1818,
+      "step": 15970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3418087503720398e-05,
+      "loss": 2.4026,
+      "step": 15980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3413835622262852e-05,
+      "loss": 2.4839,
+      "step": 15990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3409583740805308e-05,
+      "loss": 2.6198,
+      "step": 16000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.8519717454910278,
+      "eval_runtime": 176.1518,
+      "eval_samples_per_second": 14.209,
+      "eval_steps_per_second": 7.108,
+      "step": 16000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3405331859347763e-05,
+      "loss": 2.5192,
+      "step": 16010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3401079977890217e-05,
+      "loss": 2.1524,
+      "step": 16020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3396828096432672e-05,
+      "loss": 2.0034,
+      "step": 16030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3392576214975128e-05,
+      "loss": 1.9832,
+      "step": 16040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3388324333517582e-05,
+      "loss": 2.32,
+      "step": 16050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3384072452060037e-05,
+      "loss": 2.4981,
+      "step": 16060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3379820570602493e-05,
+      "loss": 2.0976,
+      "step": 16070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3375568689144949e-05,
+      "loss": 2.4842,
+      "step": 16080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3371316807687402e-05,
+      "loss": 2.2168,
+      "step": 16090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3367064926229858e-05,
+      "loss": 2.4269,
+      "step": 16100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3362813044772314e-05,
+      "loss": 2.3498,
+      "step": 16110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3358561163314767e-05,
+      "loss": 2.5344,
+      "step": 16120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3354309281857223e-05,
+      "loss": 2.1391,
+      "step": 16130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.335005740039968e-05,
+      "loss": 2.1381,
+      "step": 16140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3345805518942132e-05,
+      "loss": 1.8247,
+      "step": 16150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3341553637484588e-05,
+      "loss": 2.121,
+      "step": 16160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3337301756027044e-05,
+      "loss": 1.4757,
+      "step": 16170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3333049874569499e-05,
+      "loss": 2.389,
+      "step": 16180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3328797993111953e-05,
+      "loss": 2.8002,
+      "step": 16190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3324546111654408e-05,
+      "loss": 2.3572,
+      "step": 16200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3320294230196864e-05,
+      "loss": 2.1279,
+      "step": 16210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3316042348739318e-05,
+      "loss": 2.2369,
+      "step": 16220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3311790467281773e-05,
+      "loss": 2.3375,
+      "step": 16230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3307538585824229e-05,
+      "loss": 2.2948,
+      "step": 16240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3303286704366682e-05,
+      "loss": 1.8357,
+      "step": 16250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3299034822909138e-05,
+      "loss": 2.1718,
+      "step": 16260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3294782941451594e-05,
+      "loss": 2.1868,
+      "step": 16270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.3290531059994047e-05,
+      "loss": 2.158,
+      "step": 16280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3286279178536503e-05,
+      "loss": 2.0758,
+      "step": 16290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3282027297078959e-05,
+      "loss": 2.027,
+      "step": 16300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3277775415621415e-05,
+      "loss": 2.3991,
+      "step": 16310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3273523534163868e-05,
+      "loss": 2.4214,
+      "step": 16320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3269271652706324e-05,
+      "loss": 2.2119,
+      "step": 16330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3265019771248779e-05,
+      "loss": 2.4542,
+      "step": 16340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3260767889791233e-05,
+      "loss": 1.8715,
+      "step": 16350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3256516008333689e-05,
+      "loss": 2.2014,
+      "step": 16360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3252264126876144e-05,
+      "loss": 2.7697,
+      "step": 16370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3248012245418598e-05,
+      "loss": 2.712,
+      "step": 16380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3243760363961054e-05,
+      "loss": 2.3395,
+      "step": 16390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3239508482503509e-05,
+      "loss": 2.0136,
+      "step": 16400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3235256601045965e-05,
+      "loss": 2.5494,
+      "step": 16410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3231004719588418e-05,
+      "loss": 2.2552,
+      "step": 16420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3226752838130874e-05,
+      "loss": 1.9919,
+      "step": 16430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.322250095667333e-05,
+      "loss": 2.3593,
+      "step": 16440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3218249075215783e-05,
+      "loss": 2.2038,
+      "step": 16450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3213997193758239e-05,
+      "loss": 2.053,
+      "step": 16460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3209745312300695e-05,
+      "loss": 2.644,
+      "step": 16470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3205493430843148e-05,
+      "loss": 2.0581,
+      "step": 16480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3201241549385604e-05,
+      "loss": 2.167,
+      "step": 16490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.319698966792806e-05,
+      "loss": 2.1646,
+      "step": 16500
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 1.8592852354049683,
+      "eval_runtime": 175.6489,
+      "eval_samples_per_second": 14.25,
+      "eval_steps_per_second": 7.128,
+      "step": 16500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.3192737786470514e-05,
+      "loss": 2.2407,
+      "step": 16510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3188485905012969e-05,
+      "loss": 2.7319,
+      "step": 16520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3184234023555425e-05,
+      "loss": 2.0932,
+      "step": 16530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.317998214209788e-05,
+      "loss": 2.2724,
+      "step": 16540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3175730260640334e-05,
+      "loss": 2.4698,
+      "step": 16550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.317147837918279e-05,
+      "loss": 2.039,
+      "step": 16560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3167226497725245e-05,
+      "loss": 2.1941,
+      "step": 16570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3162974616267699e-05,
+      "loss": 2.3781,
+      "step": 16580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3158722734810154e-05,
+      "loss": 2.2378,
+      "step": 16590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.315447085335261e-05,
+      "loss": 2.2077,
+      "step": 16600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3150218971895066e-05,
+      "loss": 2.5321,
+      "step": 16610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3145967090437519e-05,
+      "loss": 1.9455,
+      "step": 16620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3141715208979975e-05,
+      "loss": 2.1657,
+      "step": 16630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3137463327522431e-05,
+      "loss": 2.3863,
+      "step": 16640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3133211446064884e-05,
+      "loss": 2.2692,
+      "step": 16650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.312895956460734e-05,
+      "loss": 1.9171,
+      "step": 16660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3124707683149796e-05,
+      "loss": 2.3214,
+      "step": 16670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3120455801692249e-05,
+      "loss": 2.3556,
+      "step": 16680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3116203920234705e-05,
+      "loss": 2.1717,
+      "step": 16690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3111952038777161e-05,
+      "loss": 2.0603,
+      "step": 16700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3107700157319616e-05,
+      "loss": 2.3345,
+      "step": 16710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.310344827586207e-05,
+      "loss": 2.1033,
+      "step": 16720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3099196394404526e-05,
+      "loss": 1.9156,
+      "step": 16730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.309494451294698e-05,
+      "loss": 1.9722,
+      "step": 16740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.3090692631489435e-05,
+      "loss": 2.5184,
+      "step": 16750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.308644075003189e-05,
+      "loss": 2.248,
+      "step": 16760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3082188868574346e-05,
+      "loss": 2.654,
+      "step": 16770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.30779369871168e-05,
+      "loss": 1.7152,
+      "step": 16780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3073685105659255e-05,
+      "loss": 2.1579,
+      "step": 16790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.306943322420171e-05,
+      "loss": 2.4243,
+      "step": 16800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3065181342744163e-05,
+      "loss": 1.8675,
+      "step": 16810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.306092946128662e-05,
+      "loss": 1.998,
+      "step": 16820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3056677579829076e-05,
+      "loss": 2.5531,
+      "step": 16830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3052425698371532e-05,
+      "loss": 2.0088,
+      "step": 16840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3048173816913985e-05,
+      "loss": 2.0293,
+      "step": 16850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3043921935456441e-05,
+      "loss": 2.1525,
+      "step": 16860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3039670053998897e-05,
+      "loss": 2.0561,
+      "step": 16870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.303541817254135e-05,
+      "loss": 2.2735,
+      "step": 16880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3031166291083806e-05,
+      "loss": 2.3812,
+      "step": 16890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.302691440962626e-05,
+      "loss": 1.9527,
+      "step": 16900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3022662528168715e-05,
+      "loss": 1.9467,
+      "step": 16910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3018410646711171e-05,
+      "loss": 2.245,
+      "step": 16920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3014158765253625e-05,
+      "loss": 1.9354,
+      "step": 16930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3009906883796082e-05,
+      "loss": 2.4325,
+      "step": 16940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.3005655002338536e-05,
+      "loss": 2.4116,
+      "step": 16950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.300140312088099e-05,
+      "loss": 2.5103,
+      "step": 16960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2997151239423447e-05,
+      "loss": 2.2477,
+      "step": 16970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.29928993579659e-05,
+      "loss": 2.0229,
+      "step": 16980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.2988647476508356e-05,
+      "loss": 2.4175,
+      "step": 16990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2984395595050812e-05,
+      "loss": 2.0822,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 1.8584306240081787,
+      "eval_runtime": 175.881,
+      "eval_samples_per_second": 14.231,
+      "eval_steps_per_second": 7.118,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2980143713593265e-05,
+      "loss": 1.9894,
+      "step": 17010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.297589183213572e-05,
+      "loss": 2.0724,
+      "step": 17020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2971639950678177e-05,
+      "loss": 1.8834,
+      "step": 17030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2967388069220633e-05,
+      "loss": 2.3956,
+      "step": 17040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2963136187763086e-05,
+      "loss": 2.1201,
+      "step": 17050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2958884306305542e-05,
+      "loss": 2.2739,
+      "step": 17060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2954632424847996e-05,
+      "loss": 2.5214,
+      "step": 17070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.295038054339045e-05,
+      "loss": 2.524,
+      "step": 17080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2946128661932907e-05,
+      "loss": 2.0987,
+      "step": 17090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2941876780475361e-05,
+      "loss": 2.6096,
+      "step": 17100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2937624899017816e-05,
+      "loss": 2.0223,
+      "step": 17110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2933373017560272e-05,
+      "loss": 2.1142,
+      "step": 17120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2929121136102726e-05,
+      "loss": 2.5442,
+      "step": 17130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2924869254645183e-05,
+      "loss": 2.6321,
+      "step": 17140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2920617373187635e-05,
+      "loss": 2.2613,
+      "step": 17150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2916365491730092e-05,
+      "loss": 2.1481,
+      "step": 17160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2912113610272548e-05,
+      "loss": 2.2593,
+      "step": 17170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2907861728815e-05,
+      "loss": 2.3222,
+      "step": 17180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2903609847357457e-05,
+      "loss": 2.1073,
+      "step": 17190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2899357965899913e-05,
+      "loss": 2.5689,
+      "step": 17200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2895106084442366e-05,
+      "loss": 2.3708,
+      "step": 17210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2890854202984822e-05,
+      "loss": 2.4703,
+      "step": 17220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.2886602321527278e-05,
+      "loss": 2.4671,
+      "step": 17230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2882350440069732e-05,
+      "loss": 2.3027,
+      "step": 17240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2878098558612187e-05,
+      "loss": 2.0178,
+      "step": 17250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2873846677154643e-05,
+      "loss": 2.0933,
+      "step": 17260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2869594795697097e-05,
+      "loss": 2.7651,
+      "step": 17270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2865342914239552e-05,
+      "loss": 2.5752,
+      "step": 17280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2861091032782006e-05,
+      "loss": 2.0839,
+      "step": 17290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2856839151324462e-05,
+      "loss": 2.5019,
+      "step": 17300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2852587269866917e-05,
+      "loss": 2.5049,
+      "step": 17310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2848335388409371e-05,
+      "loss": 2.2519,
+      "step": 17320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2844083506951828e-05,
+      "loss": 2.8233,
+      "step": 17330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2839831625494282e-05,
+      "loss": 2.3101,
+      "step": 17340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2835579744036736e-05,
+      "loss": 2.232,
+      "step": 17350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2831327862579193e-05,
+      "loss": 2.0227,
+      "step": 17360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2827075981121649e-05,
+      "loss": 2.5267,
+      "step": 17370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2822824099664102e-05,
+      "loss": 2.5093,
+      "step": 17380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2818572218206558e-05,
+      "loss": 1.6604,
+      "step": 17390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2814320336749014e-05,
+      "loss": 2.7735,
+      "step": 17400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2810068455291467e-05,
+      "loss": 1.9258,
+      "step": 17410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2805816573833923e-05,
+      "loss": 2.1722,
+      "step": 17420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2801564692376379e-05,
+      "loss": 2.4616,
+      "step": 17430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2797312810918832e-05,
+      "loss": 2.2542,
+      "step": 17440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2793060929461288e-05,
+      "loss": 1.9519,
+      "step": 17450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2788809048003742e-05,
+      "loss": 2.3988,
+      "step": 17460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.2784557166546198e-05,
+      "loss": 2.1753,
+      "step": 17470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2780305285088653e-05,
+      "loss": 2.139,
+      "step": 17480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2776053403631107e-05,
+      "loss": 2.3725,
+      "step": 17490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2771801522173563e-05,
+      "loss": 2.574,
+      "step": 17500
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 1.8318504095077515,
+      "eval_runtime": 175.9495,
+      "eval_samples_per_second": 14.226,
+      "eval_steps_per_second": 7.116,
+      "step": 17500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2767549640716018e-05,
+      "loss": 2.6461,
+      "step": 17510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2763297759258472e-05,
+      "loss": 2.26,
+      "step": 17520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2759045877800929e-05,
+      "loss": 1.8639,
+      "step": 17530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2754793996343381e-05,
+      "loss": 2.3485,
+      "step": 17540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2750542114885837e-05,
+      "loss": 2.3243,
+      "step": 17550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2746290233428294e-05,
+      "loss": 2.317,
+      "step": 17560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.274203835197075e-05,
+      "loss": 2.4058,
+      "step": 17570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2737786470513203e-05,
+      "loss": 2.0532,
+      "step": 17580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2733534589055659e-05,
+      "loss": 2.863,
+      "step": 17590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2729282707598115e-05,
+      "loss": 1.7981,
+      "step": 17600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2725030826140568e-05,
+      "loss": 2.1557,
+      "step": 17610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2720778944683024e-05,
+      "loss": 2.5958,
+      "step": 17620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2716527063225478e-05,
+      "loss": 2.1916,
+      "step": 17630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2712275181767933e-05,
+      "loss": 2.1137,
+      "step": 17640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2708023300310389e-05,
+      "loss": 2.4346,
+      "step": 17650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2703771418852843e-05,
+      "loss": 2.2496,
+      "step": 17660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.26995195373953e-05,
+      "loss": 2.2978,
+      "step": 17670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2695267655937752e-05,
+      "loss": 2.3618,
+      "step": 17680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2691015774480208e-05,
+      "loss": 2.6477,
+      "step": 17690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2686763893022665e-05,
+      "loss": 2.3733,
+      "step": 17700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2682512011565117e-05,
+      "loss": 1.8352,
+      "step": 17710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2678260130107573e-05,
+      "loss": 1.903,
+      "step": 17720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.267400824865003e-05,
+      "loss": 2.4592,
+      "step": 17730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2669756367192482e-05,
+      "loss": 2.0751,
+      "step": 17740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2665504485734938e-05,
+      "loss": 2.1852,
+      "step": 17750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2661252604277395e-05,
+      "loss": 1.8026,
+      "step": 17760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2657000722819849e-05,
+      "loss": 2.2377,
+      "step": 17770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2652748841362304e-05,
+      "loss": 2.654,
+      "step": 17780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.264849695990476e-05,
+      "loss": 1.941,
+      "step": 17790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2644245078447214e-05,
+      "loss": 2.1042,
+      "step": 17800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2639993196989669e-05,
+      "loss": 2.4165,
+      "step": 17810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2635741315532125e-05,
+      "loss": 2.0932,
+      "step": 17820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.263148943407458e-05,
+      "loss": 2.3665,
+      "step": 17830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2627237552617034e-05,
+      "loss": 2.376,
+      "step": 17840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2622985671159488e-05,
+      "loss": 2.4151,
+      "step": 17850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2618733789701944e-05,
+      "loss": 2.1847,
+      "step": 17860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2614481908244399e-05,
+      "loss": 2.3342,
+      "step": 17870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2610230026786853e-05,
+      "loss": 2.1517,
+      "step": 17880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.260597814532931e-05,
+      "loss": 2.0448,
+      "step": 17890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2601726263871766e-05,
+      "loss": 2.2522,
+      "step": 17900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2597474382414218e-05,
+      "loss": 2.1466,
+      "step": 17910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2593222500956674e-05,
+      "loss": 2.3031,
+      "step": 17920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.258897061949913e-05,
+      "loss": 2.1069,
+      "step": 17930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2584718738041583e-05,
+      "loss": 2.1193,
+      "step": 17940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.258046685658404e-05,
+      "loss": 1.9281,
+      "step": 17950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2576214975126496e-05,
+      "loss": 1.7614,
+      "step": 17960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2571963093668948e-05,
+      "loss": 1.954,
+      "step": 17970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2567711212211405e-05,
+      "loss": 2.1048,
+      "step": 17980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.256345933075386e-05,
+      "loss": 2.3576,
+      "step": 17990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2559207449296315e-05,
+      "loss": 2.5629,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 1.8561279773712158,
+      "eval_runtime": 176.0578,
+      "eval_samples_per_second": 14.217,
+      "eval_steps_per_second": 7.111,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.255495556783877e-05,
+      "loss": 2.1933,
+      "step": 18010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2550703686381224e-05,
+      "loss": 2.3697,
+      "step": 18020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.254645180492368e-05,
+      "loss": 2.3221,
+      "step": 18030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2542199923466135e-05,
+      "loss": 2.3338,
+      "step": 18040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.253794804200859e-05,
+      "loss": 2.5145,
+      "step": 18050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2533696160551045e-05,
+      "loss": 2.0809,
+      "step": 18060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2529444279093498e-05,
+      "loss": 2.5053,
+      "step": 18070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2525192397635954e-05,
+      "loss": 1.9817,
+      "step": 18080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.252094051617841e-05,
+      "loss": 2.1424,
+      "step": 18090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2516688634720867e-05,
+      "loss": 2.3833,
+      "step": 18100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.251243675326332e-05,
+      "loss": 2.2463,
+      "step": 18110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2508184871805775e-05,
+      "loss": 2.4337,
+      "step": 18120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2503932990348232e-05,
+      "loss": 2.3263,
+      "step": 18130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2499681108890684e-05,
+      "loss": 2.4375,
+      "step": 18140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.249542922743314e-05,
+      "loss": 1.8182,
+      "step": 18150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2491177345975595e-05,
+      "loss": 2.3109,
+      "step": 18160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.248692546451805e-05,
+      "loss": 2.5621,
+      "step": 18170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2482673583060506e-05,
+      "loss": 2.2866,
+      "step": 18180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.247842170160296e-05,
+      "loss": 2.0402,
+      "step": 18190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2474169820145416e-05,
+      "loss": 1.9998,
+      "step": 18200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.246991793868787e-05,
+      "loss": 1.9806,
+      "step": 18210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2465666057230325e-05,
+      "loss": 2.3714,
+      "step": 18220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2461414175772781e-05,
+      "loss": 2.431,
+      "step": 18230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2457162294315234e-05,
+      "loss": 2.3278,
+      "step": 18240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.245291041285769e-05,
+      "loss": 2.4033,
+      "step": 18250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2448658531400146e-05,
+      "loss": 2.2226,
+      "step": 18260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2444406649942599e-05,
+      "loss": 2.2681,
+      "step": 18270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2440154768485055e-05,
+      "loss": 2.4701,
+      "step": 18280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2435902887027511e-05,
+      "loss": 2.611,
+      "step": 18290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2431651005569964e-05,
+      "loss": 1.9993,
+      "step": 18300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.242739912411242e-05,
+      "loss": 1.9156,
+      "step": 18310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2423147242654876e-05,
+      "loss": 2.6014,
+      "step": 18320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2418895361197331e-05,
+      "loss": 2.1876,
+      "step": 18330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2414643479739785e-05,
+      "loss": 2.2175,
+      "step": 18340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2410391598282242e-05,
+      "loss": 2.2158,
+      "step": 18350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2406139716824696e-05,
+      "loss": 2.3834,
+      "step": 18360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.240188783536715e-05,
+      "loss": 1.9719,
+      "step": 18370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2397635953909607e-05,
+      "loss": 2.4343,
+      "step": 18380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2393384072452061e-05,
+      "loss": 2.2525,
+      "step": 18390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2389132190994516e-05,
+      "loss": 2.2265,
+      "step": 18400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.238488030953697e-05,
+      "loss": 2.1699,
+      "step": 18410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2380628428079426e-05,
+      "loss": 2.5775,
+      "step": 18420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2376376546621882e-05,
+      "loss": 2.1577,
+      "step": 18430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2372124665164335e-05,
+      "loss": 2.1021,
+      "step": 18440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2367872783706791e-05,
+      "loss": 2.3303,
+      "step": 18450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2363620902249247e-05,
+      "loss": 2.2932,
+      "step": 18460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.23593690207917e-05,
+      "loss": 2.7011,
+      "step": 18470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2355117139334156e-05,
+      "loss": 2.1523,
+      "step": 18480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2350865257876612e-05,
+      "loss": 2.3008,
+      "step": 18490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2346613376419065e-05,
+      "loss": 1.9752,
+      "step": 18500
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.8424828052520752,
+      "eval_runtime": 174.3309,
+      "eval_samples_per_second": 14.358,
+      "eval_steps_per_second": 7.182,
+      "step": 18500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2342361494961521e-05,
+      "loss": 2.284,
+      "step": 18510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2338109613503978e-05,
+      "loss": 2.1015,
+      "step": 18520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2333857732046432e-05,
+      "loss": 2.2809,
+      "step": 18530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2329605850588886e-05,
+      "loss": 2.3151,
+      "step": 18540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2325353969131341e-05,
+      "loss": 1.662,
+      "step": 18550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2321102087673797e-05,
+      "loss": 1.9858,
+      "step": 18560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2316850206216252e-05,
+      "loss": 2.5695,
+      "step": 18570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2312598324758706e-05,
+      "loss": 2.8181,
+      "step": 18580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2308346443301162e-05,
+      "loss": 2.388,
+      "step": 18590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2304094561843617e-05,
+      "loss": 2.2133,
+      "step": 18600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2299842680386071e-05,
+      "loss": 2.3169,
+      "step": 18610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2295590798928527e-05,
+      "loss": 1.9527,
+      "step": 18620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2291338917470983e-05,
+      "loss": 2.2607,
+      "step": 18630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2287087036013436e-05,
+      "loss": 2.5624,
+      "step": 18640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2282835154555892e-05,
+      "loss": 2.2322,
+      "step": 18650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2278583273098348e-05,
+      "loss": 2.1425,
+      "step": 18660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2274331391640801e-05,
+      "loss": 1.9878,
+      "step": 18670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2270079510183257e-05,
+      "loss": 2.6583,
+      "step": 18680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2265827628725713e-05,
+      "loss": 2.3085,
+      "step": 18690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2261575747268166e-05,
+      "loss": 2.4383,
+      "step": 18700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2257323865810622e-05,
+      "loss": 2.8947,
+      "step": 18710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2253071984353077e-05,
+      "loss": 2.5398,
+      "step": 18720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2248820102895533e-05,
+      "loss": 2.4291,
+      "step": 18730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2244568221437987e-05,
+      "loss": 2.5805,
+      "step": 18740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2240316339980442e-05,
+      "loss": 2.2642,
+      "step": 18750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2236064458522898e-05,
+      "loss": 2.3821,
+      "step": 18760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2231812577065353e-05,
+      "loss": 1.8382,
+      "step": 18770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2227560695607807e-05,
+      "loss": 1.8444,
+      "step": 18780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2223308814150263e-05,
+      "loss": 2.4173,
+      "step": 18790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2219056932692716e-05,
+      "loss": 2.3083,
+      "step": 18800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2214805051235172e-05,
+      "loss": 2.6496,
+      "step": 18810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2210553169777628e-05,
+      "loss": 1.9028,
+      "step": 18820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2206301288320081e-05,
+      "loss": 2.0677,
+      "step": 18830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2202049406862537e-05,
+      "loss": 2.0836,
+      "step": 18840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2197797525404993e-05,
+      "loss": 2.4567,
+      "step": 18850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.219354564394745e-05,
+      "loss": 2.4262,
+      "step": 18860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2189293762489902e-05,
+      "loss": 2.5371,
+      "step": 18870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2185041881032358e-05,
+      "loss": 2.4472,
+      "step": 18880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2180789999574813e-05,
+      "loss": 2.341,
+      "step": 18890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2176538118117267e-05,
+      "loss": 2.3559,
+      "step": 18900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2172286236659723e-05,
+      "loss": 1.8539,
+      "step": 18910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2168034355202178e-05,
+      "loss": 2.2824,
+      "step": 18920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2163782473744632e-05,
+      "loss": 1.9494,
+      "step": 18930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2159530592287088e-05,
+      "loss": 2.4157,
+      "step": 18940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2155278710829543e-05,
+      "loss": 2.6123,
+      "step": 18950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2151026829371999e-05,
+      "loss": 2.6423,
+      "step": 18960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2146774947914452e-05,
+      "loss": 1.8365,
+      "step": 18970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2142523066456908e-05,
+      "loss": 2.1457,
+      "step": 18980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2138271184999364e-05,
+      "loss": 1.8823,
+      "step": 18990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2134019303541817e-05,
+      "loss": 1.9448,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.8287702798843384,
+      "eval_runtime": 175.7832,
+      "eval_samples_per_second": 14.239,
+      "eval_steps_per_second": 7.122,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2129767422084273e-05,
+      "loss": 2.3223,
+      "step": 19010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.212551554062673e-05,
+      "loss": 2.3445,
+      "step": 19020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2121263659169182e-05,
+      "loss": 2.1137,
+      "step": 19030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2117011777711638e-05,
+      "loss": 2.4059,
+      "step": 19040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2112759896254094e-05,
+      "loss": 2.6698,
+      "step": 19050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2108508014796549e-05,
+      "loss": 2.4193,
+      "step": 19060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2104256133339003e-05,
+      "loss": 1.9789,
+      "step": 19070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.210000425188146e-05,
+      "loss": 2.1658,
+      "step": 19080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2095752370423914e-05,
+      "loss": 2.1202,
+      "step": 19090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2091500488966368e-05,
+      "loss": 2.3003,
+      "step": 19100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2087248607508823e-05,
+      "loss": 2.6162,
+      "step": 19110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2082996726051279e-05,
+      "loss": 2.3541,
+      "step": 19120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2078744844593733e-05,
+      "loss": 2.4186,
+      "step": 19130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2074492963136188e-05,
+      "loss": 2.9454,
+      "step": 19140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2070241081678644e-05,
+      "loss": 2.0989,
+      "step": 19150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.20659892002211e-05,
+      "loss": 2.4327,
+      "step": 19160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2061737318763553e-05,
+      "loss": 2.6236,
+      "step": 19170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2057485437306009e-05,
+      "loss": 2.5432,
+      "step": 19180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2053233555848465e-05,
+      "loss": 2.1143,
+      "step": 19190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2048981674390918e-05,
+      "loss": 2.2297,
+      "step": 19200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2044729792933374e-05,
+      "loss": 2.3105,
+      "step": 19210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.204047791147583e-05,
+      "loss": 2.1033,
+      "step": 19220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2036226030018283e-05,
+      "loss": 1.9961,
+      "step": 19230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.203197414856074e-05,
+      "loss": 1.9142,
+      "step": 19240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2027722267103195e-05,
+      "loss": 2.2007,
+      "step": 19250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.202347038564565e-05,
+      "loss": 2.5775,
+      "step": 19260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2019218504188104e-05,
+      "loss": 1.6784,
+      "step": 19270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2014966622730559e-05,
+      "loss": 1.8458,
+      "step": 19280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2010714741273015e-05,
+      "loss": 2.3964,
+      "step": 19290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.200646285981547e-05,
+      "loss": 2.0427,
+      "step": 19300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.2002210978357924e-05,
+      "loss": 1.8841,
+      "step": 19310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.199795909690038e-05,
+      "loss": 2.3461,
+      "step": 19320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1993707215442834e-05,
+      "loss": 2.5327,
+      "step": 19330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1989455333985289e-05,
+      "loss": 2.0839,
+      "step": 19340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1985203452527745e-05,
+      "loss": 2.2541,
+      "step": 19350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1980951571070198e-05,
+      "loss": 2.2845,
+      "step": 19360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1976699689612654e-05,
+      "loss": 2.4602,
+      "step": 19370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.197244780815511e-05,
+      "loss": 1.7821,
+      "step": 19380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1968195926697566e-05,
+      "loss": 2.02,
+      "step": 19390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1963944045240019e-05,
+      "loss": 1.92,
+      "step": 19400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1959692163782475e-05,
+      "loss": 1.9421,
+      "step": 19410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1955440282324931e-05,
+      "loss": 2.3449,
+      "step": 19420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1951188400867384e-05,
+      "loss": 2.1307,
+      "step": 19430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.194693651940984e-05,
+      "loss": 2.0286,
+      "step": 19440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1942684637952295e-05,
+      "loss": 2.3532,
+      "step": 19450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1938432756494749e-05,
+      "loss": 1.7903,
+      "step": 19460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1934180875037205e-05,
+      "loss": 2.1869,
+      "step": 19470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.192992899357966e-05,
+      "loss": 2.1925,
+      "step": 19480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1925677112122116e-05,
+      "loss": 2.3387,
+      "step": 19490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1921425230664569e-05,
+      "loss": 2.1027,
+      "step": 19500
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.8289752006530762,
+      "eval_runtime": 175.8843,
+      "eval_samples_per_second": 14.231,
+      "eval_steps_per_second": 7.118,
+      "step": 19500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1917173349207025e-05,
+      "loss": 2.2901,
+      "step": 19510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1912921467749481e-05,
+      "loss": 2.0443,
+      "step": 19520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1908669586291934e-05,
+      "loss": 2.1688,
+      "step": 19530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.190441770483439e-05,
+      "loss": 2.1367,
+      "step": 19540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1900165823376846e-05,
+      "loss": 2.14,
+      "step": 19550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1895913941919299e-05,
+      "loss": 2.0942,
+      "step": 19560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1891662060461755e-05,
+      "loss": 2.3131,
+      "step": 19570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1887410179004211e-05,
+      "loss": 1.9678,
+      "step": 19580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.1883158297546666e-05,
+      "loss": 2.3818,
+      "step": 19590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.187890641608912e-05,
+      "loss": 2.1617,
+      "step": 19600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1874654534631576e-05,
+      "loss": 2.0647,
+      "step": 19610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.187040265317403e-05,
+      "loss": 2.1498,
+      "step": 19620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1866150771716485e-05,
+      "loss": 2.09,
+      "step": 19630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1861898890258941e-05,
+      "loss": 2.3031,
+      "step": 19640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1857647008801396e-05,
+      "loss": 1.7862,
+      "step": 19650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.185339512734385e-05,
+      "loss": 2.3439,
+      "step": 19660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1849143245886305e-05,
+      "loss": 2.2409,
+      "step": 19670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.184489136442876e-05,
+      "loss": 1.838,
+      "step": 19680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1840639482971217e-05,
+      "loss": 2.435,
+      "step": 19690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.183638760151367e-05,
+      "loss": 1.8176,
+      "step": 19700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1832135720056126e-05,
+      "loss": 2.5323,
+      "step": 19710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1827883838598582e-05,
+      "loss": 2.392,
+      "step": 19720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1823631957141035e-05,
+      "loss": 2.6848,
+      "step": 19730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1819380075683491e-05,
+      "loss": 2.3956,
+      "step": 19740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1815128194225947e-05,
+      "loss": 2.0737,
+      "step": 19750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.18108763127684e-05,
+      "loss": 2.4459,
+      "step": 19760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1806624431310856e-05,
+      "loss": 1.9749,
+      "step": 19770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1802372549853312e-05,
+      "loss": 2.01,
+      "step": 19780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1798120668395767e-05,
+      "loss": 1.8578,
+      "step": 19790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1793868786938221e-05,
+      "loss": 2.1742,
+      "step": 19800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1789616905480677e-05,
+      "loss": 2.3556,
+      "step": 19810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1785365024023132e-05,
+      "loss": 2.2053,
+      "step": 19820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.1781113142565586e-05,
+      "loss": 2.1769,
+      "step": 19830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.177686126110804e-05,
+      "loss": 2.2273,
+      "step": 19840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1772609379650497e-05,
+      "loss": 2.3378,
+      "step": 19850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1768357498192951e-05,
+      "loss": 2.3385,
+      "step": 19860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1764105616735406e-05,
+      "loss": 2.1603,
+      "step": 19870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1759853735277862e-05,
+      "loss": 2.0272,
+      "step": 19880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1755601853820315e-05,
+      "loss": 2.0062,
+      "step": 19890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.175134997236277e-05,
+      "loss": 2.406,
+      "step": 19900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1747098090905227e-05,
+      "loss": 2.4801,
+      "step": 19910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1742846209447683e-05,
+      "loss": 1.9527,
+      "step": 19920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1738594327990136e-05,
+      "loss": 2.2372,
+      "step": 19930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1734342446532592e-05,
+      "loss": 2.4285,
+      "step": 19940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1730090565075048e-05,
+      "loss": 2.214,
+      "step": 19950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1725838683617501e-05,
+      "loss": 2.6894,
+      "step": 19960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1721586802159957e-05,
+      "loss": 2.2481,
+      "step": 19970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1717334920702411e-05,
+      "loss": 2.3957,
+      "step": 19980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1713083039244866e-05,
+      "loss": 2.1499,
+      "step": 19990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1708831157787322e-05,
+      "loss": 2.3084,
+      "step": 20000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 1.8395805358886719,
+      "eval_runtime": 175.6854,
+      "eval_samples_per_second": 14.247,
+      "eval_steps_per_second": 7.126,
+      "step": 20000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1704579276329777e-05,
+      "loss": 2.11,
+      "step": 20010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1700327394872233e-05,
+      "loss": 1.8904,
+      "step": 20020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1696075513414687e-05,
+      "loss": 2.3283,
+      "step": 20030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1691823631957142e-05,
+      "loss": 1.8581,
+      "step": 20040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1687571750499598e-05,
+      "loss": 2.0448,
+      "step": 20050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.168331986904205e-05,
+      "loss": 2.0497,
+      "step": 20060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1679067987584507e-05,
+      "loss": 2.1895,
+      "step": 20070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1674816106126963e-05,
+      "loss": 1.7949,
+      "step": 20080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1670564224669416e-05,
+      "loss": 2.0302,
+      "step": 20090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1666312343211872e-05,
+      "loss": 2.1369,
+      "step": 20100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1662060461754328e-05,
+      "loss": 2.1922,
+      "step": 20110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1657808580296784e-05,
+      "loss": 2.4506,
+      "step": 20120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1653556698839237e-05,
+      "loss": 2.4267,
+      "step": 20130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1649304817381693e-05,
+      "loss": 1.9311,
+      "step": 20140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1645052935924147e-05,
+      "loss": 2.3057,
+      "step": 20150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1640801054466602e-05,
+      "loss": 1.9138,
+      "step": 20160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1636549173009058e-05,
+      "loss": 2.2328,
+      "step": 20170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1632297291551512e-05,
+      "loss": 2.2676,
+      "step": 20180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1628045410093967e-05,
+      "loss": 2.4848,
+      "step": 20190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1623793528636423e-05,
+      "loss": 2.147,
+      "step": 20200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1619541647178878e-05,
+      "loss": 2.1419,
+      "step": 20210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1615289765721334e-05,
+      "loss": 2.0516,
+      "step": 20220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1611037884263786e-05,
+      "loss": 1.8975,
+      "step": 20230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1606786002806243e-05,
+      "loss": 2.4029,
+      "step": 20240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1602534121348699e-05,
+      "loss": 2.2535,
+      "step": 20250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1598282239891152e-05,
+      "loss": 2.0156,
+      "step": 20260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1594030358433608e-05,
+      "loss": 2.8164,
+      "step": 20270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1589778476976064e-05,
+      "loss": 2.3693,
+      "step": 20280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1585526595518517e-05,
+      "loss": 1.9898,
+      "step": 20290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1581274714060973e-05,
+      "loss": 2.8203,
+      "step": 20300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1577022832603429e-05,
+      "loss": 2.4059,
+      "step": 20310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1572770951145882e-05,
+      "loss": 2.1781,
+      "step": 20320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1568519069688338e-05,
+      "loss": 2.2044,
+      "step": 20330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1564267188230794e-05,
+      "loss": 2.264,
+      "step": 20340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1560015306773248e-05,
+      "loss": 2.4177,
+      "step": 20350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1555763425315703e-05,
+      "loss": 1.8219,
+      "step": 20360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1551511543858157e-05,
+      "loss": 2.5629,
+      "step": 20370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1547259662400614e-05,
+      "loss": 2.4924,
+      "step": 20380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1543007780943068e-05,
+      "loss": 2.3696,
+      "step": 20390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1538755899485522e-05,
+      "loss": 2.1506,
+      "step": 20400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1534504018027979e-05,
+      "loss": 2.3529,
+      "step": 20410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1530252136570433e-05,
+      "loss": 2.3756,
+      "step": 20420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1526000255112887e-05,
+      "loss": 2.5483,
+      "step": 20430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1521748373655344e-05,
+      "loss": 2.2056,
+      "step": 20440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.15174964921978e-05,
+      "loss": 2.2388,
+      "step": 20450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1513244610740253e-05,
+      "loss": 1.8584,
+      "step": 20460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1508992729282709e-05,
+      "loss": 2.2237,
+      "step": 20470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1504740847825165e-05,
+      "loss": 1.9827,
+      "step": 20480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1500488966367618e-05,
+      "loss": 2.5274,
+      "step": 20490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1496237084910074e-05,
+      "loss": 1.9706,
+      "step": 20500
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 1.834281086921692,
+      "eval_runtime": 175.8828,
+      "eval_samples_per_second": 14.231,
+      "eval_steps_per_second": 7.118,
+      "step": 20500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.149198520345253e-05,
+      "loss": 2.2801,
+      "step": 20510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1487733321994983e-05,
+      "loss": 2.4809,
+      "step": 20520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1483481440537439e-05,
+      "loss": 2.5073,
+      "step": 20530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1479229559079893e-05,
+      "loss": 1.6704,
+      "step": 20540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.147497767762235e-05,
+      "loss": 2.1681,
+      "step": 20550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1470725796164804e-05,
+      "loss": 2.3059,
+      "step": 20560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1466473914707258e-05,
+      "loss": 2.4117,
+      "step": 20570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1462222033249715e-05,
+      "loss": 2.028,
+      "step": 20580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1457970151792169e-05,
+      "loss": 2.3308,
+      "step": 20590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1453718270334623e-05,
+      "loss": 2.0801,
+      "step": 20600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.144946638887708e-05,
+      "loss": 2.0363,
+      "step": 20610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1445214507419532e-05,
+      "loss": 2.3273,
+      "step": 20620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1440962625961989e-05,
+      "loss": 2.7378,
+      "step": 20630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1436710744504445e-05,
+      "loss": 2.8054,
+      "step": 20640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.14324588630469e-05,
+      "loss": 2.0178,
+      "step": 20650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1428206981589354e-05,
+      "loss": 2.4087,
+      "step": 20660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.142395510013181e-05,
+      "loss": 2.0017,
+      "step": 20670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1419703218674266e-05,
+      "loss": 2.5347,
+      "step": 20680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1415451337216719e-05,
+      "loss": 2.5313,
+      "step": 20690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1411199455759175e-05,
+      "loss": 2.4035,
+      "step": 20700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.140694757430163e-05,
+      "loss": 2.2844,
+      "step": 20710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1402695692844084e-05,
+      "loss": 2.917,
+      "step": 20720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.139844381138654e-05,
+      "loss": 2.602,
+      "step": 20730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1394191929928994e-05,
+      "loss": 2.2953,
+      "step": 20740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.138994004847145e-05,
+      "loss": 2.0735,
+      "step": 20750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1385688167013905e-05,
+      "loss": 2.2651,
+      "step": 20760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.138143628555636e-05,
+      "loss": 2.5932,
+      "step": 20770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1377184404098816e-05,
+      "loss": 2.3563,
+      "step": 20780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.1372932522641268e-05,
+      "loss": 2.2868,
+      "step": 20790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1368680641183724e-05,
+      "loss": 2.1799,
+      "step": 20800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.136442875972618e-05,
+      "loss": 2.3529,
+      "step": 20810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1360176878268633e-05,
+      "loss": 2.4422,
+      "step": 20820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.135592499681109e-05,
+      "loss": 2.3017,
+      "step": 20830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1351673115353546e-05,
+      "loss": 2.6883,
+      "step": 20840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1347421233895998e-05,
+      "loss": 2.2134,
+      "step": 20850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1343169352438455e-05,
+      "loss": 2.8651,
+      "step": 20860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.133891747098091e-05,
+      "loss": 1.7896,
+      "step": 20870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1334665589523365e-05,
+      "loss": 1.8225,
+      "step": 20880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.133041370806582e-05,
+      "loss": 2.0254,
+      "step": 20890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1326161826608276e-05,
+      "loss": 2.4365,
+      "step": 20900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.132190994515073e-05,
+      "loss": 2.5786,
+      "step": 20910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1317658063693185e-05,
+      "loss": 2.1062,
+      "step": 20920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.131340618223564e-05,
+      "loss": 2.3608,
+      "step": 20930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1309154300778095e-05,
+      "loss": 2.5084,
+      "step": 20940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.130490241932055e-05,
+      "loss": 2.4546,
+      "step": 20950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1300650537863004e-05,
+      "loss": 2.3638,
+      "step": 20960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.129639865640546e-05,
+      "loss": 2.2632,
+      "step": 20970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1292146774947917e-05,
+      "loss": 2.2695,
+      "step": 20980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.128789489349037e-05,
+      "loss": 2.5105,
+      "step": 20990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1283643012032826e-05,
+      "loss": 2.3075,
+      "step": 21000
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 1.822985053062439,
+      "eval_runtime": 175.9116,
+      "eval_samples_per_second": 14.229,
+      "eval_steps_per_second": 7.117,
+      "step": 21000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1279391130575282e-05,
+      "loss": 1.9912,
+      "step": 21010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.1275139249117734e-05,
+      "loss": 2.5211,
+      "step": 21020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.127088736766019e-05,
+      "loss": 2.2959,
+      "step": 21030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1266635486202647e-05,
+      "loss": 2.1357,
+      "step": 21040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.12623836047451e-05,
+      "loss": 2.093,
+      "step": 21050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1258131723287556e-05,
+      "loss": 2.3973,
+      "step": 21060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1253879841830012e-05,
+      "loss": 2.2446,
+      "step": 21070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1249627960372466e-05,
+      "loss": 2.3044,
+      "step": 21080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.124537607891492e-05,
+      "loss": 1.7142,
+      "step": 21090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1241124197457375e-05,
+      "loss": 2.2407,
+      "step": 21100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1236872315999831e-05,
+      "loss": 2.0207,
+      "step": 21110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1232620434542286e-05,
+      "loss": 2.6111,
+      "step": 21120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.122836855308474e-05,
+      "loss": 2.5053,
+      "step": 21130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1224116671627196e-05,
+      "loss": 1.7139,
+      "step": 21140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1219864790169651e-05,
+      "loss": 2.1259,
+      "step": 21150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1215612908712105e-05,
+      "loss": 2.6204,
+      "step": 21160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1211361027254561e-05,
+      "loss": 2.0562,
+      "step": 21170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1207109145797018e-05,
+      "loss": 2.081,
+      "step": 21180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.120285726433947e-05,
+      "loss": 2.3546,
+      "step": 21190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1198605382881927e-05,
+      "loss": 2.3827,
+      "step": 21200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1194353501424383e-05,
+      "loss": 2.1586,
+      "step": 21210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1190101619966835e-05,
+      "loss": 2.1258,
+      "step": 21220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1185849738509292e-05,
+      "loss": 2.5523,
+      "step": 21230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1181597857051748e-05,
+      "loss": 2.3021,
+      "step": 21240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.11773459755942e-05,
+      "loss": 2.0766,
+      "step": 21250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1173094094136657e-05,
+      "loss": 2.3272,
+      "step": 21260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.1168842212679111e-05,
+      "loss": 2.0162,
+      "step": 21270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1164590331221567e-05,
+      "loss": 2.2479,
+      "step": 21280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1160338449764022e-05,
+      "loss": 2.1886,
+      "step": 21290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1156086568306476e-05,
+      "loss": 1.9603,
+      "step": 21300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1151834686848932e-05,
+      "loss": 2.5595,
+      "step": 21310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1147582805391385e-05,
+      "loss": 2.602,
+      "step": 21320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1143330923933841e-05,
+      "loss": 2.0311,
+      "step": 21330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1139079042476297e-05,
+      "loss": 2.3421,
+      "step": 21340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.113482716101875e-05,
+      "loss": 2.2651,
+      "step": 21350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1130575279561206e-05,
+      "loss": 2.3257,
+      "step": 21360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1126323398103662e-05,
+      "loss": 2.0084,
+      "step": 21370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1122071516646115e-05,
+      "loss": 2.7653,
+      "step": 21380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1117819635188571e-05,
+      "loss": 2.2977,
+      "step": 21390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1113567753731028e-05,
+      "loss": 2.0105,
+      "step": 21400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1109315872273482e-05,
+      "loss": 2.3345,
+      "step": 21410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1105063990815936e-05,
+      "loss": 2.4214,
+      "step": 21420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1100812109358393e-05,
+      "loss": 2.5472,
+      "step": 21430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1096560227900847e-05,
+      "loss": 2.2811,
+      "step": 21440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1092308346443302e-05,
+      "loss": 2.3565,
+      "step": 21450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1088056464985758e-05,
+      "loss": 2.026,
+      "step": 21460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1083804583528212e-05,
+      "loss": 1.721,
+      "step": 21470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1079552702070667e-05,
+      "loss": 2.2634,
+      "step": 21480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1075300820613121e-05,
+      "loss": 2.2773,
+      "step": 21490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1071048939155577e-05,
+      "loss": 1.7214,
+      "step": 21500
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 1.824805498123169,
+      "eval_runtime": 174.3778,
+      "eval_samples_per_second": 14.354,
+      "eval_steps_per_second": 7.18,
+      "step": 21500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1066797057698033e-05,
+      "loss": 2.6856,
+      "step": 21510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1062545176240486e-05,
+      "loss": 2.0154,
+      "step": 21520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1058293294782942e-05,
+      "loss": 1.9537,
+      "step": 21530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1054041413325398e-05,
+      "loss": 2.0085,
+      "step": 21540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1049789531867851e-05,
+      "loss": 2.3019,
+      "step": 21550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1045537650410307e-05,
+      "loss": 2.0823,
+      "step": 21560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1041285768952764e-05,
+      "loss": 2.0807,
+      "step": 21570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1037033887495216e-05,
+      "loss": 2.4829,
+      "step": 21580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1032782006037672e-05,
+      "loss": 2.1712,
+      "step": 21590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1028530124580129e-05,
+      "loss": 2.5054,
+      "step": 21600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1024278243122583e-05,
+      "loss": 2.1296,
+      "step": 21610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1020026361665038e-05,
+      "loss": 2.0915,
+      "step": 21620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1015774480207494e-05,
+      "loss": 2.5908,
+      "step": 21630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1011522598749948e-05,
+      "loss": 1.7967,
+      "step": 21640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1007270717292403e-05,
+      "loss": 1.9879,
+      "step": 21650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1003018835834857e-05,
+      "loss": 2.0687,
+      "step": 21660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0998766954377313e-05,
+      "loss": 2.6076,
+      "step": 21670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0994515072919768e-05,
+      "loss": 2.8764,
+      "step": 21680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0990263191462222e-05,
+      "loss": 1.9984,
+      "step": 21690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0986011310004678e-05,
+      "loss": 2.6241,
+      "step": 21700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0981759428547134e-05,
+      "loss": 2.3973,
+      "step": 21710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0977507547089587e-05,
+      "loss": 2.7101,
+      "step": 21720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0973255665632043e-05,
+      "loss": 2.1444,
+      "step": 21730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.09690037841745e-05,
+      "loss": 2.1706,
+      "step": 21740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0964751902716952e-05,
+      "loss": 1.6332,
+      "step": 21750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0960500021259408e-05,
+      "loss": 1.9248,
+      "step": 21760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0956248139801865e-05,
+      "loss": 2.0422,
+      "step": 21770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0951996258344317e-05,
+      "loss": 2.0232,
+      "step": 21780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0947744376886773e-05,
+      "loss": 2.0951,
+      "step": 21790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0943492495429228e-05,
+      "loss": 2.21,
+      "step": 21800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0939240613971684e-05,
+      "loss": 2.1676,
+      "step": 21810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0934988732514139e-05,
+      "loss": 2.6876,
+      "step": 21820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0930736851056593e-05,
+      "loss": 2.5208,
+      "step": 21830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0926484969599049e-05,
+      "loss": 2.076,
+      "step": 21840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0922233088141504e-05,
+      "loss": 2.2026,
+      "step": 21850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0917981206683958e-05,
+      "loss": 2.3495,
+      "step": 21860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0913729325226414e-05,
+      "loss": 2.1546,
+      "step": 21870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0909477443768867e-05,
+      "loss": 2.0528,
+      "step": 21880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0905225562311323e-05,
+      "loss": 2.0197,
+      "step": 21890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.090097368085378e-05,
+      "loss": 2.2806,
+      "step": 21900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0896721799396232e-05,
+      "loss": 1.7185,
+      "step": 21910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0892469917938688e-05,
+      "loss": 1.7569,
+      "step": 21920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0888218036481144e-05,
+      "loss": 2.1597,
+      "step": 21930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.08839661550236e-05,
+      "loss": 2.5093,
+      "step": 21940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0879714273566053e-05,
+      "loss": 2.3191,
+      "step": 21950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.087546239210851e-05,
+      "loss": 2.1572,
+      "step": 21960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0871210510650964e-05,
+      "loss": 2.4078,
+      "step": 21970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.0866958629193418e-05,
+      "loss": 2.315,
+      "step": 21980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0862706747735874e-05,
+      "loss": 2.5181,
+      "step": 21990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0858454866278329e-05,
+      "loss": 1.9297,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 1.8141272068023682,
+      "eval_runtime": 174.4359,
+      "eval_samples_per_second": 14.349,
+      "eval_steps_per_second": 7.177,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0854202984820783e-05,
+      "loss": 1.8454,
+      "step": 22010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.084995110336324e-05,
+      "loss": 2.1891,
+      "step": 22020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0845699221905694e-05,
+      "loss": 1.814,
+      "step": 22030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.084144734044815e-05,
+      "loss": 1.9533,
+      "step": 22040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0837195458990603e-05,
+      "loss": 2.4355,
+      "step": 22050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0832943577533059e-05,
+      "loss": 1.9416,
+      "step": 22060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0828691696075515e-05,
+      "loss": 2.0757,
+      "step": 22070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0824439814617968e-05,
+      "loss": 1.9605,
+      "step": 22080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0820187933160424e-05,
+      "loss": 2.1604,
+      "step": 22090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.081593605170288e-05,
+      "loss": 1.9458,
+      "step": 22100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0811684170245333e-05,
+      "loss": 2.3094,
+      "step": 22110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.080743228878779e-05,
+      "loss": 2.7009,
+      "step": 22120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0803180407330245e-05,
+      "loss": 2.1032,
+      "step": 22130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.07989285258727e-05,
+      "loss": 2.4193,
+      "step": 22140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0794676644415154e-05,
+      "loss": 1.9241,
+      "step": 22150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.079042476295761e-05,
+      "loss": 2.2416,
+      "step": 22160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0786172881500065e-05,
+      "loss": 2.2766,
+      "step": 22170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.078192100004252e-05,
+      "loss": 2.7925,
+      "step": 22180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0777669118584974e-05,
+      "loss": 2.0584,
+      "step": 22190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.077341723712743e-05,
+      "loss": 1.7813,
+      "step": 22200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0769165355669884e-05,
+      "loss": 1.6141,
+      "step": 22210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0764913474212339e-05,
+      "loss": 2.2464,
+      "step": 22220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0760661592754795e-05,
+      "loss": 2.0246,
+      "step": 22230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0756409711297251e-05,
+      "loss": 2.0217,
+      "step": 22240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0752157829839704e-05,
+      "loss": 2.8785,
+      "step": 22250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.074790594838216e-05,
+      "loss": 2.5433,
+      "step": 22260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0743654066924616e-05,
+      "loss": 2.0261,
+      "step": 22270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0739402185467069e-05,
+      "loss": 2.4837,
+      "step": 22280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0735150304009525e-05,
+      "loss": 2.1878,
+      "step": 22290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0730898422551981e-05,
+      "loss": 2.0769,
+      "step": 22300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0726646541094434e-05,
+      "loss": 1.976,
+      "step": 22310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.072239465963689e-05,
+      "loss": 2.3659,
+      "step": 22320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0718142778179346e-05,
+      "loss": 1.9599,
+      "step": 22330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0713890896721801e-05,
+      "loss": 1.8481,
+      "step": 22340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0709639015264255e-05,
+      "loss": 2.7933,
+      "step": 22350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.070538713380671e-05,
+      "loss": 2.5314,
+      "step": 22360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0701135252349166e-05,
+      "loss": 2.1009,
+      "step": 22370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.069688337089162e-05,
+      "loss": 2.3417,
+      "step": 22380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0692631489434075e-05,
+      "loss": 1.9509,
+      "step": 22390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0688379607976531e-05,
+      "loss": 2.1662,
+      "step": 22400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0684127726518985e-05,
+      "loss": 2.0892,
+      "step": 22410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.067987584506144e-05,
+      "loss": 2.1241,
+      "step": 22420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0675623963603896e-05,
+      "loss": 2.0205,
+      "step": 22430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0671372082146349e-05,
+      "loss": 1.8768,
+      "step": 22440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0667120200688805e-05,
+      "loss": 2.3859,
+      "step": 22450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0662868319231261e-05,
+      "loss": 2.0095,
+      "step": 22460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0658616437773717e-05,
+      "loss": 2.0586,
+      "step": 22470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.065436455631617e-05,
+      "loss": 2.232,
+      "step": 22480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0650112674858626e-05,
+      "loss": 2.09,
+      "step": 22490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0645860793401082e-05,
+      "loss": 2.412,
+      "step": 22500
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 1.8309388160705566,
+      "eval_runtime": 175.8105,
+      "eval_samples_per_second": 14.237,
+      "eval_steps_per_second": 7.121,
+      "step": 22500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0641608911943535e-05,
+      "loss": 2.3572,
+      "step": 22510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0637357030485991e-05,
+      "loss": 2.1156,
+      "step": 22520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0633105149028446e-05,
+      "loss": 2.0792,
+      "step": 22530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.06288532675709e-05,
+      "loss": 2.026,
+      "step": 22540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0624601386113356e-05,
+      "loss": 2.1075,
+      "step": 22550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.062034950465581e-05,
+      "loss": 2.1914,
+      "step": 22560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0616097623198267e-05,
+      "loss": 2.3427,
+      "step": 22570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0611845741740721e-05,
+      "loss": 2.2625,
+      "step": 22580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0607593860283176e-05,
+      "loss": 2.2144,
+      "step": 22590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0603341978825632e-05,
+      "loss": 2.6625,
+      "step": 22600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0599090097368085e-05,
+      "loss": 2.0088,
+      "step": 22610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0594838215910541e-05,
+      "loss": 2.1113,
+      "step": 22620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0590586334452997e-05,
+      "loss": 2.2633,
+      "step": 22630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.058633445299545e-05,
+      "loss": 2.4962,
+      "step": 22640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0582082571537906e-05,
+      "loss": 2.1041,
+      "step": 22650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0577830690080362e-05,
+      "loss": 2.0901,
+      "step": 22660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0573578808622818e-05,
+      "loss": 1.8119,
+      "step": 22670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0569326927165271e-05,
+      "loss": 2.5824,
+      "step": 22680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0565075045707727e-05,
+      "loss": 2.2369,
+      "step": 22690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0560823164250182e-05,
+      "loss": 1.9169,
+      "step": 22700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0556571282792636e-05,
+      "loss": 2.4513,
+      "step": 22710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0552319401335092e-05,
+      "loss": 2.5086,
+      "step": 22720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0548067519877547e-05,
+      "loss": 2.3318,
+      "step": 22730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0543815638420001e-05,
+      "loss": 2.0951,
+      "step": 22740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0539563756962456e-05,
+      "loss": 2.194,
+      "step": 22750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0535311875504912e-05,
+      "loss": 2.0352,
+      "step": 22760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0531059994047368e-05,
+      "loss": 2.3134,
+      "step": 22770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.052680811258982e-05,
+      "loss": 2.3509,
+      "step": 22780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0522556231132277e-05,
+      "loss": 2.2912,
+      "step": 22790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0518304349674733e-05,
+      "loss": 1.8441,
+      "step": 22800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0514052468217186e-05,
+      "loss": 2.019,
+      "step": 22810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0509800586759642e-05,
+      "loss": 1.9165,
+      "step": 22820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0505548705302098e-05,
+      "loss": 1.957,
+      "step": 22830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0501296823844551e-05,
+      "loss": 2.5048,
+      "step": 22840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0497044942387007e-05,
+      "loss": 2.2552,
+      "step": 22850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0492793060929463e-05,
+      "loss": 1.9776,
+      "step": 22860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0488541179471916e-05,
+      "loss": 2.4988,
+      "step": 22870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0484289298014372e-05,
+      "loss": 2.5497,
+      "step": 22880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0480037416556828e-05,
+      "loss": 2.0912,
+      "step": 22890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0475785535099283e-05,
+      "loss": 2.238,
+      "step": 22900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0471533653641737e-05,
+      "loss": 2.6684,
+      "step": 22910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0467281772184192e-05,
+      "loss": 1.9461,
+      "step": 22920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0463029890726648e-05,
+      "loss": 1.8546,
+      "step": 22930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0458778009269102e-05,
+      "loss": 2.3796,
+      "step": 22940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0454526127811557e-05,
+      "loss": 2.4635,
+      "step": 22950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0450274246354013e-05,
+      "loss": 2.6029,
+      "step": 22960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0446022364896467e-05,
+      "loss": 2.3095,
+      "step": 22970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0441770483438922e-05,
+      "loss": 2.0818,
+      "step": 22980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0437518601981378e-05,
+      "loss": 2.0784,
+      "step": 22990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0433266720523834e-05,
+      "loss": 2.4203,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 1.7879308462142944,
+      "eval_runtime": 175.9943,
+      "eval_samples_per_second": 14.222,
+      "eval_steps_per_second": 7.114,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0429014839066287e-05,
+      "loss": 2.2036,
+      "step": 23010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0424762957608743e-05,
+      "loss": 2.3537,
+      "step": 23020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0420511076151199e-05,
+      "loss": 2.2634,
+      "step": 23030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0416259194693652e-05,
+      "loss": 2.4369,
+      "step": 23040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0412007313236108e-05,
+      "loss": 2.0881,
+      "step": 23050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0407755431778564e-05,
+      "loss": 2.0322,
+      "step": 23060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0403503550321017e-05,
+      "loss": 2.1656,
+      "step": 23070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0399251668863473e-05,
+      "loss": 2.8787,
+      "step": 23080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0394999787405928e-05,
+      "loss": 2.1997,
+      "step": 23090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0390747905948384e-05,
+      "loss": 2.1858,
+      "step": 23100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0386496024490838e-05,
+      "loss": 2.5543,
+      "step": 23110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0382244143033293e-05,
+      "loss": 2.6911,
+      "step": 23120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0377992261575749e-05,
+      "loss": 1.7698,
+      "step": 23130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0373740380118202e-05,
+      "loss": 2.2497,
+      "step": 23140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0369488498660658e-05,
+      "loss": 2.3248,
+      "step": 23150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0365236617203114e-05,
+      "loss": 2.0568,
+      "step": 23160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0360984735745567e-05,
+      "loss": 2.2153,
+      "step": 23170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0356732854288023e-05,
+      "loss": 2.6964,
+      "step": 23180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0352480972830479e-05,
+      "loss": 2.2131,
+      "step": 23190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0348229091372935e-05,
+      "loss": 2.6161,
+      "step": 23200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0343977209915388e-05,
+      "loss": 2.4675,
+      "step": 23210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0339725328457844e-05,
+      "loss": 2.3536,
+      "step": 23220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0335473447000298e-05,
+      "loss": 2.7163,
+      "step": 23230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0331221565542753e-05,
+      "loss": 2.2453,
+      "step": 23240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0326969684085209e-05,
+      "loss": 2.085,
+      "step": 23250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0322717802627664e-05,
+      "loss": 2.059,
+      "step": 23260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0318465921170118e-05,
+      "loss": 2.5816,
+      "step": 23270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0314214039712574e-05,
+      "loss": 2.396,
+      "step": 23280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0309962158255029e-05,
+      "loss": 1.5413,
+      "step": 23290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0305710276797485e-05,
+      "loss": 2.3084,
+      "step": 23300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0301458395339938e-05,
+      "loss": 2.0634,
+      "step": 23310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0297206513882394e-05,
+      "loss": 2.2418,
+      "step": 23320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.029295463242485e-05,
+      "loss": 2.1639,
+      "step": 23330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0288702750967303e-05,
+      "loss": 2.8353,
+      "step": 23340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0284450869509759e-05,
+      "loss": 2.2332,
+      "step": 23350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0280198988052215e-05,
+      "loss": 2.1627,
+      "step": 23360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0275947106594668e-05,
+      "loss": 2.2034,
+      "step": 23370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0271695225137124e-05,
+      "loss": 2.5933,
+      "step": 23380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.026744334367958e-05,
+      "loss": 2.4328,
+      "step": 23390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0263191462222033e-05,
+      "loss": 2.6858,
+      "step": 23400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0258939580764489e-05,
+      "loss": 2.2375,
+      "step": 23410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0254687699306945e-05,
+      "loss": 1.9781,
+      "step": 23420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.02504358178494e-05,
+      "loss": 2.4411,
+      "step": 23430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0246183936391854e-05,
+      "loss": 2.1322,
+      "step": 23440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.024193205493431e-05,
+      "loss": 2.2164,
+      "step": 23450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0237680173476765e-05,
+      "loss": 2.2055,
+      "step": 23460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0233428292019219e-05,
+      "loss": 2.1434,
+      "step": 23470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0229176410561673e-05,
+      "loss": 1.8297,
+      "step": 23480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.022492452910413e-05,
+      "loss": 2.2009,
+      "step": 23490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0220672647646584e-05,
+      "loss": 1.8702,
+      "step": 23500
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 1.7928341627120972,
+      "eval_runtime": 175.9978,
+      "eval_samples_per_second": 14.222,
+      "eval_steps_per_second": 7.114,
+      "step": 23500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0216420766189039e-05,
+      "loss": 1.8704,
+      "step": 23510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0212168884731495e-05,
+      "loss": 2.3736,
+      "step": 23520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0207917003273951e-05,
+      "loss": 2.6456,
+      "step": 23530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0203665121816404e-05,
+      "loss": 2.1232,
+      "step": 23540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.019941324035886e-05,
+      "loss": 2.342,
+      "step": 23550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0195161358901316e-05,
+      "loss": 2.4895,
+      "step": 23560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0190909477443769e-05,
+      "loss": 2.0022,
+      "step": 23570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0186657595986225e-05,
+      "loss": 1.9872,
+      "step": 23580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0182405714528681e-05,
+      "loss": 2.2341,
+      "step": 23590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0178153833071134e-05,
+      "loss": 2.2823,
+      "step": 23600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.017390195161359e-05,
+      "loss": 1.7177,
+      "step": 23610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0169650070156044e-05,
+      "loss": 2.5706,
+      "step": 23620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.01653981886985e-05,
+      "loss": 2.6527,
+      "step": 23630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0161146307240955e-05,
+      "loss": 2.2877,
+      "step": 23640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.015689442578341e-05,
+      "loss": 2.2885,
+      "step": 23650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0152642544325866e-05,
+      "loss": 1.9024,
+      "step": 23660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.014839066286832e-05,
+      "loss": 2.1591,
+      "step": 23670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0144138781410775e-05,
+      "loss": 2.5355,
+      "step": 23680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.013988689995323e-05,
+      "loss": 2.4954,
+      "step": 23690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0135635018495683e-05,
+      "loss": 2.1401,
+      "step": 23700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.013138313703814e-05,
+      "loss": 2.05,
+      "step": 23710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0127131255580596e-05,
+      "loss": 2.4109,
+      "step": 23720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0122879374123052e-05,
+      "loss": 2.0181,
+      "step": 23730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0118627492665505e-05,
+      "loss": 2.3981,
+      "step": 23740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.011437561120796e-05,
+      "loss": 2.0846,
+      "step": 23750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0110123729750417e-05,
+      "loss": 2.0017,
+      "step": 23760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.010587184829287e-05,
+      "loss": 2.1176,
+      "step": 23770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0101619966835326e-05,
+      "loss": 2.0457,
+      "step": 23780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.009736808537778e-05,
+      "loss": 2.139,
+      "step": 23790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0093116203920235e-05,
+      "loss": 1.7997,
+      "step": 23800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0088864322462691e-05,
+      "loss": 2.1645,
+      "step": 23810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0084612441005145e-05,
+      "loss": 1.9419,
+      "step": 23820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0080360559547602e-05,
+      "loss": 2.497,
+      "step": 23830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0076108678090056e-05,
+      "loss": 2.4095,
+      "step": 23840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.007185679663251e-05,
+      "loss": 2.2074,
+      "step": 23850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0067604915174967e-05,
+      "loss": 2.2012,
+      "step": 23860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.006335303371742e-05,
+      "loss": 2.4029,
+      "step": 23870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0059101152259876e-05,
+      "loss": 2.0716,
+      "step": 23880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0054849270802332e-05,
+      "loss": 2.1061,
+      "step": 23890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0050597389344784e-05,
+      "loss": 2.175,
+      "step": 23900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.004634550788724e-05,
+      "loss": 2.2063,
+      "step": 23910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0042093626429697e-05,
+      "loss": 2.1733,
+      "step": 23920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.003784174497215e-05,
+      "loss": 2.1249,
+      "step": 23930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0033589863514606e-05,
+      "loss": 2.0847,
+      "step": 23940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0029337982057062e-05,
+      "loss": 1.8854,
+      "step": 23950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0025086100599516e-05,
+      "loss": 2.2463,
+      "step": 23960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.002083421914197e-05,
+      "loss": 1.8417,
+      "step": 23970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0016582337684427e-05,
+      "loss": 1.9418,
+      "step": 23980
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0012330456226881e-05,
+      "loss": 2.2803,
+      "step": 23990
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0008078574769336e-05,
+      "loss": 2.0295,
+      "step": 24000
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 1.7961026430130005,
+      "eval_runtime": 176.058,
+      "eval_samples_per_second": 14.217,
+      "eval_steps_per_second": 7.111,
+      "step": 24000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.000382669331179e-05,
+      "loss": 2.1783,
+      "step": 24010
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999574811854246e-06,
+      "loss": 1.8572,
+      "step": 24020
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.995322930396701e-06,
+      "loss": 2.477,
+      "step": 24030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.991071048939155e-06,
+      "loss": 2.0883,
+      "step": 24040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.986819167481612e-06,
+      "loss": 1.9533,
+      "step": 24050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.982567286024066e-06,
+      "loss": 1.791,
+      "step": 24060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.978315404566522e-06,
+      "loss": 2.0543,
+      "step": 24070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.974063523108977e-06,
+      "loss": 2.035,
+      "step": 24080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.969811641651431e-06,
+      "loss": 1.9221,
+      "step": 24090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.965559760193887e-06,
+      "loss": 2.1826,
+      "step": 24100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.961307878736342e-06,
+      "loss": 2.0506,
+      "step": 24110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.957055997278798e-06,
+      "loss": 1.9634,
+      "step": 24120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.952804115821252e-06,
+      "loss": 1.8171,
+      "step": 24130
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.948552234363707e-06,
+      "loss": 2.3548,
+      "step": 24140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.944300352906163e-06,
+      "loss": 2.1878,
+      "step": 24150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.940048471448617e-06,
+      "loss": 1.7794,
+      "step": 24160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.935796589991072e-06,
+      "loss": 2.1026,
+      "step": 24170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.931544708533526e-06,
+      "loss": 2.7091,
+      "step": 24180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.927292827075982e-06,
+      "loss": 2.2819,
+      "step": 24190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.923040945618437e-06,
+      "loss": 1.9126,
+      "step": 24200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.918789064160891e-06,
+      "loss": 2.0997,
+      "step": 24210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.914537182703346e-06,
+      "loss": 1.7042,
+      "step": 24220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.910285301245802e-06,
+      "loss": 1.9185,
+      "step": 24230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.906033419788256e-06,
+      "loss": 2.2732,
+      "step": 24240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.901781538330713e-06,
+      "loss": 2.1674,
+      "step": 24250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.897529656873167e-06,
+      "loss": 1.549,
+      "step": 24260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.893277775415621e-06,
+      "loss": 2.4823,
+      "step": 24270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.889025893958078e-06,
+      "loss": 2.7062,
+      "step": 24280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.884774012500532e-06,
+      "loss": 2.7915,
+      "step": 24290
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.880522131042988e-06,
+      "loss": 1.9146,
+      "step": 24300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.876270249585443e-06,
+      "loss": 1.7695,
+      "step": 24310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.872018368127897e-06,
+      "loss": 1.8908,
+      "step": 24320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.867766486670353e-06,
+      "loss": 2.155,
+      "step": 24330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.863514605212808e-06,
+      "loss": 2.1778,
+      "step": 24340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.859262723755262e-06,
+      "loss": 2.388,
+      "step": 24350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.855010842297718e-06,
+      "loss": 2.188,
+      "step": 24360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.850758960840173e-06,
+      "loss": 2.4158,
+      "step": 24370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.846507079382627e-06,
+      "loss": 2.0438,
+      "step": 24380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.842255197925082e-06,
+      "loss": 2.3838,
+      "step": 24390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.838003316467538e-06,
+      "loss": 1.9735,
+      "step": 24400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.833751435009992e-06,
+      "loss": 1.5923,
+      "step": 24410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.829499553552447e-06,
+      "loss": 2.4302,
+      "step": 24420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.825247672094903e-06,
+      "loss": 2.009,
+      "step": 24430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.820995790637357e-06,
+      "loss": 2.1147,
+      "step": 24440
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.816743909179814e-06,
+      "loss": 2.5158,
+      "step": 24450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.812492027722268e-06,
+      "loss": 1.8816,
+      "step": 24460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.808240146264722e-06,
+      "loss": 1.9242,
+      "step": 24470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.803988264807179e-06,
+      "loss": 2.4337,
+      "step": 24480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.799736383349633e-06,
+      "loss": 2.4206,
+      "step": 24490
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.79548450189209e-06,
+      "loss": 2.4817,
+      "step": 24500
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 1.7808321714401245,
+      "eval_runtime": 175.8437,
+      "eval_samples_per_second": 14.234,
+      "eval_steps_per_second": 7.12,
+      "step": 24500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.791232620434544e-06,
+      "loss": 2.0811,
+      "step": 24510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.786980738976998e-06,
+      "loss": 2.0437,
+      "step": 24520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.782728857519453e-06,
+      "loss": 2.5538,
+      "step": 24530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.778476976061909e-06,
+      "loss": 2.0296,
+      "step": 24540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.774225094604363e-06,
+      "loss": 2.3742,
+      "step": 24550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.769973213146818e-06,
+      "loss": 1.8981,
+      "step": 24560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.765721331689272e-06,
+      "loss": 1.7186,
+      "step": 24570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.761469450231728e-06,
+      "loss": 2.207,
+      "step": 24580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.757217568774183e-06,
+      "loss": 2.148,
+      "step": 24590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.752965687316639e-06,
+      "loss": 2.3233,
+      "step": 24600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.748713805859093e-06,
+      "loss": 2.3885,
+      "step": 24610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.744461924401548e-06,
+      "loss": 2.1731,
+      "step": 24620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.740210042944004e-06,
+      "loss": 2.1489,
+      "step": 24630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.735958161486458e-06,
+      "loss": 2.2787,
+      "step": 24640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.731706280028913e-06,
+      "loss": 2.5155,
+      "step": 24650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.727454398571369e-06,
+      "loss": 2.5187,
+      "step": 24660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.723202517113823e-06,
+      "loss": 1.9613,
+      "step": 24670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.71895063565628e-06,
+      "loss": 2.1069,
+      "step": 24680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.714698754198734e-06,
+      "loss": 1.8688,
+      "step": 24690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.710446872741189e-06,
+      "loss": 2.0668,
+      "step": 24700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.706194991283645e-06,
+      "loss": 3.0117,
+      "step": 24710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.7019431098261e-06,
+      "loss": 1.9196,
+      "step": 24720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.697691228368554e-06,
+      "loss": 2.1805,
+      "step": 24730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.693439346911008e-06,
+      "loss": 2.3032,
+      "step": 24740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.689187465453464e-06,
+      "loss": 2.1187,
+      "step": 24750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.684935583995919e-06,
+      "loss": 2.5357,
+      "step": 24760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.680683702538373e-06,
+      "loss": 1.7142,
+      "step": 24770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.67643182108083e-06,
+      "loss": 2.1176,
+      "step": 24780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.672179939623284e-06,
+      "loss": 2.0409,
+      "step": 24790
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.667928058165738e-06,
+      "loss": 2.0488,
+      "step": 24800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.663676176708194e-06,
+      "loss": 2.1542,
+      "step": 24810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.659424295250649e-06,
+      "loss": 2.1647,
+      "step": 24820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.655172413793105e-06,
+      "loss": 2.4398,
+      "step": 24830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.65092053233556e-06,
+      "loss": 1.9817,
+      "step": 24840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.646668650878014e-06,
+      "loss": 2.2528,
+      "step": 24850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.64241676942047e-06,
+      "loss": 2.4099,
+      "step": 24860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.638164887962925e-06,
+      "loss": 2.1171,
+      "step": 24870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.63391300650538e-06,
+      "loss": 2.5138,
+      "step": 24880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.629661125047835e-06,
+      "loss": 2.2478,
+      "step": 24890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.62540924359029e-06,
+      "loss": 2.1804,
+      "step": 24900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.621157362132744e-06,
+      "loss": 2.1283,
+      "step": 24910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.616905480675199e-06,
+      "loss": 2.4359,
+      "step": 24920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.612653599217655e-06,
+      "loss": 2.3146,
+      "step": 24930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.608401717760109e-06,
+      "loss": 2.1418,
+      "step": 24940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.604149836302564e-06,
+      "loss": 2.1131,
+      "step": 24950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.59989795484502e-06,
+      "loss": 1.7208,
+      "step": 24960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.595646073387474e-06,
+      "loss": 1.9904,
+      "step": 24970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.59139419192993e-06,
+      "loss": 1.8274,
+      "step": 24980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.587142310472385e-06,
+      "loss": 2.1403,
+      "step": 24990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.58289042901484e-06,
+      "loss": 2.1498,
+      "step": 25000
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 1.8270052671432495,
+      "eval_runtime": 175.899,
+      "eval_samples_per_second": 14.23,
+      "eval_steps_per_second": 7.118,
+      "step": 25000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.578638547557295e-06,
+      "loss": 2.5489,
+      "step": 25010
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.57438666609975e-06,
+      "loss": 2.049,
+      "step": 25020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.570134784642206e-06,
+      "loss": 2.0372,
+      "step": 25030
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.56588290318466e-06,
+      "loss": 1.9352,
+      "step": 25040
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.561631021727115e-06,
+      "loss": 2.8194,
+      "step": 25050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.557379140269571e-06,
+      "loss": 2.0632,
+      "step": 25060
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.553127258812026e-06,
+      "loss": 2.1373,
+      "step": 25070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.54887537735448e-06,
+      "loss": 1.5917,
+      "step": 25080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.544623495896934e-06,
+      "loss": 1.9999,
+      "step": 25090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.54037161443939e-06,
+      "loss": 2.4615,
+      "step": 25100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.536119732981845e-06,
+      "loss": 1.9341,
+      "step": 25110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.5318678515243e-06,
+      "loss": 2.0574,
+      "step": 25120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.527615970066756e-06,
+      "loss": 2.2517,
+      "step": 25130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.52336408860921e-06,
+      "loss": 2.8717,
+      "step": 25140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.519112207151665e-06,
+      "loss": 1.9216,
+      "step": 25150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.51486032569412e-06,
+      "loss": 1.7792,
+      "step": 25160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.510608444236575e-06,
+      "loss": 2.2475,
+      "step": 25170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.50635656277903e-06,
+      "loss": 1.8762,
+      "step": 25180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.502104681321486e-06,
+      "loss": 2.123,
+      "step": 25190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.49785279986394e-06,
+      "loss": 2.3417,
+      "step": 25200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.493600918406396e-06,
+      "loss": 1.6224,
+      "step": 25210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.489349036948851e-06,
+      "loss": 2.2181,
+      "step": 25220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.485097155491305e-06,
+      "loss": 2.2656,
+      "step": 25230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.480845274033762e-06,
+      "loss": 2.4773,
+      "step": 25240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.476593392576216e-06,
+      "loss": 5.3294,
+      "step": 25250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.47234151111867e-06,
+      "loss": 2.719,
+      "step": 25260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.468089629661127e-06,
+      "loss": 2.0256,
+      "step": 25270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.463837748203581e-06,
+      "loss": 2.7081,
+      "step": 25280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.459585866746035e-06,
+      "loss": 2.7167,
+      "step": 25290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.45533398528849e-06,
+      "loss": 2.3192,
+      "step": 25300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.451082103830946e-06,
+      "loss": 2.0319,
+      "step": 25310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.4468302223734e-06,
+      "loss": 2.1389,
+      "step": 25320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.442578340915855e-06,
+      "loss": 2.1409,
+      "step": 25330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.438326459458311e-06,
+      "loss": 2.0399,
+      "step": 25340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.434074578000766e-06,
+      "loss": 1.9943,
+      "step": 25350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.429822696543222e-06,
+      "loss": 1.9661,
+      "step": 25360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.425570815085676e-06,
+      "loss": 2.0869,
+      "step": 25370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.42131893362813e-06,
+      "loss": 2.0276,
+      "step": 25380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.417067052170587e-06,
+      "loss": 2.4021,
+      "step": 25390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.412815170713041e-06,
+      "loss": 2.0104,
+      "step": 25400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.408563289255497e-06,
+      "loss": 2.2962,
+      "step": 25410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.404311407797952e-06,
+      "loss": 2.1656,
+      "step": 25420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.400059526340406e-06,
+      "loss": 2.1771,
+      "step": 25430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.39580764488286e-06,
+      "loss": 2.49,
+      "step": 25440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.391555763425317e-06,
+      "loss": 2.2185,
+      "step": 25450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.387303881967771e-06,
+      "loss": 2.1741,
+      "step": 25460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.383052000510226e-06,
+      "loss": 1.9142,
+      "step": 25470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.37880011905268e-06,
+      "loss": 2.1593,
+      "step": 25480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.374548237595137e-06,
+      "loss": 2.6488,
+      "step": 25490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.370296356137591e-06,
+      "loss": 1.5643,
+      "step": 25500
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 1.8130146265029907,
+      "eval_runtime": 174.1071,
+      "eval_samples_per_second": 14.376,
+      "eval_steps_per_second": 7.191,
+      "step": 25500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.366044474680047e-06,
+      "loss": 2.4988,
+      "step": 25510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.361792593222502e-06,
+      "loss": 1.8701,
+      "step": 25520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.357540711764956e-06,
+      "loss": 1.9622,
+      "step": 25530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.353288830307412e-06,
+      "loss": 2.2419,
+      "step": 25540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.349036948849867e-06,
+      "loss": 2.479,
+      "step": 25550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.344785067392323e-06,
+      "loss": 2.3276,
+      "step": 25560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.340533185934777e-06,
+      "loss": 1.6181,
+      "step": 25570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.336281304477232e-06,
+      "loss": 2.2207,
+      "step": 25580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.332029423019688e-06,
+      "loss": 2.4773,
+      "step": 25590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.327777541562142e-06,
+      "loss": 2.1505,
+      "step": 25600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.323525660104597e-06,
+      "loss": 1.9943,
+      "step": 25610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.319273778647053e-06,
+      "loss": 2.2931,
+      "step": 25620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.315021897189507e-06,
+      "loss": 1.637,
+      "step": 25630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.310770015731962e-06,
+      "loss": 1.9188,
+      "step": 25640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.306518134274416e-06,
+      "loss": 1.9153,
+      "step": 25650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.302266252816872e-06,
+      "loss": 2.1277,
+      "step": 25660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.298014371359327e-06,
+      "loss": 2.6407,
+      "step": 25670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.293762489901781e-06,
+      "loss": 2.0261,
+      "step": 25680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.289510608444238e-06,
+      "loss": 1.9497,
+      "step": 25690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.285258726986692e-06,
+      "loss": 2.6189,
+      "step": 25700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.281006845529146e-06,
+      "loss": 1.9009,
+      "step": 25710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.276754964071603e-06,
+      "loss": 2.254,
+      "step": 25720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.272503082614057e-06,
+      "loss": 2.2595,
+      "step": 25730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.268251201156513e-06,
+      "loss": 2.3933,
+      "step": 25740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.263999319698968e-06,
+      "loss": 2.318,
+      "step": 25750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.259747438241422e-06,
+      "loss": 2.0637,
+      "step": 25760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.255495556783878e-06,
+      "loss": 2.4044,
+      "step": 25770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.251243675326333e-06,
+      "loss": 2.54,
+      "step": 25780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.246991793868789e-06,
+      "loss": 2.075,
+      "step": 25790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.242739912411243e-06,
+      "loss": 2.491,
+      "step": 25800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.238488030953698e-06,
+      "loss": 1.9432,
+      "step": 25810
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.234236149496152e-06,
+      "loss": 2.0367,
+      "step": 25820
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.229984268038607e-06,
+      "loss": 2.1673,
+      "step": 25830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.225732386581063e-06,
+      "loss": 2.1345,
+      "step": 25840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.221480505123517e-06,
+      "loss": 2.0569,
+      "step": 25850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.217228623665972e-06,
+      "loss": 1.986,
+      "step": 25860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.212976742208428e-06,
+      "loss": 2.3173,
+      "step": 25870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.208724860750882e-06,
+      "loss": 2.5024,
+      "step": 25880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.204472979293339e-06,
+      "loss": 2.3327,
+      "step": 25890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.200221097835793e-06,
+      "loss": 1.8746,
+      "step": 25900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.195969216378247e-06,
+      "loss": 2.4627,
+      "step": 25910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.191717334920704e-06,
+      "loss": 1.8956,
+      "step": 25920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.187465453463158e-06,
+      "loss": 1.9613,
+      "step": 25930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.183213572005614e-06,
+      "loss": 1.7535,
+      "step": 25940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.178961690548069e-06,
+      "loss": 2.539,
+      "step": 25950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.174709809090523e-06,
+      "loss": 1.6916,
+      "step": 25960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.17045792763298e-06,
+      "loss": 2.0812,
+      "step": 25970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.166206046175434e-06,
+      "loss": 2.3501,
+      "step": 25980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.161954164717888e-06,
+      "loss": 2.1016,
+      "step": 25990
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.157702283260343e-06,
+      "loss": 2.1681,
+      "step": 26000
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 1.8029661178588867,
+      "eval_runtime": 175.9282,
+      "eval_samples_per_second": 14.227,
+      "eval_steps_per_second": 7.117,
+      "step": 26000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.153450401802799e-06,
+      "loss": 2.029,
+      "step": 26010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.149198520345253e-06,
+      "loss": 2.1774,
+      "step": 26020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.144946638887708e-06,
+      "loss": 2.2816,
+      "step": 26030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.140694757430164e-06,
+      "loss": 2.2293,
+      "step": 26040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.136442875972618e-06,
+      "loss": 2.6347,
+      "step": 26050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.132190994515073e-06,
+      "loss": 2.0672,
+      "step": 26060
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.127939113057529e-06,
+      "loss": 2.284,
+      "step": 26070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.123687231599983e-06,
+      "loss": 2.3167,
+      "step": 26080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.11943535014244e-06,
+      "loss": 1.5823,
+      "step": 26090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.115183468684894e-06,
+      "loss": 2.4238,
+      "step": 26100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.110931587227349e-06,
+      "loss": 2.1433,
+      "step": 26110
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.106679705769805e-06,
+      "loss": 1.8668,
+      "step": 26120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.102427824312259e-06,
+      "loss": 2.5338,
+      "step": 26130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.098175942854715e-06,
+      "loss": 1.9807,
+      "step": 26140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.09392406139717e-06,
+      "loss": 2.1115,
+      "step": 26150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.089672179939624e-06,
+      "loss": 1.9726,
+      "step": 26160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.085420298482079e-06,
+      "loss": 2.0894,
+      "step": 26170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.081168417024535e-06,
+      "loss": 2.2522,
+      "step": 26180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.07691653556699e-06,
+      "loss": 2.2052,
+      "step": 26190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.072664654109444e-06,
+      "loss": 2.1807,
+      "step": 26200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.068412772651898e-06,
+      "loss": 1.9953,
+      "step": 26210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.064160891194354e-06,
+      "loss": 2.5388,
+      "step": 26220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.059909009736809e-06,
+      "loss": 2.431,
+      "step": 26230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.055657128279263e-06,
+      "loss": 2.4511,
+      "step": 26240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.05140524682172e-06,
+      "loss": 2.5282,
+      "step": 26250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.047153365364174e-06,
+      "loss": 2.3258,
+      "step": 26260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.04290148390663e-06,
+      "loss": 2.2077,
+      "step": 26270
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.038649602449084e-06,
+      "loss": 2.1033,
+      "step": 26280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.034397720991539e-06,
+      "loss": 2.064,
+      "step": 26290
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.030145839533995e-06,
+      "loss": 2.1307,
+      "step": 26300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.02589395807645e-06,
+      "loss": 2.1043,
+      "step": 26310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.021642076618906e-06,
+      "loss": 1.842,
+      "step": 26320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.01739019516136e-06,
+      "loss": 1.8477,
+      "step": 26330
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.013138313703815e-06,
+      "loss": 1.968,
+      "step": 26340
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.008886432246269e-06,
+      "loss": 2.2406,
+      "step": 26350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.004634550788725e-06,
+      "loss": 2.3495,
+      "step": 26360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.00038266933118e-06,
+      "loss": 2.318,
+      "step": 26370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.996130787873634e-06,
+      "loss": 1.9728,
+      "step": 26380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.991878906416089e-06,
+      "loss": 2.7301,
+      "step": 26390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.987627024958545e-06,
+      "loss": 2.3417,
+      "step": 26400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.983375143501e-06,
+      "loss": 2.547,
+      "step": 26410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.979123262043455e-06,
+      "loss": 2.2439,
+      "step": 26420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.97487138058591e-06,
+      "loss": 2.1761,
+      "step": 26430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.970619499128364e-06,
+      "loss": 1.8553,
+      "step": 26440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.96636761767082e-06,
+      "loss": 2.4548,
+      "step": 26450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.962115736213275e-06,
+      "loss": 1.7375,
+      "step": 26460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.957863854755731e-06,
+      "loss": 2.3438,
+      "step": 26470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.953611973298186e-06,
+      "loss": 2.4082,
+      "step": 26480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.94936009184064e-06,
+      "loss": 2.2813,
+      "step": 26490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.945108210383096e-06,
+      "loss": 2.1074,
+      "step": 26500
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 1.7794862985610962,
+      "eval_runtime": 175.953,
+      "eval_samples_per_second": 14.225,
+      "eval_steps_per_second": 7.116,
+      "step": 26500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.94085632892555e-06,
+      "loss": 2.3367,
+      "step": 26510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.936604447468005e-06,
+      "loss": 2.0808,
+      "step": 26520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.932352566010461e-06,
+      "loss": 2.0238,
+      "step": 26530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.928100684552916e-06,
+      "loss": 1.9315,
+      "step": 26540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.92384880309537e-06,
+      "loss": 2.1414,
+      "step": 26550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.919596921637825e-06,
+      "loss": 2.2046,
+      "step": 26560
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.91534504018028e-06,
+      "loss": 1.559,
+      "step": 26570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.911093158722735e-06,
+      "loss": 2.3783,
+      "step": 26580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.90684127726519e-06,
+      "loss": 2.259,
+      "step": 26590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.902589395807646e-06,
+      "loss": 2.2287,
+      "step": 26600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.8983375143501e-06,
+      "loss": 2.2234,
+      "step": 26610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.894085632892556e-06,
+      "loss": 2.3311,
+      "step": 26620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.889833751435011e-06,
+      "loss": 2.4397,
+      "step": 26630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.885581869977465e-06,
+      "loss": 2.0832,
+      "step": 26640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.881329988519921e-06,
+      "loss": 1.9653,
+      "step": 26650
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.877078107062376e-06,
+      "loss": 1.7654,
+      "step": 26660
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.87282622560483e-06,
+      "loss": 2.2557,
+      "step": 26670
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.868574344147287e-06,
+      "loss": 2.0998,
+      "step": 26680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.864322462689741e-06,
+      "loss": 1.5496,
+      "step": 26690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.860070581232197e-06,
+      "loss": 2.0313,
+      "step": 26700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.855818699774652e-06,
+      "loss": 2.3958,
+      "step": 26710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.851566818317106e-06,
+      "loss": 2.2315,
+      "step": 26720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.84731493685956e-06,
+      "loss": 2.4052,
+      "step": 26730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.843063055402015e-06,
+      "loss": 2.1748,
+      "step": 26740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.838811173944471e-06,
+      "loss": 2.2979,
+      "step": 26750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.834559292486926e-06,
+      "loss": 2.0609,
+      "step": 26760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.83030741102938e-06,
+      "loss": 2.4294,
+      "step": 26770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.826055529571836e-06,
+      "loss": 2.4692,
+      "step": 26780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.82180364811429e-06,
+      "loss": 2.3065,
+      "step": 26790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.817551766656747e-06,
+      "loss": 1.8414,
+      "step": 26800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.813299885199201e-06,
+      "loss": 1.9798,
+      "step": 26810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.809048003741656e-06,
+      "loss": 2.6128,
+      "step": 26820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.804796122284112e-06,
+      "loss": 2.4113,
+      "step": 26830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.800544240826566e-06,
+      "loss": 2.3281,
+      "step": 26840
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.796292359369022e-06,
+      "loss": 2.0856,
+      "step": 26850
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.792040477911477e-06,
+      "loss": 1.9102,
+      "step": 26860
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.787788596453931e-06,
+      "loss": 1.8436,
+      "step": 26870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.783536714996388e-06,
+      "loss": 2.187,
+      "step": 26880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.779284833538842e-06,
+      "loss": 2.4518,
+      "step": 26890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.775032952081296e-06,
+      "loss": 2.7198,
+      "step": 26900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.770781070623751e-06,
+      "loss": 2.3112,
+      "step": 26910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.766529189166207e-06,
+      "loss": 2.0266,
+      "step": 26920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.762277307708662e-06,
+      "loss": 2.4087,
+      "step": 26930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.758025426251116e-06,
+      "loss": 1.793,
+      "step": 26940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.753773544793572e-06,
+      "loss": 1.9159,
+      "step": 26950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.749521663336027e-06,
+      "loss": 2.0469,
+      "step": 26960
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.745269781878481e-06,
+      "loss": 2.1949,
+      "step": 26970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.741017900420937e-06,
+      "loss": 2.1446,
+      "step": 26980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.736766018963392e-06,
+      "loss": 2.1481,
+      "step": 26990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.732514137505848e-06,
+      "loss": 2.3402,
+      "step": 27000
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 1.7784305810928345,
+      "eval_runtime": 175.9346,
+      "eval_samples_per_second": 14.227,
+      "eval_steps_per_second": 7.116,
+      "step": 27000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.728262256048302e-06,
+      "loss": 1.9918,
+      "step": 27010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.724010374590757e-06,
+      "loss": 1.9544,
+      "step": 27020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.719758493133213e-06,
+      "loss": 2.1023,
+      "step": 27030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.715506611675667e-06,
+      "loss": 2.3112,
+      "step": 27040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.711254730218124e-06,
+      "loss": 2.2315,
+      "step": 27050
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.707002848760578e-06,
+      "loss": 2.2551,
+      "step": 27060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.702750967303032e-06,
+      "loss": 1.783,
+      "step": 27070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.698499085845487e-06,
+      "loss": 2.2734,
+      "step": 27080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.694247204387943e-06,
+      "loss": 2.235,
+      "step": 27090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.689995322930397e-06,
+      "loss": 2.0812,
+      "step": 27100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.685743441472852e-06,
+      "loss": 2.3111,
+      "step": 27110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.681491560015306e-06,
+      "loss": 1.9091,
+      "step": 27120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.677239678557763e-06,
+      "loss": 2.1717,
+      "step": 27130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.672987797100217e-06,
+      "loss": 1.6394,
+      "step": 27140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.668735915642673e-06,
+      "loss": 2.1665,
+      "step": 27150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.664484034185128e-06,
+      "loss": 2.6547,
+      "step": 27160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.660232152727582e-06,
+      "loss": 2.5331,
+      "step": 27170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.655980271270038e-06,
+      "loss": 2.1644,
+      "step": 27180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.651728389812493e-06,
+      "loss": 2.5835,
+      "step": 27190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.647476508354947e-06,
+      "loss": 1.8968,
+      "step": 27200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.643224626897403e-06,
+      "loss": 2.1733,
+      "step": 27210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.638972745439858e-06,
+      "loss": 2.2962,
+      "step": 27220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.634720863982314e-06,
+      "loss": 1.7098,
+      "step": 27230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.630468982524768e-06,
+      "loss": 2.4761,
+      "step": 27240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.626217101067223e-06,
+      "loss": 1.9359,
+      "step": 27250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.621965219609677e-06,
+      "loss": 1.9284,
+      "step": 27260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.617713338152133e-06,
+      "loss": 1.934,
+      "step": 27270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.613461456694588e-06,
+      "loss": 2.2519,
+      "step": 27280
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.609209575237042e-06,
+      "loss": 2.2796,
+      "step": 27290
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.604957693779497e-06,
+      "loss": 2.0495,
+      "step": 27300
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.600705812321953e-06,
+      "loss": 1.8942,
+      "step": 27310
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.596453930864407e-06,
+      "loss": 1.9881,
+      "step": 27320
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.592202049406864e-06,
+      "loss": 2.444,
+      "step": 27330
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.587950167949318e-06,
+      "loss": 2.2539,
+      "step": 27340
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.583698286491773e-06,
+      "loss": 1.6349,
+      "step": 27350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.579446405034229e-06,
+      "loss": 1.8034,
+      "step": 27360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.575194523576683e-06,
+      "loss": 1.714,
+      "step": 27370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.57094264211914e-06,
+      "loss": 1.8281,
+      "step": 27380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.566690760661594e-06,
+      "loss": 1.9654,
+      "step": 27390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.562438879204048e-06,
+      "loss": 1.8929,
+      "step": 27400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.558186997746504e-06,
+      "loss": 1.9911,
+      "step": 27410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.553935116288959e-06,
+      "loss": 2.025,
+      "step": 27420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.549683234831413e-06,
+      "loss": 2.3029,
+      "step": 27430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.54543135337387e-06,
+      "loss": 2.0765,
+      "step": 27440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.541179471916324e-06,
+      "loss": 2.17,
+      "step": 27450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.536927590458778e-06,
+      "loss": 2.0451,
+      "step": 27460
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.532675709001233e-06,
+      "loss": 2.2552,
+      "step": 27470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.528423827543689e-06,
+      "loss": 2.0923,
+      "step": 27480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.524171946086143e-06,
+      "loss": 2.2716,
+      "step": 27490
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.519920064628598e-06,
+      "loss": 1.8374,
+      "step": 27500
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 1.7960634231567383,
+      "eval_runtime": 175.5325,
+      "eval_samples_per_second": 14.259,
+      "eval_steps_per_second": 7.133,
+      "step": 27500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.515668183171054e-06,
+      "loss": 1.9933,
+      "step": 27510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.511416301713508e-06,
+      "loss": 2.3631,
+      "step": 27520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.507164420255965e-06,
+      "loss": 2.209,
+      "step": 27530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.502912538798419e-06,
+      "loss": 2.4285,
+      "step": 27540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.498660657340874e-06,
+      "loss": 2.2684,
+      "step": 27550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.49440877588333e-06,
+      "loss": 2.0219,
+      "step": 27560
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.490156894425784e-06,
+      "loss": 1.8807,
+      "step": 27570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.48590501296824e-06,
+      "loss": 2.6944,
+      "step": 27580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.481653131510695e-06,
+      "loss": 2.3637,
+      "step": 27590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.47740125005315e-06,
+      "loss": 2.2071,
+      "step": 27600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.473149368595605e-06,
+      "loss": 2.0359,
+      "step": 27610
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.46889748713806e-06,
+      "loss": 2.1155,
+      "step": 27620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.464645605680514e-06,
+      "loss": 2.3826,
+      "step": 27630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.460393724222969e-06,
+      "loss": 1.8784,
+      "step": 27640
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.456141842765423e-06,
+      "loss": 2.3723,
+      "step": 27650
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.45188996130788e-06,
+      "loss": 1.7547,
+      "step": 27660
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.447638079850334e-06,
+      "loss": 1.9922,
+      "step": 27670
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.443386198392788e-06,
+      "loss": 2.2056,
+      "step": 27680
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.439134316935244e-06,
+      "loss": 2.4756,
+      "step": 27690
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.434882435477699e-06,
+      "loss": 1.7505,
+      "step": 27700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.430630554020155e-06,
+      "loss": 2.1735,
+      "step": 27710
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.42637867256261e-06,
+      "loss": 2.3311,
+      "step": 27720
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.422126791105064e-06,
+      "loss": 2.1869,
+      "step": 27730
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.41787490964752e-06,
+      "loss": 1.7889,
+      "step": 27740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.413623028189975e-06,
+      "loss": 2.1902,
+      "step": 27750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.40937114673243e-06,
+      "loss": 2.3086,
+      "step": 27760
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.405119265274885e-06,
+      "loss": 2.6292,
+      "step": 27770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.40086738381734e-06,
+      "loss": 2.3843,
+      "step": 27780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.396615502359796e-06,
+      "loss": 1.9254,
+      "step": 27790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.39236362090225e-06,
+      "loss": 1.703,
+      "step": 27800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.388111739444705e-06,
+      "loss": 1.6527,
+      "step": 27810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.38385985798716e-06,
+      "loss": 2.2103,
+      "step": 27820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.379607976529615e-06,
+      "loss": 2.1751,
+      "step": 27830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.37535609507207e-06,
+      "loss": 2.2004,
+      "step": 27840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.371104213614524e-06,
+      "loss": 1.9038,
+      "step": 27850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.36685233215698e-06,
+      "loss": 2.2364,
+      "step": 27860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.362600450699435e-06,
+      "loss": 1.9736,
+      "step": 27870
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.35834856924189e-06,
+      "loss": 1.9249,
+      "step": 27880
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.354096687784345e-06,
+      "loss": 1.7786,
+      "step": 27890
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.3498448063268e-06,
+      "loss": 2.1232,
+      "step": 27900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.345592924869256e-06,
+      "loss": 2.3536,
+      "step": 27910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.34134104341171e-06,
+      "loss": 2.2933,
+      "step": 27920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.337089161954165e-06,
+      "loss": 1.7072,
+      "step": 27930
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.332837280496621e-06,
+      "loss": 2.2741,
+      "step": 27940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.328585399039076e-06,
+      "loss": 2.356,
+      "step": 27950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.324333517581532e-06,
+      "loss": 2.1711,
+      "step": 27960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.320081636123986e-06,
+      "loss": 1.9022,
+      "step": 27970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.31582975466644e-06,
+      "loss": 2.5043,
+      "step": 27980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.311577873208895e-06,
+      "loss": 2.411,
+      "step": 27990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.307325991751351e-06,
+      "loss": 1.9705,
+      "step": 28000
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 1.7835320234298706,
+      "eval_runtime": 175.7479,
+      "eval_samples_per_second": 14.242,
+      "eval_steps_per_second": 7.124,
+      "step": 28000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.303074110293806e-06,
+      "loss": 2.3563,
+      "step": 28010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.29882222883626e-06,
+      "loss": 2.5579,
+      "step": 28020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.294570347378715e-06,
+      "loss": 2.2202,
+      "step": 28030
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.29031846592117e-06,
+      "loss": 2.2095,
+      "step": 28040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.286066584463625e-06,
+      "loss": 2.0755,
+      "step": 28050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.281814703006081e-06,
+      "loss": 2.3718,
+      "step": 28060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.277562821548536e-06,
+      "loss": 2.4411,
+      "step": 28070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.27331094009099e-06,
+      "loss": 2.5163,
+      "step": 28080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.269059058633446e-06,
+      "loss": 2.1317,
+      "step": 28090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.264807177175901e-06,
+      "loss": 2.2966,
+      "step": 28100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.260555295718357e-06,
+      "loss": 2.1478,
+      "step": 28110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.256303414260812e-06,
+      "loss": 2.1012,
+      "step": 28120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.252051532803266e-06,
+      "loss": 1.8812,
+      "step": 28130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.247799651345722e-06,
+      "loss": 1.7168,
+      "step": 28140
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.243547769888177e-06,
+      "loss": 2.199,
+      "step": 28150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.239295888430631e-06,
+      "loss": 2.252,
+      "step": 28160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.235044006973086e-06,
+      "loss": 1.89,
+      "step": 28170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.230792125515542e-06,
+      "loss": 2.1628,
+      "step": 28180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.226540244057996e-06,
+      "loss": 2.1145,
+      "step": 28190
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.22228836260045e-06,
+      "loss": 2.167,
+      "step": 28200
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.218036481142905e-06,
+      "loss": 2.5227,
+      "step": 28210
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.213784599685361e-06,
+      "loss": 2.1437,
+      "step": 28220
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.209532718227816e-06,
+      "loss": 2.0659,
+      "step": 28230
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.205280836770272e-06,
+      "loss": 2.4093,
+      "step": 28240
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.201028955312726e-06,
+      "loss": 2.0918,
+      "step": 28250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.19677707385518e-06,
+      "loss": 2.1433,
+      "step": 28260
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.192525192397637e-06,
+      "loss": 2.0139,
+      "step": 28270
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.188273310940091e-06,
+      "loss": 2.1022,
+      "step": 28280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.184021429482548e-06,
+      "loss": 2.0868,
+      "step": 28290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.179769548025002e-06,
+      "loss": 1.7078,
+      "step": 28300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.175517666567456e-06,
+      "loss": 1.9889,
+      "step": 28310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.171265785109913e-06,
+      "loss": 2.3366,
+      "step": 28320
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.167013903652367e-06,
+      "loss": 2.2599,
+      "step": 28330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.162762022194821e-06,
+      "loss": 2.2115,
+      "step": 28340
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.158510140737278e-06,
+      "loss": 1.867,
+      "step": 28350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.154258259279732e-06,
+      "loss": 1.7084,
+      "step": 28360
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.150006377822187e-06,
+      "loss": 2.4791,
+      "step": 28370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.145754496364641e-06,
+      "loss": 1.8911,
+      "step": 28380
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.141502614907097e-06,
+      "loss": 2.0083,
+      "step": 28390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.137250733449552e-06,
+      "loss": 2.436,
+      "step": 28400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.132998851992006e-06,
+      "loss": 2.5692,
+      "step": 28410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.128746970534462e-06,
+      "loss": 1.878,
+      "step": 28420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.124495089076917e-06,
+      "loss": 1.6927,
+      "step": 28430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.120243207619373e-06,
+      "loss": 2.2977,
+      "step": 28440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.115991326161827e-06,
+      "loss": 2.2699,
+      "step": 28450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.111739444704282e-06,
+      "loss": 1.844,
+      "step": 28460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.107487563246738e-06,
+      "loss": 2.162,
+      "step": 28470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.103235681789192e-06,
+      "loss": 2.3989,
+      "step": 28480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.098983800331649e-06,
+      "loss": 1.8529,
+      "step": 28490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.094731918874103e-06,
+      "loss": 2.0539,
+      "step": 28500
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.7728729248046875,
+      "eval_runtime": 176.2414,
+      "eval_samples_per_second": 14.202,
+      "eval_steps_per_second": 7.104,
+      "step": 28500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.090480037416557e-06,
+      "loss": 1.6733,
+      "step": 28510
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.086228155959014e-06,
+      "loss": 2.2849,
+      "step": 28520
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.081976274501468e-06,
+      "loss": 2.5538,
+      "step": 28530
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.077724393043923e-06,
+      "loss": 2.4862,
+      "step": 28540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.073472511586377e-06,
+      "loss": 2.3368,
+      "step": 28550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.069220630128831e-06,
+      "loss": 2.6631,
+      "step": 28560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.064968748671288e-06,
+      "loss": 2.1342,
+      "step": 28570
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.060716867213742e-06,
+      "loss": 2.0555,
+      "step": 28580
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.056464985756198e-06,
+      "loss": 2.3854,
+      "step": 28590
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.052213104298653e-06,
+      "loss": 2.4276,
+      "step": 28600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.047961222841107e-06,
+      "loss": 2.2421,
+      "step": 28610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.043709341383563e-06,
+      "loss": 1.6778,
+      "step": 28620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.039457459926018e-06,
+      "loss": 2.1492,
+      "step": 28630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.035205578468474e-06,
+      "loss": 1.6066,
+      "step": 28640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.030953697010928e-06,
+      "loss": 2.134,
+      "step": 28650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.026701815553383e-06,
+      "loss": 1.6651,
+      "step": 28660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.022449934095839e-06,
+      "loss": 2.3821,
+      "step": 28670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.018198052638293e-06,
+      "loss": 1.8024,
+      "step": 28680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.013946171180748e-06,
+      "loss": 2.3084,
+      "step": 28690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.009694289723204e-06,
+      "loss": 2.1194,
+      "step": 28700
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.005442408265658e-06,
+      "loss": 1.9815,
+      "step": 28710
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.001190526808113e-06,
+      "loss": 1.8908,
+      "step": 28720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.996938645350567e-06,
+      "loss": 2.0239,
+      "step": 28730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.992686763893024e-06,
+      "loss": 2.3303,
+      "step": 28740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.988434882435478e-06,
+      "loss": 1.8689,
+      "step": 28750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.984183000977932e-06,
+      "loss": 2.2181,
+      "step": 28760
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.979931119520389e-06,
+      "loss": 2.3179,
+      "step": 28770
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.975679238062843e-06,
+      "loss": 2.0093,
+      "step": 28780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.971427356605298e-06,
+      "loss": 2.3236,
+      "step": 28790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.967175475147754e-06,
+      "loss": 2.1852,
+      "step": 28800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.962923593690208e-06,
+      "loss": 2.062,
+      "step": 28810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.958671712232664e-06,
+      "loss": 2.6228,
+      "step": 28820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.954419830775119e-06,
+      "loss": 2.2448,
+      "step": 28830
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.950167949317573e-06,
+      "loss": 2.3091,
+      "step": 28840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.94591606786003e-06,
+      "loss": 2.1589,
+      "step": 28850
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.941664186402484e-06,
+      "loss": 1.9526,
+      "step": 28860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.93741230494494e-06,
+      "loss": 2.165,
+      "step": 28870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.933160423487394e-06,
+      "loss": 2.0298,
+      "step": 28880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.928908542029849e-06,
+      "loss": 2.2625,
+      "step": 28890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.924656660572303e-06,
+      "loss": 2.1089,
+      "step": 28900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.92040477911476e-06,
+      "loss": 2.4739,
+      "step": 28910
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.916152897657214e-06,
+      "loss": 1.879,
+      "step": 28920
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.911901016199668e-06,
+      "loss": 2.0006,
+      "step": 28930
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.907649134742123e-06,
+      "loss": 2.1949,
+      "step": 28940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.903397253284579e-06,
+      "loss": 2.3032,
+      "step": 28950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.899145371827033e-06,
+      "loss": 2.0933,
+      "step": 28960
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.89489349036949e-06,
+      "loss": 1.83,
+      "step": 28970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.890641608911944e-06,
+      "loss": 2.0929,
+      "step": 28980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.886389727454399e-06,
+      "loss": 2.0977,
+      "step": 28990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.882137845996855e-06,
+      "loss": 2.0148,
+      "step": 29000
+    },
+    {
+      "epoch": 1.22,
+      "eval_loss": 1.784656047821045,
+      "eval_runtime": 174.4708,
+      "eval_samples_per_second": 14.346,
+      "eval_steps_per_second": 7.176,
+      "step": 29000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.87788596453931e-06,
+      "loss": 2.2318,
+      "step": 29010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.873634083081765e-06,
+      "loss": 2.0887,
+      "step": 29020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.86938220162422e-06,
+      "loss": 2.6538,
+      "step": 29030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.865130320166674e-06,
+      "loss": 2.1592,
+      "step": 29040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.86087843870913e-06,
+      "loss": 2.2841,
+      "step": 29050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.856626557251585e-06,
+      "loss": 2.4645,
+      "step": 29060
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.85237467579404e-06,
+      "loss": 2.272,
+      "step": 29070
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.848122794336494e-06,
+      "loss": 2.0791,
+      "step": 29080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.84387091287895e-06,
+      "loss": 2.5805,
+      "step": 29090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.839619031421404e-06,
+      "loss": 2.1916,
+      "step": 29100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.835367149963859e-06,
+      "loss": 1.6838,
+      "step": 29110
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.831115268506315e-06,
+      "loss": 2.1771,
+      "step": 29120
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.82686338704877e-06,
+      "loss": 2.0263,
+      "step": 29130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.822611505591224e-06,
+      "loss": 2.1806,
+      "step": 29140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.81835962413368e-06,
+      "loss": 2.3806,
+      "step": 29150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.814107742676135e-06,
+      "loss": 2.0126,
+      "step": 29160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.80985586121859e-06,
+      "loss": 1.817,
+      "step": 29170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.805603979761045e-06,
+      "loss": 2.336,
+      "step": 29180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.8013520983035e-06,
+      "loss": 2.2979,
+      "step": 29190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.797100216845956e-06,
+      "loss": 2.3,
+      "step": 29200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.79284833538841e-06,
+      "loss": 1.6785,
+      "step": 29210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.788596453930865e-06,
+      "loss": 2.1275,
+      "step": 29220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.78434457247332e-06,
+      "loss": 2.4896,
+      "step": 29230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.780092691015775e-06,
+      "loss": 2.3844,
+      "step": 29240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.77584080955823e-06,
+      "loss": 2.2897,
+      "step": 29250
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.771588928100686e-06,
+      "loss": 1.7411,
+      "step": 29260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.76733704664314e-06,
+      "loss": 2.4193,
+      "step": 29270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.763085165185595e-06,
+      "loss": 2.0962,
+      "step": 29280
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.75883328372805e-06,
+      "loss": 2.1699,
+      "step": 29290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.754581402270505e-06,
+      "loss": 2.2598,
+      "step": 29300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.75032952081296e-06,
+      "loss": 2.0741,
+      "step": 29310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.746077639355414e-06,
+      "loss": 2.0311,
+      "step": 29320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.74182575789787e-06,
+      "loss": 2.4243,
+      "step": 29330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.737573876440325e-06,
+      "loss": 2.3444,
+      "step": 29340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.733321994982781e-06,
+      "loss": 2.1163,
+      "step": 29350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.729070113525236e-06,
+      "loss": 2.3206,
+      "step": 29360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.72481823206769e-06,
+      "loss": 1.7765,
+      "step": 29370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.720566350610146e-06,
+      "loss": 2.348,
+      "step": 29380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.7163144691526e-06,
+      "loss": 2.2471,
+      "step": 29390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.712062587695057e-06,
+      "loss": 2.2791,
+      "step": 29400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.707810706237511e-06,
+      "loss": 1.6969,
+      "step": 29410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.703558824779966e-06,
+      "loss": 2.3942,
+      "step": 29420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.699306943322422e-06,
+      "loss": 1.9153,
+      "step": 29430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.695055061864876e-06,
+      "loss": 2.3807,
+      "step": 29440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.69080318040733e-06,
+      "loss": 2.1779,
+      "step": 29450
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.686551298949785e-06,
+      "loss": 1.9697,
+      "step": 29460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.68229941749224e-06,
+      "loss": 2.3212,
+      "step": 29470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.678047536034696e-06,
+      "loss": 1.9044,
+      "step": 29480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.67379565457715e-06,
+      "loss": 2.4155,
+      "step": 29490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.669543773119606e-06,
+      "loss": 1.9734,
+      "step": 29500
+    },
+    {
+      "epoch": 1.24,
+      "eval_loss": 1.762258768081665,
+      "eval_runtime": 175.5265,
+      "eval_samples_per_second": 14.26,
+      "eval_steps_per_second": 7.133,
+      "step": 29500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.665291891662061e-06,
+      "loss": 2.1499,
+      "step": 29510
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.661040010204515e-06,
+      "loss": 2.0907,
+      "step": 29520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.656788128746971e-06,
+      "loss": 1.8271,
+      "step": 29530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.652536247289426e-06,
+      "loss": 2.0215,
+      "step": 29540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.648284365831882e-06,
+      "loss": 2.1363,
+      "step": 29550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.644032484374337e-06,
+      "loss": 2.2156,
+      "step": 29560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.639780602916791e-06,
+      "loss": 1.9902,
+      "step": 29570
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.635528721459247e-06,
+      "loss": 1.8233,
+      "step": 29580
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.631276840001702e-06,
+      "loss": 1.9334,
+      "step": 29590
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.627024958544157e-06,
+      "loss": 2.4853,
+      "step": 29600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.622773077086611e-06,
+      "loss": 2.425,
+      "step": 29610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.618521195629066e-06,
+      "loss": 2.1252,
+      "step": 29620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.614269314171522e-06,
+      "loss": 2.0679,
+      "step": 29630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.6100174327139765e-06,
+      "loss": 2.2087,
+      "step": 29640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.605765551256432e-06,
+      "loss": 2.0734,
+      "step": 29650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.601513669798887e-06,
+      "loss": 2.2149,
+      "step": 29660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5972617883413415e-06,
+      "loss": 2.3035,
+      "step": 29670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.593009906883797e-06,
+      "loss": 2.2851,
+      "step": 29680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.588758025426251e-06,
+      "loss": 2.4831,
+      "step": 29690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5845061439687075e-06,
+      "loss": 2.1056,
+      "step": 29700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.580254262511162e-06,
+      "loss": 1.8801,
+      "step": 29710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.576002381053616e-06,
+      "loss": 2.2619,
+      "step": 29720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5717504995960725e-06,
+      "loss": 2.1201,
+      "step": 29730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.567498618138527e-06,
+      "loss": 2.043,
+      "step": 29740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5632467366809814e-06,
+      "loss": 1.9556,
+      "step": 29750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.558994855223437e-06,
+      "loss": 2.2447,
+      "step": 29760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.554742973765892e-06,
+      "loss": 1.8696,
+      "step": 29770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.550491092308347e-06,
+      "loss": 2.0987,
+      "step": 29780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.546239210850802e-06,
+      "loss": 2.1271,
+      "step": 29790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.541987329393256e-06,
+      "loss": 2.0724,
+      "step": 29800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.5377354479357124e-06,
+      "loss": 2.1989,
+      "step": 29810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.533483566478167e-06,
+      "loss": 2.011,
+      "step": 29820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.529231685020623e-06,
+      "loss": 2.1491,
+      "step": 29830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.5249798035630775e-06,
+      "loss": 2.0264,
+      "step": 29840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.520727922105532e-06,
+      "loss": 2.0702,
+      "step": 29850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.516476040647987e-06,
+      "loss": 2.1978,
+      "step": 29860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.512224159190442e-06,
+      "loss": 2.2314,
+      "step": 29870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.507972277732898e-06,
+      "loss": 2.3425,
+      "step": 29880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.503720396275352e-06,
+      "loss": 1.9546,
+      "step": 29890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.499468514817807e-06,
+      "loss": 2.7283,
+      "step": 29900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.495216633360263e-06,
+      "loss": 2.3941,
+      "step": 29910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.490964751902717e-06,
+      "loss": 1.8176,
+      "step": 29920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.486712870445173e-06,
+      "loss": 2.1529,
+      "step": 29930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.482460988987628e-06,
+      "loss": 2.0351,
+      "step": 29940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.4782091075300825e-06,
+      "loss": 1.8016,
+      "step": 29950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.473957226072538e-06,
+      "loss": 1.6143,
+      "step": 29960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.469705344614992e-06,
+      "loss": 1.9063,
+      "step": 29970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.465453463157448e-06,
+      "loss": 2.4023,
+      "step": 29980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.461201581699903e-06,
+      "loss": 1.9606,
+      "step": 29990
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.456949700242357e-06,
+      "loss": 2.2853,
+      "step": 30000
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 1.768416404724121,
+      "eval_runtime": 175.6228,
+      "eval_samples_per_second": 14.252,
+      "eval_steps_per_second": 7.129,
+      "step": 30000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.4526978187848135e-06,
+      "loss": 1.8119,
+      "step": 30010
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.448445937327268e-06,
+      "loss": 1.8843,
+      "step": 30020
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.444194055869723e-06,
+      "loss": 2.4566,
+      "step": 30030
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.439942174412178e-06,
+      "loss": 1.6584,
+      "step": 30040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.435690292954633e-06,
+      "loss": 2.5253,
+      "step": 30050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.431438411497088e-06,
+      "loss": 2.1786,
+      "step": 30060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.427186530039543e-06,
+      "loss": 2.2506,
+      "step": 30070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.422934648581999e-06,
+      "loss": 2.1316,
+      "step": 30080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.418682767124453e-06,
+      "loss": 2.0686,
+      "step": 30090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.414430885666908e-06,
+      "loss": 1.7463,
+      "step": 30100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.410179004209363e-06,
+      "loss": 2.2856,
+      "step": 30110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.405927122751818e-06,
+      "loss": 1.9305,
+      "step": 30120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.401675241294274e-06,
+      "loss": 1.7928,
+      "step": 30130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.397423359836728e-06,
+      "loss": 2.4415,
+      "step": 30140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.393171478379183e-06,
+      "loss": 2.2208,
+      "step": 30150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.388919596921639e-06,
+      "loss": 2.0496,
+      "step": 30160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.384667715464093e-06,
+      "loss": 2.1804,
+      "step": 30170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.380415834006549e-06,
+      "loss": 2.5044,
+      "step": 30180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.376163952549004e-06,
+      "loss": 2.0653,
+      "step": 30190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.371912071091458e-06,
+      "loss": 2.1761,
+      "step": 30200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.367660189633914e-06,
+      "loss": 2.1186,
+      "step": 30210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.363408308176369e-06,
+      "loss": 2.1766,
+      "step": 30220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.359156426718823e-06,
+      "loss": 2.4063,
+      "step": 30230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.354904545261279e-06,
+      "loss": 2.0167,
+      "step": 30240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.350652663803733e-06,
+      "loss": 2.4051,
+      "step": 30250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.346400782346189e-06,
+      "loss": 2.0316,
+      "step": 30260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.342148900888644e-06,
+      "loss": 1.8998,
+      "step": 30270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.337897019431098e-06,
+      "loss": 2.305,
+      "step": 30280
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.333645137973554e-06,
+      "loss": 2.0012,
+      "step": 30290
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.329393256516009e-06,
+      "loss": 2.2966,
+      "step": 30300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.325141375058464e-06,
+      "loss": 2.3181,
+      "step": 30310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.320889493600919e-06,
+      "loss": 2.2078,
+      "step": 30320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.316637612143374e-06,
+      "loss": 2.0477,
+      "step": 30330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.312385730685829e-06,
+      "loss": 2.0226,
+      "step": 30340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.308133849228284e-06,
+      "loss": 2.1437,
+      "step": 30350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.30388196777074e-06,
+      "loss": 2.5629,
+      "step": 30360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.299630086313194e-06,
+      "loss": 2.4503,
+      "step": 30370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.295378204855649e-06,
+      "loss": 1.9968,
+      "step": 30380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.291126323398104e-06,
+      "loss": 2.0472,
+      "step": 30390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.286874441940559e-06,
+      "loss": 2.2337,
+      "step": 30400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.282622560483015e-06,
+      "loss": 1.8869,
+      "step": 30410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.278370679025469e-06,
+      "loss": 2.1899,
+      "step": 30420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.2741187975679236e-06,
+      "loss": 1.478,
+      "step": 30430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.26986691611038e-06,
+      "loss": 2.2113,
+      "step": 30440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.265615034652834e-06,
+      "loss": 1.8112,
+      "step": 30450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.26136315319529e-06,
+      "loss": 2.0767,
+      "step": 30460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.257111271737745e-06,
+      "loss": 2.2814,
+      "step": 30470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.252859390280199e-06,
+      "loss": 2.4961,
+      "step": 30480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.2486075088226546e-06,
+      "loss": 2.1519,
+      "step": 30490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.24435562736511e-06,
+      "loss": 2.2354,
+      "step": 30500
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 1.7716997861862183,
+      "eval_runtime": 175.8883,
+      "eval_samples_per_second": 14.231,
+      "eval_steps_per_second": 7.118,
+      "step": 30500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.240103745907565e-06,
+      "loss": 2.3328,
+      "step": 30510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.23585186445002e-06,
+      "loss": 1.8702,
+      "step": 30520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.231599982992474e-06,
+      "loss": 2.2167,
+      "step": 30530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.22734810153493e-06,
+      "loss": 1.9658,
+      "step": 30540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.223096220077385e-06,
+      "loss": 1.9015,
+      "step": 30550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.21884433861984e-06,
+      "loss": 1.6753,
+      "step": 30560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.214592457162295e-06,
+      "loss": 2.3337,
+      "step": 30570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.21034057570475e-06,
+      "loss": 2.136,
+      "step": 30580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.206088694247205e-06,
+      "loss": 2.0738,
+      "step": 30590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.2018368127896595e-06,
+      "loss": 2.1366,
+      "step": 30600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.197584931332116e-06,
+      "loss": 2.1618,
+      "step": 30610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.19333304987457e-06,
+      "loss": 1.9825,
+      "step": 30620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.189081168417025e-06,
+      "loss": 1.8846,
+      "step": 30630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.184829286959481e-06,
+      "loss": 2.3258,
+      "step": 30640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.180577405501935e-06,
+      "loss": 2.2125,
+      "step": 30650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1763255240443905e-06,
+      "loss": 1.7932,
+      "step": 30660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.172073642586845e-06,
+      "loss": 2.1522,
+      "step": 30670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1678217611293e-06,
+      "loss": 2.0979,
+      "step": 30680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.163569879671756e-06,
+      "loss": 1.6993,
+      "step": 30690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.15931799821421e-06,
+      "loss": 2.257,
+      "step": 30700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.155066116756666e-06,
+      "loss": 2.1497,
+      "step": 30710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.150814235299121e-06,
+      "loss": 2.1923,
+      "step": 30720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.146562353841575e-06,
+      "loss": 1.9727,
+      "step": 30730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.142310472384031e-06,
+      "loss": 2.3352,
+      "step": 30740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.138058590926486e-06,
+      "loss": 1.7339,
+      "step": 30750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.13380670946894e-06,
+      "loss": 1.9481,
+      "step": 30760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.1295548280113955e-06,
+      "loss": 1.8164,
+      "step": 30770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.12530294655385e-06,
+      "loss": 1.7699,
+      "step": 30780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.121051065096306e-06,
+      "loss": 1.7728,
+      "step": 30790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.1167991836387606e-06,
+      "loss": 2.3271,
+      "step": 30800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.112547302181215e-06,
+      "loss": 2.4951,
+      "step": 30810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.108295420723671e-06,
+      "loss": 2.127,
+      "step": 30820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.104043539266126e-06,
+      "loss": 2.3694,
+      "step": 30830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.099791657808581e-06,
+      "loss": 2.6008,
+      "step": 30840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.095539776351036e-06,
+      "loss": 2.1604,
+      "step": 30850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.091287894893491e-06,
+      "loss": 2.303,
+      "step": 30860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.087036013435946e-06,
+      "loss": 1.8119,
+      "step": 30870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0827841319784005e-06,
+      "loss": 2.0222,
+      "step": 30880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.078532250520857e-06,
+      "loss": 2.493,
+      "step": 30890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.074280369063311e-06,
+      "loss": 1.7992,
+      "step": 30900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0700284876057655e-06,
+      "loss": 2.3118,
+      "step": 30910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.065776606148222e-06,
+      "loss": 2.258,
+      "step": 30920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.061524724690676e-06,
+      "loss": 2.1641,
+      "step": 30930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0572728432331314e-06,
+      "loss": 1.9602,
+      "step": 30940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.053020961775586e-06,
+      "loss": 2.3728,
+      "step": 30950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.048769080318041e-06,
+      "loss": 2.2433,
+      "step": 30960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.0445171988604965e-06,
+      "loss": 2.1174,
+      "step": 30970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.040265317402951e-06,
+      "loss": 2.2292,
+      "step": 30980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.036013435945407e-06,
+      "loss": 2.113,
+      "step": 30990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.031761554487862e-06,
+      "loss": 2.2651,
+      "step": 31000
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 1.7629889249801636,
+      "eval_runtime": 175.7029,
+      "eval_samples_per_second": 14.246,
+      "eval_steps_per_second": 7.126,
+      "step": 31000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.027509673030316e-06,
+      "loss": 2.0244,
+      "step": 31010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.023257791572771e-06,
+      "loss": 2.4088,
+      "step": 31020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.019005910115227e-06,
+      "loss": 1.9128,
+      "step": 31030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.014754028657682e-06,
+      "loss": 2.1604,
+      "step": 31040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.010502147200136e-06,
+      "loss": 1.9239,
+      "step": 31050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.006250265742591e-06,
+      "loss": 1.816,
+      "step": 31060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.001998384285047e-06,
+      "loss": 2.4882,
+      "step": 31070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9977465028275015e-06,
+      "loss": 2.2132,
+      "step": 31080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.993494621369958e-06,
+      "loss": 2.2542,
+      "step": 31090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.989242739912412e-06,
+      "loss": 1.9381,
+      "step": 31100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9849908584548666e-06,
+      "loss": 2.0868,
+      "step": 31110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.980738976997322e-06,
+      "loss": 2.2442,
+      "step": 31120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.976487095539777e-06,
+      "loss": 2.2274,
+      "step": 31130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9722352140822325e-06,
+      "loss": 2.1612,
+      "step": 31140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.967983332624687e-06,
+      "loss": 2.375,
+      "step": 31150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.963731451167141e-06,
+      "loss": 2.2004,
+      "step": 31160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9594795697095975e-06,
+      "loss": 2.187,
+      "step": 31170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.955227688252052e-06,
+      "loss": 2.2825,
+      "step": 31180
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.950975806794507e-06,
+      "loss": 2.3688,
+      "step": 31190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.946723925336963e-06,
+      "loss": 2.1461,
+      "step": 31200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.942472043879417e-06,
+      "loss": 1.8028,
+      "step": 31210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.938220162421872e-06,
+      "loss": 2.0668,
+      "step": 31220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.933968280964327e-06,
+      "loss": 1.7507,
+      "step": 31230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.929716399506782e-06,
+      "loss": 2.0785,
+      "step": 31240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9254645180492374e-06,
+      "loss": 2.0424,
+      "step": 31250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.921212636591692e-06,
+      "loss": 2.3862,
+      "step": 31260
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.916960755134148e-06,
+      "loss": 2.0227,
+      "step": 31270
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.9127088736766025e-06,
+      "loss": 2.5728,
+      "step": 31280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.908456992219057e-06,
+      "loss": 2.1277,
+      "step": 31290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.904205110761512e-06,
+      "loss": 1.9434,
+      "step": 31300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.899953229303968e-06,
+      "loss": 2.2722,
+      "step": 31310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.895701347846423e-06,
+      "loss": 1.9561,
+      "step": 31320
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.891449466388877e-06,
+      "loss": 1.9599,
+      "step": 31330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.887197584931332e-06,
+      "loss": 2.0738,
+      "step": 31340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.882945703473788e-06,
+      "loss": 1.8086,
+      "step": 31350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.878693822016242e-06,
+      "loss": 1.7633,
+      "step": 31360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8744419405586986e-06,
+      "loss": 1.924,
+      "step": 31370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.870190059101153e-06,
+      "loss": 2.0792,
+      "step": 31380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8659381776436075e-06,
+      "loss": 2.3389,
+      "step": 31390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.861686296186063e-06,
+      "loss": 2.4297,
+      "step": 31400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.857434414728518e-06,
+      "loss": 1.7596,
+      "step": 31410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.853182533270973e-06,
+      "loss": 2.2129,
+      "step": 31420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.848930651813428e-06,
+      "loss": 2.4238,
+      "step": 31430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.844678770355882e-06,
+      "loss": 2.1243,
+      "step": 31440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8404268888983385e-06,
+      "loss": 2.0298,
+      "step": 31450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.836175007440793e-06,
+      "loss": 1.9073,
+      "step": 31460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.831923125983248e-06,
+      "loss": 2.0428,
+      "step": 31470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8276712445257035e-06,
+      "loss": 2.3143,
+      "step": 31480
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.823419363068158e-06,
+      "loss": 1.8345,
+      "step": 31490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.819167481610613e-06,
+      "loss": 2.4983,
+      "step": 31500
+    },
+    {
+      "epoch": 1.33,
+      "eval_loss": 1.7589433193206787,
+      "eval_runtime": 175.8885,
+      "eval_samples_per_second": 14.231,
+      "eval_steps_per_second": 7.118,
+      "step": 31500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.814915600153068e-06,
+      "loss": 2.1111,
+      "step": 31510
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.810663718695524e-06,
+      "loss": 1.6782,
+      "step": 31520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.806411837237978e-06,
+      "loss": 2.6629,
+      "step": 31530
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.802159955780433e-06,
+      "loss": 2.2417,
+      "step": 31540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.797908074322889e-06,
+      "loss": 2.1622,
+      "step": 31550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.7936561928653434e-06,
+      "loss": 2.2477,
+      "step": 31560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.789404311407799e-06,
+      "loss": 1.5379,
+      "step": 31570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.785152429950253e-06,
+      "loss": 1.9817,
+      "step": 31580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.7809005484927085e-06,
+      "loss": 2.1176,
+      "step": 31590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.776648667035164e-06,
+      "loss": 2.0681,
+      "step": 31600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.772396785577618e-06,
+      "loss": 1.9443,
+      "step": 31610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.7681449041200744e-06,
+      "loss": 2.6853,
+      "step": 31620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.763893022662529e-06,
+      "loss": 2.517,
+      "step": 31630
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.759641141204983e-06,
+      "loss": 1.8634,
+      "step": 31640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.7553892597474395e-06,
+      "loss": 1.9171,
+      "step": 31650
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.751137378289894e-06,
+      "loss": 1.9872,
+      "step": 31660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.746885496832349e-06,
+      "loss": 2.1077,
+      "step": 31670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.742633615374804e-06,
+      "loss": 2.1553,
+      "step": 31680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.738381733917258e-06,
+      "loss": 2.2523,
+      "step": 31690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.734129852459714e-06,
+      "loss": 2.293,
+      "step": 31700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.729877971002169e-06,
+      "loss": 2.3029,
+      "step": 31710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.725626089544625e-06,
+      "loss": 1.7122,
+      "step": 31720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.721374208087079e-06,
+      "loss": 2.3072,
+      "step": 31730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.717122326629534e-06,
+      "loss": 2.0192,
+      "step": 31740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.712870445171989e-06,
+      "loss": 2.2539,
+      "step": 31750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.7086185637144445e-06,
+      "loss": 2.0044,
+      "step": 31760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.704366682256899e-06,
+      "loss": 2.2183,
+      "step": 31770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.700114800799354e-06,
+      "loss": 1.8438,
+      "step": 31780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.695862919341809e-06,
+      "loss": 2.1252,
+      "step": 31790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.691611037884265e-06,
+      "loss": 1.9709,
+      "step": 31800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.687359156426719e-06,
+      "loss": 2.1972,
+      "step": 31810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.683107274969174e-06,
+      "loss": 2.5023,
+      "step": 31820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.67885539351163e-06,
+      "loss": 2.1028,
+      "step": 31830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.674603512054084e-06,
+      "loss": 1.8368,
+      "step": 31840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.67035163059654e-06,
+      "loss": 1.742,
+      "step": 31850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.666099749138994e-06,
+      "loss": 2.1388,
+      "step": 31860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.6618478676814494e-06,
+      "loss": 1.8245,
+      "step": 31870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.657595986223905e-06,
+      "loss": 2.1489,
+      "step": 31880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.653344104766359e-06,
+      "loss": 2.0893,
+      "step": 31890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.649092223308815e-06,
+      "loss": 2.1849,
+      "step": 31900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.64484034185127e-06,
+      "loss": 2.0908,
+      "step": 31910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.640588460393724e-06,
+      "loss": 1.8421,
+      "step": 31920
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.63633657893618e-06,
+      "loss": 2.0227,
+      "step": 31930
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.632084697478635e-06,
+      "loss": 2.4361,
+      "step": 31940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.62783281602109e-06,
+      "loss": 2.0913,
+      "step": 31950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.623580934563545e-06,
+      "loss": 2.2489,
+      "step": 31960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.619329053105999e-06,
+      "loss": 2.0853,
+      "step": 31970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.615077171648455e-06,
+      "loss": 1.5733,
+      "step": 31980
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.61082529019091e-06,
+      "loss": 2.4172,
+      "step": 31990
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.606573408733366e-06,
+      "loss": 1.9044,
+      "step": 32000
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 1.7534154653549194,
+      "eval_runtime": 175.7062,
+      "eval_samples_per_second": 14.245,
+      "eval_steps_per_second": 7.126,
+      "step": 32000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.60232152727582e-06,
+      "loss": 1.9327,
+      "step": 32010
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.598069645818275e-06,
+      "loss": 2.2584,
+      "step": 32020
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.59381776436073e-06,
+      "loss": 2.2586,
+      "step": 32030
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.589565882903185e-06,
+      "loss": 1.9329,
+      "step": 32040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.585314001445641e-06,
+      "loss": 1.9196,
+      "step": 32050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.581062119988095e-06,
+      "loss": 1.9311,
+      "step": 32060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.57681023853055e-06,
+      "loss": 2.0028,
+      "step": 32070
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.572558357073006e-06,
+      "loss": 2.0391,
+      "step": 32080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.56830647561546e-06,
+      "loss": 2.1205,
+      "step": 32090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.5640545941579155e-06,
+      "loss": 1.9866,
+      "step": 32100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.559802712700371e-06,
+      "loss": 2.078,
+      "step": 32110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.555550831242825e-06,
+      "loss": 2.1073,
+      "step": 32120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.551298949785281e-06,
+      "loss": 2.1966,
+      "step": 32130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.547047068327735e-06,
+      "loss": 2.0649,
+      "step": 32140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.542795186870191e-06,
+      "loss": 2.1104,
+      "step": 32150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.538543305412646e-06,
+      "loss": 2.4335,
+      "step": 32160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.5342914239551e-06,
+      "loss": 1.6751,
+      "step": 32170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.530039542497556e-06,
+      "loss": 2.1906,
+      "step": 32180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.525787661040011e-06,
+      "loss": 1.8149,
+      "step": 32190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.521535779582466e-06,
+      "loss": 2.2962,
+      "step": 32200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.5172838981249205e-06,
+      "loss": 2.3522,
+      "step": 32210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.513032016667376e-06,
+      "loss": 1.9007,
+      "step": 32220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.508780135209831e-06,
+      "loss": 1.5771,
+      "step": 32230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.5045282537522856e-06,
+      "loss": 1.8895,
+      "step": 32240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.50027637229474e-06,
+      "loss": 2.3177,
+      "step": 32250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.496024490837196e-06,
+      "loss": 1.8772,
+      "step": 32260
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.491772609379651e-06,
+      "loss": 1.78,
+      "step": 32270
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.487520727922107e-06,
+      "loss": 2.3023,
+      "step": 32280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.483268846464561e-06,
+      "loss": 2.0458,
+      "step": 32290
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.479016965007016e-06,
+      "loss": 1.8302,
+      "step": 32300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.474765083549471e-06,
+      "loss": 1.8139,
+      "step": 32310
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.470513202091926e-06,
+      "loss": 2.1669,
+      "step": 32320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.466261320634382e-06,
+      "loss": 2.1452,
+      "step": 32330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.462009439176836e-06,
+      "loss": 2.1107,
+      "step": 32340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4577575577192905e-06,
+      "loss": 1.9855,
+      "step": 32350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.453505676261747e-06,
+      "loss": 2.0678,
+      "step": 32360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.449253794804201e-06,
+      "loss": 1.7296,
+      "step": 32370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4450019133466565e-06,
+      "loss": 1.9016,
+      "step": 32380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.440750031889112e-06,
+      "loss": 2.5799,
+      "step": 32390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.436498150431566e-06,
+      "loss": 2.2696,
+      "step": 32400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.4322462689740215e-06,
+      "loss": 2.1848,
+      "step": 32410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.427994387516476e-06,
+      "loss": 2.1398,
+      "step": 32420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.423742506058932e-06,
+      "loss": 1.9232,
+      "step": 32430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.419490624601387e-06,
+      "loss": 2.2663,
+      "step": 32440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.415238743143841e-06,
+      "loss": 2.1383,
+      "step": 32450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.410986861686297e-06,
+      "loss": 2.0458,
+      "step": 32460
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.406734980228752e-06,
+      "loss": 1.9462,
+      "step": 32470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.402483098771207e-06,
+      "loss": 2.4849,
+      "step": 32480
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3982312173136614e-06,
+      "loss": 2.114,
+      "step": 32490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.393979335856117e-06,
+      "loss": 2.0785,
+      "step": 32500
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 1.7712970972061157,
+      "eval_runtime": 174.1828,
+      "eval_samples_per_second": 14.37,
+      "eval_steps_per_second": 7.188,
+      "step": 32500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.389727454398572e-06,
+      "loss": 2.2324,
+      "step": 32510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3854755729410265e-06,
+      "loss": 2.4942,
+      "step": 32520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.381223691483483e-06,
+      "loss": 2.2043,
+      "step": 32530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.376971810025937e-06,
+      "loss": 2.3819,
+      "step": 32540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3727199285683916e-06,
+      "loss": 2.0652,
+      "step": 32550
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.368468047110848e-06,
+      "loss": 2.3862,
+      "step": 32560
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.364216165653302e-06,
+      "loss": 2.0991,
+      "step": 32570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3599642841957575e-06,
+      "loss": 2.0504,
+      "step": 32580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.355712402738212e-06,
+      "loss": 1.8913,
+      "step": 32590
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.351460521280666e-06,
+      "loss": 1.8265,
+      "step": 32600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.3472086398231226e-06,
+      "loss": 1.9062,
+      "step": 32610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.342956758365577e-06,
+      "loss": 1.8239,
+      "step": 32620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.338704876908033e-06,
+      "loss": 1.8398,
+      "step": 32630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.334452995450488e-06,
+      "loss": 2.1208,
+      "step": 32640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.330201113992942e-06,
+      "loss": 2.2409,
+      "step": 32650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.325949232535397e-06,
+      "loss": 1.9637,
+      "step": 32660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.321697351077853e-06,
+      "loss": 1.8296,
+      "step": 32670
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.317445469620308e-06,
+      "loss": 2.0955,
+      "step": 32680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.3131935881627625e-06,
+      "loss": 1.8445,
+      "step": 32690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.308941706705217e-06,
+      "loss": 2.1954,
+      "step": 32700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.304689825247673e-06,
+      "loss": 1.9115,
+      "step": 32710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.3004379437901275e-06,
+      "loss": 1.792,
+      "step": 32720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.296186062332583e-06,
+      "loss": 2.1596,
+      "step": 32730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.291934180875038e-06,
+      "loss": 1.7065,
+      "step": 32740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.287682299417493e-06,
+      "loss": 1.9132,
+      "step": 32750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.283430417959948e-06,
+      "loss": 2.6364,
+      "step": 32760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.279178536502402e-06,
+      "loss": 2.143,
+      "step": 32770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.274926655044858e-06,
+      "loss": 2.4071,
+      "step": 32780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.270674773587313e-06,
+      "loss": 2.0815,
+      "step": 32790
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2664228921297674e-06,
+      "loss": 2.3196,
+      "step": 32800
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.262171010672224e-06,
+      "loss": 2.3816,
+      "step": 32810
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.257919129214678e-06,
+      "loss": 2.175,
+      "step": 32820
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2536672477571325e-06,
+      "loss": 1.9625,
+      "step": 32830
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.249415366299588e-06,
+      "loss": 1.8776,
+      "step": 32840
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.245163484842043e-06,
+      "loss": 2.041,
+      "step": 32850
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.240911603384498e-06,
+      "loss": 2.3219,
+      "step": 32860
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.236659721926953e-06,
+      "loss": 2.2202,
+      "step": 32870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.232407840469407e-06,
+      "loss": 2.2712,
+      "step": 32880
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2281559590118635e-06,
+      "loss": 1.8479,
+      "step": 32890
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.223904077554318e-06,
+      "loss": 2.4935,
+      "step": 32900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.219652196096774e-06,
+      "loss": 2.167,
+      "step": 32910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.2154003146392286e-06,
+      "loss": 1.871,
+      "step": 32920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.211148433181683e-06,
+      "loss": 1.8298,
+      "step": 32930
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.206896551724138e-06,
+      "loss": 2.3033,
+      "step": 32940
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.202644670266594e-06,
+      "loss": 2.0429,
+      "step": 32950
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.198392788809049e-06,
+      "loss": 1.8533,
+      "step": 32960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.194140907351503e-06,
+      "loss": 1.7284,
+      "step": 32970
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.189889025893958e-06,
+      "loss": 2.4582,
+      "step": 32980
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.185637144436414e-06,
+      "loss": 1.625,
+      "step": 32990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1813852629788685e-06,
+      "loss": 1.9835,
+      "step": 33000
+    },
+    {
+      "epoch": 1.39,
+      "eval_loss": 1.7666312456130981,
+      "eval_runtime": 175.9092,
+      "eval_samples_per_second": 14.229,
+      "eval_steps_per_second": 7.117,
+      "step": 33000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.177133381521324e-06,
+      "loss": 2.0395,
+      "step": 33010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.172881500063779e-06,
+      "loss": 1.8279,
+      "step": 33020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1686296186062335e-06,
+      "loss": 2.1629,
+      "step": 33030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.164377737148689e-06,
+      "loss": 2.4225,
+      "step": 33040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.160125855691143e-06,
+      "loss": 1.9997,
+      "step": 33050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1558739742335994e-06,
+      "loss": 2.0743,
+      "step": 33060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.151622092776054e-06,
+      "loss": 1.9704,
+      "step": 33070
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.147370211318508e-06,
+      "loss": 2.1134,
+      "step": 33080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.1431183298609645e-06,
+      "loss": 1.9697,
+      "step": 33090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.138866448403419e-06,
+      "loss": 2.3882,
+      "step": 33100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.134614566945874e-06,
+      "loss": 2.5031,
+      "step": 33110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.130362685488329e-06,
+      "loss": 1.9838,
+      "step": 33120
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.126110804030784e-06,
+      "loss": 2.5724,
+      "step": 33130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.121858922573239e-06,
+      "loss": 2.175,
+      "step": 33140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.117607041115694e-06,
+      "loss": 1.6917,
+      "step": 33150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.11335515965815e-06,
+      "loss": 2.0187,
+      "step": 33160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.109103278200604e-06,
+      "loss": 2.1692,
+      "step": 33170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.104851396743059e-06,
+      "loss": 2.4369,
+      "step": 33180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.100599515285515e-06,
+      "loss": 1.7183,
+      "step": 33190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0963476338279695e-06,
+      "loss": 1.6395,
+      "step": 33200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.092095752370425e-06,
+      "loss": 2.3003,
+      "step": 33210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.087843870912879e-06,
+      "loss": 2.0762,
+      "step": 33220
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0835919894553346e-06,
+      "loss": 2.5726,
+      "step": 33230
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.07934010799779e-06,
+      "loss": 2.1681,
+      "step": 33240
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.075088226540244e-06,
+      "loss": 2.3166,
+      "step": 33250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0708363450827005e-06,
+      "loss": 2.4197,
+      "step": 33260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.066584463625155e-06,
+      "loss": 2.1967,
+      "step": 33270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.062332582167609e-06,
+      "loss": 2.246,
+      "step": 33280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.058080700710065e-06,
+      "loss": 2.095,
+      "step": 33290
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.05382881925252e-06,
+      "loss": 2.4741,
+      "step": 33300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0495769377949745e-06,
+      "loss": 1.7434,
+      "step": 33310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.04532505633743e-06,
+      "loss": 2.0141,
+      "step": 33320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.041073174879884e-06,
+      "loss": 2.2197,
+      "step": 33330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.03682129342234e-06,
+      "loss": 2.3397,
+      "step": 33340
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.032569411964795e-06,
+      "loss": 2.1607,
+      "step": 33350
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.028317530507249e-06,
+      "loss": 2.1818,
+      "step": 33360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.0240656490497054e-06,
+      "loss": 2.0139,
+      "step": 33370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.01981376759216e-06,
+      "loss": 2.3018,
+      "step": 33380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.015561886134615e-06,
+      "loss": 2.0767,
+      "step": 33390
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.01131000467707e-06,
+      "loss": 2.0438,
+      "step": 33400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.007058123219525e-06,
+      "loss": 2.016,
+      "step": 33410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.00280624176198e-06,
+      "loss": 1.6808,
+      "step": 33420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.998554360304435e-06,
+      "loss": 1.8655,
+      "step": 33430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.994302478846891e-06,
+      "loss": 1.9778,
+      "step": 33440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.990050597389345e-06,
+      "loss": 2.5145,
+      "step": 33450
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.9857987159318e-06,
+      "loss": 2.1583,
+      "step": 33460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.981546834474256e-06,
+      "loss": 2.1911,
+      "step": 33470
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.97729495301671e-06,
+      "loss": 2.1878,
+      "step": 33480
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.973043071559166e-06,
+      "loss": 1.7057,
+      "step": 33490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.96879119010162e-06,
+      "loss": 2.5451,
+      "step": 33500
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 1.7640694379806519,
+      "eval_runtime": 175.7004,
+      "eval_samples_per_second": 14.246,
+      "eval_steps_per_second": 7.126,
+      "step": 33500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.964539308644075e-06,
+      "loss": 1.8074,
+      "step": 33510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.960287427186531e-06,
+      "loss": 2.3821,
+      "step": 33520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.956035545728985e-06,
+      "loss": 2.203,
+      "step": 33530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.951783664271441e-06,
+      "loss": 2.1969,
+      "step": 33540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.947531782813896e-06,
+      "loss": 2.5106,
+      "step": 33550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.94327990135635e-06,
+      "loss": 2.1026,
+      "step": 33560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.939028019898806e-06,
+      "loss": 1.931,
+      "step": 33570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.934776138441261e-06,
+      "loss": 2.279,
+      "step": 33580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.930524256983716e-06,
+      "loss": 2.0317,
+      "step": 33590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.926272375526171e-06,
+      "loss": 2.4391,
+      "step": 33600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.922020494068625e-06,
+      "loss": 2.2738,
+      "step": 33610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.917768612611081e-06,
+      "loss": 2.0268,
+      "step": 33620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.913516731153536e-06,
+      "loss": 2.2811,
+      "step": 33630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.909264849695991e-06,
+      "loss": 1.6955,
+      "step": 33640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.905012968238446e-06,
+      "loss": 1.9845,
+      "step": 33650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.900761086780901e-06,
+      "loss": 2.0359,
+      "step": 33660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.896509205323356e-06,
+      "loss": 2.185,
+      "step": 33670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.892257323865811e-06,
+      "loss": 2.0223,
+      "step": 33680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.888005442408267e-06,
+      "loss": 1.8995,
+      "step": 33690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.883753560950721e-06,
+      "loss": 2.2747,
+      "step": 33700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.879501679493176e-06,
+      "loss": 2.0273,
+      "step": 33710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.875249798035632e-06,
+      "loss": 1.6985,
+      "step": 33720
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.870997916578086e-06,
+      "loss": 1.9237,
+      "step": 33730
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.866746035120542e-06,
+      "loss": 2.2394,
+      "step": 33740
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.862494153662996e-06,
+      "loss": 2.4825,
+      "step": 33750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.858242272205451e-06,
+      "loss": 2.0602,
+      "step": 33760
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.853990390747907e-06,
+      "loss": 2.2383,
+      "step": 33770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.849738509290361e-06,
+      "loss": 2.3586,
+      "step": 33780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8454866278328156e-06,
+      "loss": 1.9765,
+      "step": 33790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.841234746375272e-06,
+      "loss": 2.0437,
+      "step": 33800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.836982864917726e-06,
+      "loss": 1.6747,
+      "step": 33810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.832730983460182e-06,
+      "loss": 2.1325,
+      "step": 33820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.828479102002637e-06,
+      "loss": 2.1148,
+      "step": 33830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.824227220545091e-06,
+      "loss": 1.809,
+      "step": 33840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8199753390875466e-06,
+      "loss": 2.4527,
+      "step": 33850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.815723457630002e-06,
+      "loss": 2.1952,
+      "step": 33860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.811471576172457e-06,
+      "loss": 1.9291,
+      "step": 33870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.807219694714912e-06,
+      "loss": 2.3615,
+      "step": 33880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.802967813257366e-06,
+      "loss": 2.2254,
+      "step": 33890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.798715931799822e-06,
+      "loss": 1.9964,
+      "step": 33900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.794464050342277e-06,
+      "loss": 2.2679,
+      "step": 33910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.790212168884732e-06,
+      "loss": 2.2388,
+      "step": 33920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.785960287427187e-06,
+      "loss": 1.9743,
+      "step": 33930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.781708405969642e-06,
+      "loss": 2.4368,
+      "step": 33940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.777456524512097e-06,
+      "loss": 1.5072,
+      "step": 33950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7732046430545515e-06,
+      "loss": 2.0864,
+      "step": 33960
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.768952761597008e-06,
+      "loss": 2.4148,
+      "step": 33970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.764700880139462e-06,
+      "loss": 1.8377,
+      "step": 33980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.760448998681917e-06,
+      "loss": 2.3857,
+      "step": 33990
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.756197117224373e-06,
+      "loss": 2.2245,
+      "step": 34000
+    },
+    {
+      "epoch": 1.43,
+      "eval_loss": 1.7518954277038574,
+      "eval_runtime": 175.8609,
+      "eval_samples_per_second": 14.233,
+      "eval_steps_per_second": 7.119,
+      "step": 34000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.751945235766827e-06,
+      "loss": 1.9569,
+      "step": 34010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7476933543092825e-06,
+      "loss": 2.3212,
+      "step": 34020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.743441472851737e-06,
+      "loss": 1.8699,
+      "step": 34030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.739189591394192e-06,
+      "loss": 2.2364,
+      "step": 34040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.734937709936648e-06,
+      "loss": 1.7592,
+      "step": 34050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.730685828479102e-06,
+      "loss": 2.54,
+      "step": 34060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.726433947021558e-06,
+      "loss": 1.9775,
+      "step": 34070
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.722182065564013e-06,
+      "loss": 1.7832,
+      "step": 34080
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.717930184106467e-06,
+      "loss": 1.8524,
+      "step": 34090
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.713678302648923e-06,
+      "loss": 2.1769,
+      "step": 34100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.709426421191378e-06,
+      "loss": 1.97,
+      "step": 34110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.705174539733833e-06,
+      "loss": 2.0002,
+      "step": 34120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7009226582762875e-06,
+      "loss": 2.2234,
+      "step": 34130
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.696670776818743e-06,
+      "loss": 2.2216,
+      "step": 34140
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.692418895361198e-06,
+      "loss": 2.0222,
+      "step": 34150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6881670139036526e-06,
+      "loss": 2.3931,
+      "step": 34160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.683915132446109e-06,
+      "loss": 2.0872,
+      "step": 34170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.679663250988563e-06,
+      "loss": 1.8038,
+      "step": 34180
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.675411369531018e-06,
+      "loss": 1.7638,
+      "step": 34190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.671159488073473e-06,
+      "loss": 1.9803,
+      "step": 34200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.666907606615928e-06,
+      "loss": 2.4514,
+      "step": 34210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6626557251583835e-06,
+      "loss": 2.0363,
+      "step": 34220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.658403843700838e-06,
+      "loss": 2.0239,
+      "step": 34230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6541519622432925e-06,
+      "loss": 2.4211,
+      "step": 34240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.649900080785749e-06,
+      "loss": 1.9727,
+      "step": 34250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.645648199328203e-06,
+      "loss": 2.3525,
+      "step": 34260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.641396317870658e-06,
+      "loss": 2.1225,
+      "step": 34270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.637144436413114e-06,
+      "loss": 1.8737,
+      "step": 34280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.632892554955568e-06,
+      "loss": 2.1636,
+      "step": 34290
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6286406734980234e-06,
+      "loss": 2.3626,
+      "step": 34300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.624388792040478e-06,
+      "loss": 1.9545,
+      "step": 34310
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.620136910582933e-06,
+      "loss": 2.0651,
+      "step": 34320
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.6158850291253885e-06,
+      "loss": 1.6739,
+      "step": 34330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.611633147667843e-06,
+      "loss": 2.0615,
+      "step": 34340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.607381266210299e-06,
+      "loss": 1.6146,
+      "step": 34350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.603129384752754e-06,
+      "loss": 2.1332,
+      "step": 34360
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.598877503295208e-06,
+      "loss": 1.9554,
+      "step": 34370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.594625621837664e-06,
+      "loss": 2.0376,
+      "step": 34380
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.590373740380119e-06,
+      "loss": 2.5308,
+      "step": 34390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.586121858922574e-06,
+      "loss": 2.0548,
+      "step": 34400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.581869977465028e-06,
+      "loss": 1.9477,
+      "step": 34410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.577618096007483e-06,
+      "loss": 2.1934,
+      "step": 34420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.573366214549939e-06,
+      "loss": 2.3392,
+      "step": 34430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5691143330923935e-06,
+      "loss": 2.3634,
+      "step": 34440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.56486245163485e-06,
+      "loss": 1.9666,
+      "step": 34450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.560610570177304e-06,
+      "loss": 2.2298,
+      "step": 34460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5563586887197585e-06,
+      "loss": 1.9168,
+      "step": 34470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.552106807262214e-06,
+      "loss": 1.6325,
+      "step": 34480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.547854925804669e-06,
+      "loss": 2.262,
+      "step": 34490
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5436030443471245e-06,
+      "loss": 1.6855,
+      "step": 34500
+    },
+    {
+      "epoch": 1.45,
+      "eval_loss": 1.7510865926742554,
+      "eval_runtime": 176.082,
+      "eval_samples_per_second": 14.215,
+      "eval_steps_per_second": 7.11,
+      "step": 34500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.539351162889579e-06,
+      "loss": 2.4015,
+      "step": 34510
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.535099281432033e-06,
+      "loss": 2.3955,
+      "step": 34520
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.5308473999744895e-06,
+      "loss": 2.1583,
+      "step": 34530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.526595518516944e-06,
+      "loss": 2.2425,
+      "step": 34540
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.522343637059399e-06,
+      "loss": 2.2618,
+      "step": 34550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.518091755601855e-06,
+      "loss": 1.7956,
+      "step": 34560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.513839874144309e-06,
+      "loss": 2.1824,
+      "step": 34570
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.509587992686764e-06,
+      "loss": 2.0588,
+      "step": 34580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.505336111229219e-06,
+      "loss": 2.1407,
+      "step": 34590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.501084229771675e-06,
+      "loss": 1.9659,
+      "step": 34600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4968323483141294e-06,
+      "loss": 2.2308,
+      "step": 34610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.492580466856584e-06,
+      "loss": 2.0814,
+      "step": 34620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.48832858539904e-06,
+      "loss": 2.0365,
+      "step": 34630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4840767039414945e-06,
+      "loss": 2.7061,
+      "step": 34640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.47982482248395e-06,
+      "loss": 2.011,
+      "step": 34650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.475572941026404e-06,
+      "loss": 2.0643,
+      "step": 34660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4713210595688596e-06,
+      "loss": 1.9561,
+      "step": 34670
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.467069178111315e-06,
+      "loss": 2.2013,
+      "step": 34680
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.462817296653769e-06,
+      "loss": 2.221,
+      "step": 34690
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4585654151962255e-06,
+      "loss": 2.2442,
+      "step": 34700
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.45431353373868e-06,
+      "loss": 2.303,
+      "step": 34710
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.450061652281134e-06,
+      "loss": 2.4457,
+      "step": 34720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4458097708235906e-06,
+      "loss": 2.1348,
+      "step": 34730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.441557889366045e-06,
+      "loss": 2.097,
+      "step": 34740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4373060079085e-06,
+      "loss": 2.2935,
+      "step": 34750
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.433054126450955e-06,
+      "loss": 2.3818,
+      "step": 34760
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.42880224499341e-06,
+      "loss": 2.4383,
+      "step": 34770
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.424550363535865e-06,
+      "loss": 2.2461,
+      "step": 34780
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.42029848207832e-06,
+      "loss": 2.4632,
+      "step": 34790
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.416046600620774e-06,
+      "loss": 2.1457,
+      "step": 34800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.4117947191632305e-06,
+      "loss": 2.0426,
+      "step": 34810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.407542837705685e-06,
+      "loss": 2.1143,
+      "step": 34820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.40329095624814e-06,
+      "loss": 2.0542,
+      "step": 34830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.3990390747905955e-06,
+      "loss": 1.601,
+      "step": 34840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.39478719333305e-06,
+      "loss": 1.88,
+      "step": 34850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.390535311875505e-06,
+      "loss": 1.8141,
+      "step": 34860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.38628343041796e-06,
+      "loss": 1.9591,
+      "step": 34870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.382031548960416e-06,
+      "loss": 2.2129,
+      "step": 34880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.37777966750287e-06,
+      "loss": 2.1054,
+      "step": 34890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.373527786045325e-06,
+      "loss": 2.0846,
+      "step": 34900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.369275904587781e-06,
+      "loss": 2.2735,
+      "step": 34910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.3650240231302354e-06,
+      "loss": 1.8749,
+      "step": 34920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.360772141672691e-06,
+      "loss": 2.1951,
+      "step": 34930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.356520260215145e-06,
+      "loss": 2.1941,
+      "step": 34940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.3522683787576005e-06,
+      "loss": 2.3601,
+      "step": 34950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.348016497300056e-06,
+      "loss": 1.3729,
+      "step": 34960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.34376461584251e-06,
+      "loss": 2.1726,
+      "step": 34970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.339512734384966e-06,
+      "loss": 1.9625,
+      "step": 34980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.335260852927421e-06,
+      "loss": 1.9156,
+      "step": 34990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.331008971469875e-06,
+      "loss": 2.1513,
+      "step": 35000
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 1.7616750001907349,
+      "eval_runtime": 175.2829,
+      "eval_samples_per_second": 14.28,
+      "eval_steps_per_second": 7.143,
+      "step": 35000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.3267570900123315e-06,
+      "loss": 2.07,
+      "step": 35010
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.322505208554786e-06,
+      "loss": 2.4348,
+      "step": 35020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.318253327097241e-06,
+      "loss": 2.5164,
+      "step": 35030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.314001445639696e-06,
+      "loss": 2.2986,
+      "step": 35040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.309749564182151e-06,
+      "loss": 2.2002,
+      "step": 35050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.305497682724606e-06,
+      "loss": 2.5061,
+      "step": 35060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.301245801267061e-06,
+      "loss": 2.4472,
+      "step": 35070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.296993919809517e-06,
+      "loss": 2.0103,
+      "step": 35080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.292742038351971e-06,
+      "loss": 2.0856,
+      "step": 35090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.288490156894426e-06,
+      "loss": 2.0458,
+      "step": 35100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.284238275436881e-06,
+      "loss": 2.1446,
+      "step": 35110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.2799863939793365e-06,
+      "loss": 1.7831,
+      "step": 35120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.275734512521792e-06,
+      "loss": 2.1475,
+      "step": 35130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.271482631064246e-06,
+      "loss": 2.0982,
+      "step": 35140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.267230749606701e-06,
+      "loss": 1.8416,
+      "step": 35150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.262978868149157e-06,
+      "loss": 1.9284,
+      "step": 35160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.258726986691611e-06,
+      "loss": 2.3603,
+      "step": 35170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.254475105234067e-06,
+      "loss": 2.4798,
+      "step": 35180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.250223223776522e-06,
+      "loss": 2.2247,
+      "step": 35190
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.245971342318976e-06,
+      "loss": 2.3189,
+      "step": 35200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.241719460861432e-06,
+      "loss": 2.1824,
+      "step": 35210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.237467579403886e-06,
+      "loss": 2.3432,
+      "step": 35220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.233215697946342e-06,
+      "loss": 2.0543,
+      "step": 35230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.228963816488797e-06,
+      "loss": 1.8353,
+      "step": 35240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.224711935031251e-06,
+      "loss": 2.5007,
+      "step": 35250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.220460053573707e-06,
+      "loss": 2.0003,
+      "step": 35260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.216208172116162e-06,
+      "loss": 2.3392,
+      "step": 35270
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.211956290658617e-06,
+      "loss": 2.0539,
+      "step": 35280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.207704409201072e-06,
+      "loss": 2.1073,
+      "step": 35290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.203452527743527e-06,
+      "loss": 2.4152,
+      "step": 35300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.199200646285982e-06,
+      "loss": 1.3909,
+      "step": 35310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.194948764828437e-06,
+      "loss": 2.1223,
+      "step": 35320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.190696883370891e-06,
+      "loss": 2.0514,
+      "step": 35330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.186445001913347e-06,
+      "loss": 2.0323,
+      "step": 35340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.182193120455802e-06,
+      "loss": 2.138,
+      "step": 35350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.177941238998258e-06,
+      "loss": 2.0379,
+      "step": 35360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.173689357540712e-06,
+      "loss": 2.1295,
+      "step": 35370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.169437476083167e-06,
+      "loss": 1.7035,
+      "step": 35380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.165185594625622e-06,
+      "loss": 2.0688,
+      "step": 35390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.160933713168077e-06,
+      "loss": 2.0947,
+      "step": 35400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.156681831710533e-06,
+      "loss": 1.8796,
+      "step": 35410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.152429950252987e-06,
+      "loss": 2.1555,
+      "step": 35420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.148178068795442e-06,
+      "loss": 2.1799,
+      "step": 35430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.143926187337898e-06,
+      "loss": 2.5292,
+      "step": 35440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.139674305880352e-06,
+      "loss": 2.3218,
+      "step": 35450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.1354224244228075e-06,
+      "loss": 1.4921,
+      "step": 35460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.131170542965263e-06,
+      "loss": 2.1965,
+      "step": 35470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.126918661507717e-06,
+      "loss": 2.4084,
+      "step": 35480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.122666780050173e-06,
+      "loss": 2.4196,
+      "step": 35490
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.118414898592627e-06,
+      "loss": 1.9576,
+      "step": 35500
+    },
+    {
+      "epoch": 1.49,
+      "eval_loss": 1.747657299041748,
+      "eval_runtime": 176.1258,
+      "eval_samples_per_second": 14.211,
+      "eval_steps_per_second": 7.109,
+      "step": 35500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.114163017135083e-06,
+      "loss": 2.3121,
+      "step": 35510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.109911135677538e-06,
+      "loss": 2.4393,
+      "step": 35520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.105659254219992e-06,
+      "loss": 2.4803,
+      "step": 35530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.101407372762448e-06,
+      "loss": 2.0533,
+      "step": 35540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.097155491304903e-06,
+      "loss": 2.0527,
+      "step": 35550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.092903609847358e-06,
+      "loss": 2.2304,
+      "step": 35560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0886517283898125e-06,
+      "loss": 2.1561,
+      "step": 35570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.084399846932268e-06,
+      "loss": 2.2062,
+      "step": 35580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.080147965474723e-06,
+      "loss": 2.0185,
+      "step": 35590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0758960840171776e-06,
+      "loss": 2.4212,
+      "step": 35600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.071644202559634e-06,
+      "loss": 2.05,
+      "step": 35610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.067392321102088e-06,
+      "loss": 1.958,
+      "step": 35620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.063140439644543e-06,
+      "loss": 2.0309,
+      "step": 35630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.058888558186999e-06,
+      "loss": 1.5439,
+      "step": 35640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.054636676729453e-06,
+      "loss": 2.072,
+      "step": 35650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0503847952719086e-06,
+      "loss": 2.22,
+      "step": 35660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.046132913814363e-06,
+      "loss": 1.9343,
+      "step": 35670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.041881032356818e-06,
+      "loss": 2.2888,
+      "step": 35680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.037629150899274e-06,
+      "loss": 1.9664,
+      "step": 35690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.033377269441728e-06,
+      "loss": 2.4235,
+      "step": 35700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.029125387984184e-06,
+      "loss": 2.2587,
+      "step": 35710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.024873506526639e-06,
+      "loss": 2.1917,
+      "step": 35720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.020621625069093e-06,
+      "loss": 2.0016,
+      "step": 35730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0163697436115485e-06,
+      "loss": 2.4665,
+      "step": 35740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.012117862154004e-06,
+      "loss": 2.1793,
+      "step": 35750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.007865980696459e-06,
+      "loss": 2.0033,
+      "step": 35760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0036140992389135e-06,
+      "loss": 1.8352,
+      "step": 35770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.999362217781369e-06,
+      "loss": 2.1996,
+      "step": 35780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.995110336323823e-06,
+      "loss": 1.9625,
+      "step": 35790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.990858454866279e-06,
+      "loss": 1.7395,
+      "step": 35800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.986606573408734e-06,
+      "loss": 2.5111,
+      "step": 35810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.982354691951189e-06,
+      "loss": 2.0511,
+      "step": 35820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.978102810493644e-06,
+      "loss": 2.3589,
+      "step": 35830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.973850929036099e-06,
+      "loss": 1.9662,
+      "step": 35840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9695990475785534e-06,
+      "loss": 1.9488,
+      "step": 35850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.965347166121009e-06,
+      "loss": 2.0706,
+      "step": 35860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.961095284663464e-06,
+      "loss": 2.1325,
+      "step": 35870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.956843403205919e-06,
+      "loss": 1.6996,
+      "step": 35880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.952591521748374e-06,
+      "loss": 2.2667,
+      "step": 35890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.948339640290829e-06,
+      "loss": 1.6537,
+      "step": 35900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.944087758833284e-06,
+      "loss": 2.1472,
+      "step": 35910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.93983587737574e-06,
+      "loss": 1.527,
+      "step": 35920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.935583995918194e-06,
+      "loss": 2.4018,
+      "step": 35930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9313321144606495e-06,
+      "loss": 2.0018,
+      "step": 35940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.927080233003104e-06,
+      "loss": 1.7936,
+      "step": 35950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.922828351545559e-06,
+      "loss": 2.4642,
+      "step": 35960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9185764700880146e-06,
+      "loss": 2.4036,
+      "step": 35970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.914324588630469e-06,
+      "loss": 1.8644,
+      "step": 35980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.910072707172924e-06,
+      "loss": 2.0751,
+      "step": 35990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.90582082571538e-06,
+      "loss": 2.0943,
+      "step": 36000
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 1.7490991353988647,
+      "eval_runtime": 175.4591,
+      "eval_samples_per_second": 14.265,
+      "eval_steps_per_second": 7.136,
+      "step": 36000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.901568944257835e-06,
+      "loss": 2.0113,
+      "step": 36010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.897317062800289e-06,
+      "loss": 1.8348,
+      "step": 36020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.893065181342745e-06,
+      "loss": 2.6659,
+      "step": 36030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.888813299885199e-06,
+      "loss": 2.4203,
+      "step": 36040
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.8845614184276545e-06,
+      "loss": 2.3977,
+      "step": 36050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.88030953697011e-06,
+      "loss": 2.0347,
+      "step": 36060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.876057655512565e-06,
+      "loss": 1.9492,
+      "step": 36070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.8718057740550195e-06,
+      "loss": 1.8611,
+      "step": 36080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.867553892597475e-06,
+      "loss": 1.6959,
+      "step": 36090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.86330201113993e-06,
+      "loss": 1.7844,
+      "step": 36100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.859050129682385e-06,
+      "loss": 2.132,
+      "step": 36110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.85479824822484e-06,
+      "loss": 2.2562,
+      "step": 36120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.850546366767294e-06,
+      "loss": 1.8299,
+      "step": 36130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.84629448530975e-06,
+      "loss": 2.1926,
+      "step": 36140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.842042603852205e-06,
+      "loss": 1.9032,
+      "step": 36150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.83779072239466e-06,
+      "loss": 2.3041,
+      "step": 36160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.833538840937116e-06,
+      "loss": 2.111,
+      "step": 36170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.82928695947957e-06,
+      "loss": 2.187,
+      "step": 36180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.825035078022025e-06,
+      "loss": 2.0693,
+      "step": 36190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.820783196564481e-06,
+      "loss": 1.9847,
+      "step": 36200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.816531315106935e-06,
+      "loss": 2.1018,
+      "step": 36210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.8122794336493896e-06,
+      "loss": 2.1058,
+      "step": 36220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.808027552191845e-06,
+      "loss": 1.9707,
+      "step": 36230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.8037756707343e-06,
+      "loss": 1.8293,
+      "step": 36240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7995237892767555e-06,
+      "loss": 2.3156,
+      "step": 36250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.795271907819211e-06,
+      "loss": 1.685,
+      "step": 36260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.791020026361665e-06,
+      "loss": 1.6848,
+      "step": 36270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7867681449041206e-06,
+      "loss": 1.7966,
+      "step": 36280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.782516263446576e-06,
+      "loss": 2.0305,
+      "step": 36290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.77826438198903e-06,
+      "loss": 1.9382,
+      "step": 36300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.774012500531486e-06,
+      "loss": 2.0167,
+      "step": 36310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.76976061907394e-06,
+      "loss": 1.9864,
+      "step": 36320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.765508737616395e-06,
+      "loss": 2.4167,
+      "step": 36330
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.761256856158851e-06,
+      "loss": 2.4879,
+      "step": 36340
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.757004974701306e-06,
+      "loss": 1.9834,
+      "step": 36350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.752753093243761e-06,
+      "loss": 2.2833,
+      "step": 36360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.748501211786216e-06,
+      "loss": 1.9415,
+      "step": 36370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.744249330328671e-06,
+      "loss": 2.3043,
+      "step": 36380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7399974488711255e-06,
+      "loss": 1.8226,
+      "step": 36390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.735745567413581e-06,
+      "loss": 2.1448,
+      "step": 36400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.731493685956036e-06,
+      "loss": 2.5569,
+      "step": 36410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.727241804498491e-06,
+      "loss": 2.1507,
+      "step": 36420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.722989923040946e-06,
+      "loss": 1.9932,
+      "step": 36430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.718738041583401e-06,
+      "loss": 2.0433,
+      "step": 36440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.7144861601258565e-06,
+      "loss": 2.4259,
+      "step": 36450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.710234278668311e-06,
+      "loss": 2.389,
+      "step": 36460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.705982397210766e-06,
+      "loss": 2.0994,
+      "step": 36470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.701730515753221e-06,
+      "loss": 2.5288,
+      "step": 36480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.697478634295676e-06,
+      "loss": 2.4069,
+      "step": 36490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.693226752838131e-06,
+      "loss": 1.669,
+      "step": 36500
+    },
+    {
+      "epoch": 1.54,
+      "eval_loss": 1.754565954208374,
+      "eval_runtime": 175.7316,
+      "eval_samples_per_second": 14.243,
+      "eval_steps_per_second": 7.125,
+      "step": 36500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.688974871380586e-06,
+      "loss": 2.0982,
+      "step": 36510
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.684722989923041e-06,
+      "loss": 2.4972,
+      "step": 36520
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.680471108465496e-06,
+      "loss": 2.5963,
+      "step": 36530
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.676219227007952e-06,
+      "loss": 2.0617,
+      "step": 36540
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.671967345550407e-06,
+      "loss": 1.9341,
+      "step": 36550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6677154640928615e-06,
+      "loss": 2.0048,
+      "step": 36560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.663463582635317e-06,
+      "loss": 2.2699,
+      "step": 36570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.659211701177771e-06,
+      "loss": 1.9895,
+      "step": 36580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6549598197202265e-06,
+      "loss": 2.3447,
+      "step": 36590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.650707938262682e-06,
+      "loss": 1.969,
+      "step": 36600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.646456056805136e-06,
+      "loss": 2.2418,
+      "step": 36610
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.642204175347592e-06,
+      "loss": 2.2551,
+      "step": 36620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.637952293890047e-06,
+      "loss": 1.6052,
+      "step": 36630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.633700412432502e-06,
+      "loss": 1.8645,
+      "step": 36640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.629448530974957e-06,
+      "loss": 2.3318,
+      "step": 36650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.625196649517412e-06,
+      "loss": 2.1373,
+      "step": 36660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6209447680598665e-06,
+      "loss": 2.3078,
+      "step": 36670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.616692886602322e-06,
+      "loss": 2.3501,
+      "step": 36680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.612441005144777e-06,
+      "loss": 1.6993,
+      "step": 36690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.6081891236872315e-06,
+      "loss": 1.8396,
+      "step": 36700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.603937242229687e-06,
+      "loss": 2.2297,
+      "step": 36710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.599685360772142e-06,
+      "loss": 1.9559,
+      "step": 36720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5954334793145974e-06,
+      "loss": 2.153,
+      "step": 36730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.591181597857052e-06,
+      "loss": 2.2837,
+      "step": 36740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.586929716399507e-06,
+      "loss": 1.6944,
+      "step": 36750
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.582677834941962e-06,
+      "loss": 1.879,
+      "step": 36760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.578425953484417e-06,
+      "loss": 1.9212,
+      "step": 36770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.574174072026872e-06,
+      "loss": 2.3385,
+      "step": 36780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5699221905693276e-06,
+      "loss": 2.0329,
+      "step": 36790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.565670309111782e-06,
+      "loss": 2.0217,
+      "step": 36800
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.561418427654237e-06,
+      "loss": 2.2787,
+      "step": 36810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.557166546196693e-06,
+      "loss": 1.9281,
+      "step": 36820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.552914664739148e-06,
+      "loss": 2.0044,
+      "step": 36830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.548662783281602e-06,
+      "loss": 2.0486,
+      "step": 36840
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.544410901824058e-06,
+      "loss": 2.3855,
+      "step": 36850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.540159020366512e-06,
+      "loss": 1.7565,
+      "step": 36860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5359071389089675e-06,
+      "loss": 2.4534,
+      "step": 36870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.531655257451423e-06,
+      "loss": 1.9945,
+      "step": 36880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.527403375993878e-06,
+      "loss": 2.3056,
+      "step": 36890
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.5231514945363325e-06,
+      "loss": 2.261,
+      "step": 36900
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.518899613078788e-06,
+      "loss": 1.9996,
+      "step": 36910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.514647731621243e-06,
+      "loss": 2.1015,
+      "step": 36920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.510395850163698e-06,
+      "loss": 2.1077,
+      "step": 36930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.506143968706153e-06,
+      "loss": 1.9946,
+      "step": 36940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.501892087248607e-06,
+      "loss": 2.2828,
+      "step": 36950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.497640205791063e-06,
+      "loss": 2.2453,
+      "step": 36960
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.493388324333518e-06,
+      "loss": 2.032,
+      "step": 36970
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.489136442875973e-06,
+      "loss": 2.162,
+      "step": 36980
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.484884561418428e-06,
+      "loss": 2.1001,
+      "step": 36990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.480632679960883e-06,
+      "loss": 2.1434,
+      "step": 37000
+    },
+    {
+      "epoch": 1.56,
+      "eval_loss": 1.7376418113708496,
+      "eval_runtime": 176.1379,
+      "eval_samples_per_second": 14.21,
+      "eval_steps_per_second": 7.108,
+      "step": 37000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.476380798503338e-06,
+      "loss": 2.0959,
+      "step": 37010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.472128917045793e-06,
+      "loss": 2.4057,
+      "step": 37020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.467877035588248e-06,
+      "loss": 2.2582,
+      "step": 37030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.463625154130703e-06,
+      "loss": 1.7448,
+      "step": 37040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.459373272673158e-06,
+      "loss": 2.0855,
+      "step": 37050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.455121391215613e-06,
+      "loss": 2.2708,
+      "step": 37060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4508695097580685e-06,
+      "loss": 2.0224,
+      "step": 37070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.446617628300524e-06,
+      "loss": 2.2547,
+      "step": 37080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.442365746842978e-06,
+      "loss": 1.9236,
+      "step": 37090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4381138653854336e-06,
+      "loss": 1.7799,
+      "step": 37100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.433861983927889e-06,
+      "loss": 2.1988,
+      "step": 37110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.429610102470343e-06,
+      "loss": 2.3449,
+      "step": 37120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.425358221012799e-06,
+      "loss": 2.2117,
+      "step": 37130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.421106339555253e-06,
+      "loss": 2.1477,
+      "step": 37140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.416854458097708e-06,
+      "loss": 2.2606,
+      "step": 37150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.412602576640164e-06,
+      "loss": 1.779,
+      "step": 37160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.408350695182619e-06,
+      "loss": 1.9888,
+      "step": 37170
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.404098813725074e-06,
+      "loss": 2.2328,
+      "step": 37180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.399846932267529e-06,
+      "loss": 2.0528,
+      "step": 37190
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.395595050809984e-06,
+      "loss": 1.7668,
+      "step": 37200
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3913431693524385e-06,
+      "loss": 1.8782,
+      "step": 37210
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.387091287894894e-06,
+      "loss": 2.1357,
+      "step": 37220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.382839406437348e-06,
+      "loss": 2.2905,
+      "step": 37230
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.378587524979804e-06,
+      "loss": 2.0164,
+      "step": 37240
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.374335643522259e-06,
+      "loss": 2.2011,
+      "step": 37250
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.370083762064714e-06,
+      "loss": 2.1939,
+      "step": 37260
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3658318806071695e-06,
+      "loss": 1.9964,
+      "step": 37270
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.361579999149624e-06,
+      "loss": 1.7766,
+      "step": 37280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.357328117692079e-06,
+      "loss": 2.1014,
+      "step": 37290
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.353076236234534e-06,
+      "loss": 2.0096,
+      "step": 37300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.348824354776989e-06,
+      "loss": 2.3916,
+      "step": 37310
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.344572473319444e-06,
+      "loss": 2.0797,
+      "step": 37320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.340320591861899e-06,
+      "loss": 2.3945,
+      "step": 37330
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.336068710404354e-06,
+      "loss": 1.627,
+      "step": 37340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3318168289468094e-06,
+      "loss": 2.3023,
+      "step": 37350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.327564947489265e-06,
+      "loss": 1.9037,
+      "step": 37360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.32331306603172e-06,
+      "loss": 1.7088,
+      "step": 37370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3190611845741745e-06,
+      "loss": 1.7849,
+      "step": 37380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.314809303116629e-06,
+      "loss": 1.8289,
+      "step": 37390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.310557421659084e-06,
+      "loss": 2.3683,
+      "step": 37400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3063055402015396e-06,
+      "loss": 2.1933,
+      "step": 37410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.302053658743995e-06,
+      "loss": 2.2794,
+      "step": 37420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.297801777286449e-06,
+      "loss": 2.0994,
+      "step": 37430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.293549895828905e-06,
+      "loss": 2.2796,
+      "step": 37440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.28929801437136e-06,
+      "loss": 1.8149,
+      "step": 37450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.285046132913815e-06,
+      "loss": 2.1684,
+      "step": 37460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.28079425145627e-06,
+      "loss": 1.8682,
+      "step": 37470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.276542369998725e-06,
+      "loss": 1.9193,
+      "step": 37480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2722904885411795e-06,
+      "loss": 2.0892,
+      "step": 37490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.268038607083635e-06,
+      "loss": 2.1553,
+      "step": 37500
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 1.7463552951812744,
+      "eval_runtime": 176.0244,
+      "eval_samples_per_second": 14.22,
+      "eval_steps_per_second": 7.113,
+      "step": 37500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.26378672562609e-06,
+      "loss": 1.6898,
+      "step": 37510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2595348441685445e-06,
+      "loss": 2.1169,
+      "step": 37520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.255282962711e-06,
+      "loss": 2.3435,
+      "step": 37530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.251031081253455e-06,
+      "loss": 2.0221,
+      "step": 37540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2467791997959105e-06,
+      "loss": 1.7718,
+      "step": 37550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.242527318338365e-06,
+      "loss": 2.4137,
+      "step": 37560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.23827543688082e-06,
+      "loss": 1.5029,
+      "step": 37570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.234023555423275e-06,
+      "loss": 2.308,
+      "step": 37580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.22977167396573e-06,
+      "loss": 2.1589,
+      "step": 37590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.225519792508185e-06,
+      "loss": 2.1282,
+      "step": 37600
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.221267911050641e-06,
+      "loss": 2.0603,
+      "step": 37610
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.217016029593095e-06,
+      "loss": 2.3884,
+      "step": 37620
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.21276414813555e-06,
+      "loss": 2.3481,
+      "step": 37630
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.208512266678006e-06,
+      "loss": 2.3606,
+      "step": 37640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.20426038522046e-06,
+      "loss": 2.076,
+      "step": 37650
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2000085037629154e-06,
+      "loss": 1.9552,
+      "step": 37660
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.19575662230537e-06,
+      "loss": 2.5605,
+      "step": 37670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.191504740847825e-06,
+      "loss": 2.4783,
+      "step": 37680
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1872528593902805e-06,
+      "loss": 1.8744,
+      "step": 37690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.183000977932736e-06,
+      "loss": 1.8266,
+      "step": 37700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.17874909647519e-06,
+      "loss": 1.784,
+      "step": 37710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1744972150176456e-06,
+      "loss": 1.855,
+      "step": 37720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.170245333560101e-06,
+      "loss": 1.9852,
+      "step": 37730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.165993452102556e-06,
+      "loss": 1.9374,
+      "step": 37740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.161741570645011e-06,
+      "loss": 2.0234,
+      "step": 37750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.157489689187466e-06,
+      "loss": 1.7482,
+      "step": 37760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.15323780772992e-06,
+      "loss": 2.1098,
+      "step": 37770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.148985926272376e-06,
+      "loss": 2.4678,
+      "step": 37780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.144734044814831e-06,
+      "loss": 2.0931,
+      "step": 37790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.140482163357286e-06,
+      "loss": 2.2251,
+      "step": 37800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.136230281899741e-06,
+      "loss": 2.319,
+      "step": 37810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.131978400442196e-06,
+      "loss": 2.2545,
+      "step": 37820
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.127726518984651e-06,
+      "loss": 2.139,
+      "step": 37830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.123474637527106e-06,
+      "loss": 2.2262,
+      "step": 37840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.119222756069561e-06,
+      "loss": 2.0406,
+      "step": 37850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.114970874612016e-06,
+      "loss": 1.7894,
+      "step": 37860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.110718993154471e-06,
+      "loss": 2.4756,
+      "step": 37870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.106467111696926e-06,
+      "loss": 2.2345,
+      "step": 37880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.1022152302393815e-06,
+      "loss": 2.3746,
+      "step": 37890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.097963348781837e-06,
+      "loss": 1.7428,
+      "step": 37900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.093711467324291e-06,
+      "loss": 2.3988,
+      "step": 37910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.089459585866747e-06,
+      "loss": 1.9973,
+      "step": 37920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.085207704409201e-06,
+      "loss": 2.0614,
+      "step": 37930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.080955822951656e-06,
+      "loss": 1.8093,
+      "step": 37940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.076703941494112e-06,
+      "loss": 2.2133,
+      "step": 37950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.072452060036566e-06,
+      "loss": 1.9063,
+      "step": 37960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0682001785790214e-06,
+      "loss": 2.6421,
+      "step": 37970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.063948297121477e-06,
+      "loss": 2.2117,
+      "step": 37980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.059696415663932e-06,
+      "loss": 2.1171,
+      "step": 37990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0554445342063865e-06,
+      "loss": 2.06,
+      "step": 38000
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 1.745969533920288,
+      "eval_runtime": 175.921,
+      "eval_samples_per_second": 14.228,
+      "eval_steps_per_second": 7.117,
+      "step": 38000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.051192652748842e-06,
+      "loss": 1.6237,
+      "step": 38010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.046940771291297e-06,
+      "loss": 2.3635,
+      "step": 38020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0426888898337516e-06,
+      "loss": 2.0552,
+      "step": 38030
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.038437008376207e-06,
+      "loss": 2.4618,
+      "step": 38040
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.034185126918661e-06,
+      "loss": 2.0934,
+      "step": 38050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.029933245461117e-06,
+      "loss": 1.8939,
+      "step": 38060
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.025681364003572e-06,
+      "loss": 1.8378,
+      "step": 38070
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.021429482546027e-06,
+      "loss": 1.5897,
+      "step": 38080
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.0171776010884826e-06,
+      "loss": 2.397,
+      "step": 38090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.012925719630937e-06,
+      "loss": 2.1118,
+      "step": 38100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.008673838173392e-06,
+      "loss": 2.2938,
+      "step": 38110
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.004421956715847e-06,
+      "loss": 2.3173,
+      "step": 38120
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.000170075258302e-06,
+      "loss": 2.3025,
+      "step": 38130
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.995918193800757e-06,
+      "loss": 2.0702,
+      "step": 38140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.991666312343212e-06,
+      "loss": 2.1073,
+      "step": 38150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.987414430885667e-06,
+      "loss": 2.7411,
+      "step": 38160
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9831625494281225e-06,
+      "loss": 2.1455,
+      "step": 38170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.978910667970578e-06,
+      "loss": 1.9867,
+      "step": 38180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.974658786513032e-06,
+      "loss": 2.1154,
+      "step": 38190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9704069050554875e-06,
+      "loss": 2.1478,
+      "step": 38200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.966155023597942e-06,
+      "loss": 1.744,
+      "step": 38210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961903142140397e-06,
+      "loss": 2.3651,
+      "step": 38220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957651260682853e-06,
+      "loss": 2.4052,
+      "step": 38230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.953399379225307e-06,
+      "loss": 1.8449,
+      "step": 38240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.949147497767762e-06,
+      "loss": 1.8959,
+      "step": 38250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.944895616310218e-06,
+      "loss": 1.8913,
+      "step": 38260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.940643734852673e-06,
+      "loss": 2.0492,
+      "step": 38270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.936391853395128e-06,
+      "loss": 2.2655,
+      "step": 38280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.932139971937583e-06,
+      "loss": 2.3627,
+      "step": 38290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.927888090480037e-06,
+      "loss": 2.2161,
+      "step": 38300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9236362090224925e-06,
+      "loss": 2.0692,
+      "step": 38310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.919384327564948e-06,
+      "loss": 1.6689,
+      "step": 38320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.915132446107403e-06,
+      "loss": 2.0331,
+      "step": 38330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9108805646498576e-06,
+      "loss": 2.0686,
+      "step": 38340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.906628683192313e-06,
+      "loss": 2.5842,
+      "step": 38350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.902376801734768e-06,
+      "loss": 2.0441,
+      "step": 38360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8981249202772235e-06,
+      "loss": 2.0406,
+      "step": 38370
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.893873038819678e-06,
+      "loss": 2.2905,
+      "step": 38380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.889621157362133e-06,
+      "loss": 1.9497,
+      "step": 38390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.885369275904588e-06,
+      "loss": 1.7564,
+      "step": 38400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.881117394447043e-06,
+      "loss": 1.895,
+      "step": 38410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.876865512989498e-06,
+      "loss": 2.4167,
+      "step": 38420
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.872613631531954e-06,
+      "loss": 2.4465,
+      "step": 38430
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.868361750074408e-06,
+      "loss": 2.1046,
+      "step": 38440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.864109868616863e-06,
+      "loss": 2.1757,
+      "step": 38450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.859857987159319e-06,
+      "loss": 2.6139,
+      "step": 38460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.855606105701773e-06,
+      "loss": 2.2124,
+      "step": 38470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8513542242442285e-06,
+      "loss": 2.0222,
+      "step": 38480
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.847102342786683e-06,
+      "loss": 2.1389,
+      "step": 38490
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.842850461329138e-06,
+      "loss": 2.2977,
+      "step": 38500
+    },
+    {
+      "epoch": 1.62,
+      "eval_loss": 1.7398837804794312,
+      "eval_runtime": 174.2372,
+      "eval_samples_per_second": 14.365,
+      "eval_steps_per_second": 7.186,
+      "step": 38500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8385985798715935e-06,
+      "loss": 2.4658,
+      "step": 38510
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.834346698414049e-06,
+      "loss": 1.8469,
+      "step": 38520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.830094816956503e-06,
+      "loss": 2.4372,
+      "step": 38530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.825842935498959e-06,
+      "loss": 1.6964,
+      "step": 38540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.821591054041414e-06,
+      "loss": 1.8705,
+      "step": 38550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.817339172583868e-06,
+      "loss": 2.2879,
+      "step": 38560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.813087291126324e-06,
+      "loss": 2.1464,
+      "step": 38570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8088354096687785e-06,
+      "loss": 1.8558,
+      "step": 38580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.804583528211234e-06,
+      "loss": 2.0456,
+      "step": 38590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8003316467536887e-06,
+      "loss": 2.2801,
+      "step": 38600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.796079765296144e-06,
+      "loss": 1.9854,
+      "step": 38610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.791827883838599e-06,
+      "loss": 2.5356,
+      "step": 38620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.787576002381054e-06,
+      "loss": 2.319,
+      "step": 38630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7833241209235087e-06,
+      "loss": 1.9153,
+      "step": 38640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.779072239465964e-06,
+      "loss": 1.8595,
+      "step": 38650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7748203580084193e-06,
+      "loss": 2.1707,
+      "step": 38660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.770568476550874e-06,
+      "loss": 1.9023,
+      "step": 38670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.766316595093329e-06,
+      "loss": 2.3358,
+      "step": 38680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.762064713635784e-06,
+      "loss": 2.0008,
+      "step": 38690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7578128321782392e-06,
+      "loss": 1.6161,
+      "step": 38700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7535609507206945e-06,
+      "loss": 1.9317,
+      "step": 38710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7493090692631494e-06,
+      "loss": 1.9281,
+      "step": 38720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7450571878056043e-06,
+      "loss": 2.2011,
+      "step": 38730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.740805306348059e-06,
+      "loss": 2.0454,
+      "step": 38740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7365534248905145e-06,
+      "loss": 2.2612,
+      "step": 38750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7323015434329694e-06,
+      "loss": 2.2228,
+      "step": 38760
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7280496619754243e-06,
+      "loss": 2.2572,
+      "step": 38770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.723797780517879e-06,
+      "loss": 1.7262,
+      "step": 38780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7195458990603345e-06,
+      "loss": 1.8567,
+      "step": 38790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7152940176027898e-06,
+      "loss": 2.0451,
+      "step": 38800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7110421361452446e-06,
+      "loss": 2.3198,
+      "step": 38810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7067902546876995e-06,
+      "loss": 2.1443,
+      "step": 38820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.7025383732301544e-06,
+      "loss": 1.7047,
+      "step": 38830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6982864917726097e-06,
+      "loss": 2.2287,
+      "step": 38840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.694034610315065e-06,
+      "loss": 2.5186,
+      "step": 38850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.68978272885752e-06,
+      "loss": 1.7028,
+      "step": 38860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6855308473999748e-06,
+      "loss": 2.2628,
+      "step": 38870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6812789659424297e-06,
+      "loss": 2.1352,
+      "step": 38880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.677027084484885e-06,
+      "loss": 2.2716,
+      "step": 38890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.67277520302734e-06,
+      "loss": 2.4243,
+      "step": 38900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.668523321569795e-06,
+      "loss": 2.1119,
+      "step": 38910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6642714401122496e-06,
+      "loss": 1.5971,
+      "step": 38920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.660019558654705e-06,
+      "loss": 2.2422,
+      "step": 38930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6557676771971602e-06,
+      "loss": 2.1503,
+      "step": 38940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.651515795739615e-06,
+      "loss": 2.3962,
+      "step": 38950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6472639142820704e-06,
+      "loss": 1.8617,
+      "step": 38960
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.643012032824525e-06,
+      "loss": 1.9732,
+      "step": 38970
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.63876015136698e-06,
+      "loss": 2.1593,
+      "step": 38980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6345082699094355e-06,
+      "loss": 1.8509,
+      "step": 38990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6302563884518904e-06,
+      "loss": 2.1659,
+      "step": 39000
+    },
+    {
+      "epoch": 1.64,
+      "eval_loss": 1.7403854131698608,
+      "eval_runtime": 175.8224,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 7.121,
+      "step": 39000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.626004506994345e-06,
+      "loss": 2.571,
+      "step": 39010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6217526255368e-06,
+      "loss": 2.0228,
+      "step": 39020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6175007440792554e-06,
+      "loss": 2.1392,
+      "step": 39030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6132488626217103e-06,
+      "loss": 2.2451,
+      "step": 39040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6089969811641656e-06,
+      "loss": 2.3339,
+      "step": 39050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.60474509970662e-06,
+      "loss": 2.0396,
+      "step": 39060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.6004932182490754e-06,
+      "loss": 2.571,
+      "step": 39070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5962413367915307e-06,
+      "loss": 2.0907,
+      "step": 39080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.5919894553339856e-06,
+      "loss": 2.1145,
+      "step": 39090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.587737573876441e-06,
+      "loss": 1.8426,
+      "step": 39100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5834856924188953e-06,
+      "loss": 1.8662,
+      "step": 39110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5792338109613506e-06,
+      "loss": 1.8386,
+      "step": 39120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.574981929503806e-06,
+      "loss": 1.9898,
+      "step": 39130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.570730048046261e-06,
+      "loss": 1.7737,
+      "step": 39140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.566478166588716e-06,
+      "loss": 2.8594,
+      "step": 39150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5622262851311706e-06,
+      "loss": 2.5668,
+      "step": 39160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.557974403673626e-06,
+      "loss": 2.1744,
+      "step": 39170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5537225222160808e-06,
+      "loss": 2.2878,
+      "step": 39180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.549470640758536e-06,
+      "loss": 1.7849,
+      "step": 39190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5452187593009914e-06,
+      "loss": 1.9586,
+      "step": 39200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.540966877843446e-06,
+      "loss": 2.1255,
+      "step": 39210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.536714996385901e-06,
+      "loss": 2.091,
+      "step": 39220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.532463114928356e-06,
+      "loss": 2.0458,
+      "step": 39230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5282112334708113e-06,
+      "loss": 2.0761,
+      "step": 39240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.523959352013266e-06,
+      "loss": 2.0046,
+      "step": 39250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.519707470555721e-06,
+      "loss": 2.2256,
+      "step": 39260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.515455589098176e-06,
+      "loss": 2.3956,
+      "step": 39270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5112037076406313e-06,
+      "loss": 1.9632,
+      "step": 39280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.5069518261830866e-06,
+      "loss": 1.9887,
+      "step": 39290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.502699944725541e-06,
+      "loss": 2.1451,
+      "step": 39300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4984480632679964e-06,
+      "loss": 2.2727,
+      "step": 39310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4941961818104512e-06,
+      "loss": 1.9085,
+      "step": 39320
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.4899443003529065e-06,
+      "loss": 2.1372,
+      "step": 39330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.485692418895362e-06,
+      "loss": 2.0505,
+      "step": 39340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4814405374378163e-06,
+      "loss": 2.2831,
+      "step": 39350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4771886559802716e-06,
+      "loss": 1.6421,
+      "step": 39360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4729367745227265e-06,
+      "loss": 1.5879,
+      "step": 39370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.468684893065182e-06,
+      "loss": 2.1306,
+      "step": 39380
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.464433011607637e-06,
+      "loss": 2.0623,
+      "step": 39390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4601811301500916e-06,
+      "loss": 1.9673,
+      "step": 39400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4559292486925464e-06,
+      "loss": 1.9179,
+      "step": 39410
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4516773672350018e-06,
+      "loss": 2.3303,
+      "step": 39420
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.447425485777457e-06,
+      "loss": 1.8826,
+      "step": 39430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.443173604319912e-06,
+      "loss": 2.3108,
+      "step": 39440
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.438921722862367e-06,
+      "loss": 2.4321,
+      "step": 39450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4346698414048217e-06,
+      "loss": 2.2305,
+      "step": 39460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.430417959947277e-06,
+      "loss": 2.3232,
+      "step": 39470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4261660784897323e-06,
+      "loss": 2.3438,
+      "step": 39480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4219141970321868e-06,
+      "loss": 2.0287,
+      "step": 39490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.417662315574642e-06,
+      "loss": 1.6394,
+      "step": 39500
+    },
+    {
+      "epoch": 1.66,
+      "eval_loss": 1.739558219909668,
+      "eval_runtime": 176.2216,
+      "eval_samples_per_second": 14.204,
+      "eval_steps_per_second": 7.105,
+      "step": 39500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.413410434117097e-06,
+      "loss": 1.9423,
+      "step": 39510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.4091585526595523e-06,
+      "loss": 2.2243,
+      "step": 39520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.404906671202007e-06,
+      "loss": 2.1207,
+      "step": 39530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.400654789744462e-06,
+      "loss": 2.2572,
+      "step": 39540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.396402908286917e-06,
+      "loss": 2.3871,
+      "step": 39550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3921510268293722e-06,
+      "loss": 2.1014,
+      "step": 39560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3878991453718275e-06,
+      "loss": 2.235,
+      "step": 39570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3836472639142824e-06,
+      "loss": 2.0478,
+      "step": 39580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3793953824567373e-06,
+      "loss": 2.2684,
+      "step": 39590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.375143500999192e-06,
+      "loss": 2.2871,
+      "step": 39600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3708916195416475e-06,
+      "loss": 2.3627,
+      "step": 39610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3666397380841028e-06,
+      "loss": 2.2261,
+      "step": 39620
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3623878566265577e-06,
+      "loss": 1.8954,
+      "step": 39630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3581359751690125e-06,
+      "loss": 1.9103,
+      "step": 39640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3538840937114674e-06,
+      "loss": 2.0318,
+      "step": 39650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3496322122539227e-06,
+      "loss": 2.5308,
+      "step": 39660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3453803307963776e-06,
+      "loss": 2.1057,
+      "step": 39670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.341128449338833e-06,
+      "loss": 1.9694,
+      "step": 39680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3368765678812874e-06,
+      "loss": 2.1225,
+      "step": 39690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3326246864237427e-06,
+      "loss": 1.6829,
+      "step": 39700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.328372804966198e-06,
+      "loss": 1.8024,
+      "step": 39710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.324120923508653e-06,
+      "loss": 2.2156,
+      "step": 39720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.319869042051108e-06,
+      "loss": 2.145,
+      "step": 39730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3156171605935626e-06,
+      "loss": 1.945,
+      "step": 39740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.311365279136018e-06,
+      "loss": 1.4955,
+      "step": 39750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3071133976784732e-06,
+      "loss": 1.5375,
+      "step": 39760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.302861516220928e-06,
+      "loss": 1.9554,
+      "step": 39770
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.298609634763383e-06,
+      "loss": 1.4847,
+      "step": 39780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.294357753305838e-06,
+      "loss": 1.9126,
+      "step": 39790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.290105871848293e-06,
+      "loss": 2.4683,
+      "step": 39800
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.285853990390748e-06,
+      "loss": 2.5539,
+      "step": 39810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2816021089332034e-06,
+      "loss": 2.2314,
+      "step": 39820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.277350227475658e-06,
+      "loss": 2.4671,
+      "step": 39830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.273098346018113e-06,
+      "loss": 2.0932,
+      "step": 39840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2688464645605685e-06,
+      "loss": 1.9059,
+      "step": 39850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2645945831030233e-06,
+      "loss": 1.9942,
+      "step": 39860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2603427016454786e-06,
+      "loss": 2.345,
+      "step": 39870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.256090820187933e-06,
+      "loss": 1.9271,
+      "step": 39880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2518389387303884e-06,
+      "loss": 2.1968,
+      "step": 39890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2475870572728437e-06,
+      "loss": 2.1401,
+      "step": 39900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2433351758152986e-06,
+      "loss": 1.9934,
+      "step": 39910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.239083294357754e-06,
+      "loss": 1.9551,
+      "step": 39920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2348314129002084e-06,
+      "loss": 1.7779,
+      "step": 39930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2305795314426637e-06,
+      "loss": 2.2155,
+      "step": 39940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2263276499851185e-06,
+      "loss": 2.4332,
+      "step": 39950
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.222075768527574e-06,
+      "loss": 2.0219,
+      "step": 39960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.217823887070029e-06,
+      "loss": 1.7372,
+      "step": 39970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2135720056124836e-06,
+      "loss": 2.3289,
+      "step": 39980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.209320124154939e-06,
+      "loss": 2.1965,
+      "step": 39990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.205068242697394e-06,
+      "loss": 2.1019,
+      "step": 40000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 1.736899495124817,
+      "eval_runtime": 176.1277,
+      "eval_samples_per_second": 14.211,
+      "eval_steps_per_second": 7.108,
+      "step": 40000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.200816361239849e-06,
+      "loss": 2.2945,
+      "step": 40010
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.1965644797823036e-06,
+      "loss": 1.9417,
+      "step": 40020
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.192312598324759e-06,
+      "loss": 2.1428,
+      "step": 40030
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.188060716867214e-06,
+      "loss": 2.4317,
+      "step": 40040
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.183808835409669e-06,
+      "loss": 2.1994,
+      "step": 40050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1795569539521244e-06,
+      "loss": 1.6647,
+      "step": 40060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.175305072494579e-06,
+      "loss": 2.3434,
+      "step": 40070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.171053191037034e-06,
+      "loss": 1.9678,
+      "step": 40080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.166801309579489e-06,
+      "loss": 2.0424,
+      "step": 40090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1625494281219443e-06,
+      "loss": 1.8569,
+      "step": 40100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1582975466643996e-06,
+      "loss": 1.929,
+      "step": 40110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.154045665206854e-06,
+      "loss": 1.7804,
+      "step": 40120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1497937837493094e-06,
+      "loss": 2.2012,
+      "step": 40130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1455419022917643e-06,
+      "loss": 2.3599,
+      "step": 40140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1412900208342196e-06,
+      "loss": 2.2751,
+      "step": 40150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.137038139376675e-06,
+      "loss": 2.0306,
+      "step": 40160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1327862579191293e-06,
+      "loss": 1.9452,
+      "step": 40170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1285343764615842e-06,
+      "loss": 2.5462,
+      "step": 40180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1242824950040395e-06,
+      "loss": 1.9403,
+      "step": 40190
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.120030613546495e-06,
+      "loss": 1.5562,
+      "step": 40200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1157787320889497e-06,
+      "loss": 1.988,
+      "step": 40210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1115268506314046e-06,
+      "loss": 1.9707,
+      "step": 40220
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1072749691738595e-06,
+      "loss": 2.1308,
+      "step": 40230
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1030230877163148e-06,
+      "loss": 2.4308,
+      "step": 40240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.09877120625877e-06,
+      "loss": 2.0105,
+      "step": 40250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0945193248012245e-06,
+      "loss": 1.8751,
+      "step": 40260
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.09026744334368e-06,
+      "loss": 1.9968,
+      "step": 40270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0860155618861347e-06,
+      "loss": 1.909,
+      "step": 40280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.08176368042859e-06,
+      "loss": 1.3458,
+      "step": 40290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0775117989710453e-06,
+      "loss": 2.4988,
+      "step": 40300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0732599175135e-06,
+      "loss": 2.0061,
+      "step": 40310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0690080360559547e-06,
+      "loss": 2.3653,
+      "step": 40320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.06475615459841e-06,
+      "loss": 2.0347,
+      "step": 40330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0605042731408653e-06,
+      "loss": 2.1352,
+      "step": 40340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.05625239168332e-06,
+      "loss": 2.0918,
+      "step": 40350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.052000510225775e-06,
+      "loss": 1.8342,
+      "step": 40360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.04774862876823e-06,
+      "loss": 2.0742,
+      "step": 40370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0434967473106852e-06,
+      "loss": 2.4816,
+      "step": 40380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0392448658531405e-06,
+      "loss": 2.3585,
+      "step": 40390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0349929843955954e-06,
+      "loss": 2.0589,
+      "step": 40400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0307411029380503e-06,
+      "loss": 1.9231,
+      "step": 40410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.026489221480505e-06,
+      "loss": 1.6204,
+      "step": 40420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0222373400229605e-06,
+      "loss": 1.6995,
+      "step": 40430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0179854585654154e-06,
+      "loss": 2.1029,
+      "step": 40440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0137335771078707e-06,
+      "loss": 2.1059,
+      "step": 40450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.009481695650325e-06,
+      "loss": 2.1145,
+      "step": 40460
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0052298141927804e-06,
+      "loss": 1.5657,
+      "step": 40470
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0009779327352358e-06,
+      "loss": 1.6706,
+      "step": 40480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9967260512776906e-06,
+      "loss": 2.0865,
+      "step": 40490
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.992474169820146e-06,
+      "loss": 1.6778,
+      "step": 40500
+    },
+    {
+      "epoch": 1.7,
+      "eval_loss": 1.747696876525879,
+      "eval_runtime": 175.7241,
+      "eval_samples_per_second": 14.244,
+      "eval_steps_per_second": 7.125,
+      "step": 40500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9882222883626004e-06,
+      "loss": 2.2589,
+      "step": 40510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9839704069050557e-06,
+      "loss": 1.9951,
+      "step": 40520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.979718525447511e-06,
+      "loss": 2.075,
+      "step": 40530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.975466643989966e-06,
+      "loss": 2.4767,
+      "step": 40540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9712147625324208e-06,
+      "loss": 2.0628,
+      "step": 40550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9669628810748757e-06,
+      "loss": 2.1439,
+      "step": 40560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.962710999617331e-06,
+      "loss": 2.5834,
+      "step": 40570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.958459118159786e-06,
+      "loss": 1.9482,
+      "step": 40580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.954207236702241e-06,
+      "loss": 2.0323,
+      "step": 40590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9499553552446956e-06,
+      "loss": 1.833,
+      "step": 40600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.945703473787151e-06,
+      "loss": 1.7037,
+      "step": 40610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9414515923296062e-06,
+      "loss": 2.0919,
+      "step": 40620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.937199710872061e-06,
+      "loss": 2.0546,
+      "step": 40630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9329478294145164e-06,
+      "loss": 2.1635,
+      "step": 40640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.928695947956971e-06,
+      "loss": 2.1947,
+      "step": 40650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.924444066499426e-06,
+      "loss": 2.3327,
+      "step": 40660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9201921850418815e-06,
+      "loss": 1.9751,
+      "step": 40670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9159403035843364e-06,
+      "loss": 2.0792,
+      "step": 40680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9116884221267917e-06,
+      "loss": 2.0507,
+      "step": 40690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.907436540669246e-06,
+      "loss": 2.0128,
+      "step": 40700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.9031846592117014e-06,
+      "loss": 2.0428,
+      "step": 40710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8989327777541563e-06,
+      "loss": 2.24,
+      "step": 40720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8946808962966116e-06,
+      "loss": 2.4284,
+      "step": 40730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.890429014839067e-06,
+      "loss": 2.2595,
+      "step": 40740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8861771333815214e-06,
+      "loss": 1.9509,
+      "step": 40750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8819252519239767e-06,
+      "loss": 1.7006,
+      "step": 40760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8776733704664316e-06,
+      "loss": 2.0867,
+      "step": 40770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.873421489008887e-06,
+      "loss": 1.9262,
+      "step": 40780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8691696075513413e-06,
+      "loss": 1.9245,
+      "step": 40790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8649177260937966e-06,
+      "loss": 2.0081,
+      "step": 40800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.860665844636252e-06,
+      "loss": 2.3462,
+      "step": 40810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.856413963178707e-06,
+      "loss": 1.8247,
+      "step": 40820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.852162081721162e-06,
+      "loss": 2.1083,
+      "step": 40830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8479102002636166e-06,
+      "loss": 2.024,
+      "step": 40840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.843658318806072e-06,
+      "loss": 2.2802,
+      "step": 40850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8394064373485268e-06,
+      "loss": 2.3956,
+      "step": 40860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.835154555890982e-06,
+      "loss": 2.0755,
+      "step": 40870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8309026744334374e-06,
+      "loss": 2.1163,
+      "step": 40880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.826650792975892e-06,
+      "loss": 2.0023,
+      "step": 40890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.822398911518347e-06,
+      "loss": 2.0573,
+      "step": 40900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.818147030060802e-06,
+      "loss": 1.7237,
+      "step": 40910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8138951486032573e-06,
+      "loss": 2.2315,
+      "step": 40920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8096432671457126e-06,
+      "loss": 2.4632,
+      "step": 40930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.805391385688167e-06,
+      "loss": 2.3754,
+      "step": 40940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8011395042306224e-06,
+      "loss": 1.8312,
+      "step": 40950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7968876227730773e-06,
+      "loss": 2.0822,
+      "step": 40960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7926357413155326e-06,
+      "loss": 1.9008,
+      "step": 40970
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7883838598579875e-06,
+      "loss": 1.6843,
+      "step": 40980
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7841319784004424e-06,
+      "loss": 2.1599,
+      "step": 40990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.7798800969428972e-06,
+      "loss": 2.0825,
+      "step": 41000
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 1.7327312231063843,
+      "eval_runtime": 175.8219,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 7.121,
+      "step": 41000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7756282154853525e-06,
+      "loss": 1.9554,
+      "step": 41010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.771376334027808e-06,
+      "loss": 2.1843,
+      "step": 41020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7671244525702623e-06,
+      "loss": 2.2919,
+      "step": 41030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7628725711127176e-06,
+      "loss": 2.1009,
+      "step": 41040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7586206896551725e-06,
+      "loss": 1.9775,
+      "step": 41050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.754368808197628e-06,
+      "loss": 2.2389,
+      "step": 41060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.750116926740083e-06,
+      "loss": 2.0114,
+      "step": 41070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7458650452825376e-06,
+      "loss": 2.124,
+      "step": 41080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7416131638249924e-06,
+      "loss": 2.2319,
+      "step": 41090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7373612823674478e-06,
+      "loss": 2.1945,
+      "step": 41100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.733109400909903e-06,
+      "loss": 2.0162,
+      "step": 41110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.728857519452358e-06,
+      "loss": 2.0864,
+      "step": 41120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.724605637994813e-06,
+      "loss": 2.4334,
+      "step": 41130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7203537565372677e-06,
+      "loss": 2.4696,
+      "step": 41140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.716101875079723e-06,
+      "loss": 2.141,
+      "step": 41150
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.7118499936221783e-06,
+      "loss": 2.658,
+      "step": 41160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.707598112164633e-06,
+      "loss": 2.2342,
+      "step": 41170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.703346230707088e-06,
+      "loss": 1.8181,
+      "step": 41180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.699094349249543e-06,
+      "loss": 2.1224,
+      "step": 41190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6948424677919983e-06,
+      "loss": 2.8241,
+      "step": 41200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6905905863344536e-06,
+      "loss": 2.0384,
+      "step": 41210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.6863387048769085e-06,
+      "loss": 2.2898,
+      "step": 41220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.682086823419363e-06,
+      "loss": 2.2362,
+      "step": 41230
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6778349419618182e-06,
+      "loss": 2.3073,
+      "step": 41240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6735830605042735e-06,
+      "loss": 2.0308,
+      "step": 41250
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6693311790467284e-06,
+      "loss": 2.4844,
+      "step": 41260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6650792975891833e-06,
+      "loss": 2.058,
+      "step": 41270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.660827416131638e-06,
+      "loss": 1.623,
+      "step": 41280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6565755346740935e-06,
+      "loss": 2.1991,
+      "step": 41290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6523236532165488e-06,
+      "loss": 2.2852,
+      "step": 41300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6480717717590037e-06,
+      "loss": 2.4844,
+      "step": 41310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6438198903014585e-06,
+      "loss": 1.9889,
+      "step": 41320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6395680088439134e-06,
+      "loss": 2.1942,
+      "step": 41330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6353161273863687e-06,
+      "loss": 2.0585,
+      "step": 41340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6310642459288236e-06,
+      "loss": 2.1481,
+      "step": 41350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.626812364471279e-06,
+      "loss": 2.0574,
+      "step": 41360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6225604830137334e-06,
+      "loss": 1.9496,
+      "step": 41370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6183086015561887e-06,
+      "loss": 2.2692,
+      "step": 41380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.614056720098644e-06,
+      "loss": 1.6283,
+      "step": 41390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.609804838641099e-06,
+      "loss": 1.812,
+      "step": 41400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.605552957183554e-06,
+      "loss": 2.2804,
+      "step": 41410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.6013010757260086e-06,
+      "loss": 1.9433,
+      "step": 41420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.597049194268464e-06,
+      "loss": 2.2863,
+      "step": 41430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5927973128109192e-06,
+      "loss": 1.9486,
+      "step": 41440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.588545431353374e-06,
+      "loss": 1.8636,
+      "step": 41450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5842935498958294e-06,
+      "loss": 2.1634,
+      "step": 41460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.580041668438284e-06,
+      "loss": 2.0405,
+      "step": 41470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.575789786980739e-06,
+      "loss": 1.8633,
+      "step": 41480
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.571537905523194e-06,
+      "loss": 1.7699,
+      "step": 41490
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5672860240656494e-06,
+      "loss": 2.5621,
+      "step": 41500
+    },
+    {
+      "epoch": 1.75,
+      "eval_loss": 1.7315673828125,
+      "eval_runtime": 175.8241,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 7.121,
+      "step": 41500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5630341426081047e-06,
+      "loss": 2.2221,
+      "step": 41510
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.558782261150559e-06,
+      "loss": 2.1501,
+      "step": 41520
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5545303796930144e-06,
+      "loss": 2.3411,
+      "step": 41530
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5502784982354693e-06,
+      "loss": 2.1032,
+      "step": 41540
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5460266167779246e-06,
+      "loss": 1.8093,
+      "step": 41550
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.541774735320379e-06,
+      "loss": 2.1871,
+      "step": 41560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5375228538628344e-06,
+      "loss": 2.5805,
+      "step": 41570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5332709724052897e-06,
+      "loss": 2.0588,
+      "step": 41580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5290190909477446e-06,
+      "loss": 2.0337,
+      "step": 41590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5247672094902e-06,
+      "loss": 1.6492,
+      "step": 41600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5205153280326544e-06,
+      "loss": 2.1416,
+      "step": 41610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5162634465751097e-06,
+      "loss": 2.5335,
+      "step": 41620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5120115651175645e-06,
+      "loss": 1.8925,
+      "step": 41630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.50775968366002e-06,
+      "loss": 1.8441,
+      "step": 41640
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.503507802202475e-06,
+      "loss": 2.11,
+      "step": 41650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.49925592074493e-06,
+      "loss": 2.0362,
+      "step": 41660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.495004039287385e-06,
+      "loss": 1.9229,
+      "step": 41670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.49075215782984e-06,
+      "loss": 2.3089,
+      "step": 41680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.486500276372295e-06,
+      "loss": 2.028,
+      "step": 41690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.48224839491475e-06,
+      "loss": 2.1496,
+      "step": 41700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4779965134572053e-06,
+      "loss": 1.9989,
+      "step": 41710
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.47374463199966e-06,
+      "loss": 1.616,
+      "step": 41720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.469492750542115e-06,
+      "loss": 1.8583,
+      "step": 41730
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.46524086908457e-06,
+      "loss": 2.1548,
+      "step": 41740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4609889876270252e-06,
+      "loss": 1.9858,
+      "step": 41750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.45673710616948e-06,
+      "loss": 2.3754,
+      "step": 41760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.452485224711935e-06,
+      "loss": 2.2034,
+      "step": 41770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4482333432543903e-06,
+      "loss": 2.0934,
+      "step": 41780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.443981461796845e-06,
+      "loss": 1.7186,
+      "step": 41790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4397295803393005e-06,
+      "loss": 2.0101,
+      "step": 41800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4354776988817554e-06,
+      "loss": 2.2067,
+      "step": 41810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4312258174242103e-06,
+      "loss": 2.0457,
+      "step": 41820
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4269739359666656e-06,
+      "loss": 2.1525,
+      "step": 41830
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4227220545091204e-06,
+      "loss": 2.1126,
+      "step": 41840
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4184701730515758e-06,
+      "loss": 1.9887,
+      "step": 41850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4142182915940306e-06,
+      "loss": 2.3211,
+      "step": 41860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4099664101364855e-06,
+      "loss": 1.9687,
+      "step": 41870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.405714528678941e-06,
+      "loss": 1.8773,
+      "step": 41880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4014626472213957e-06,
+      "loss": 1.8474,
+      "step": 41890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3972107657638506e-06,
+      "loss": 1.8167,
+      "step": 41900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3929588843063055e-06,
+      "loss": 2.4848,
+      "step": 41910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.3887070028487608e-06,
+      "loss": 2.096,
+      "step": 41920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.384455121391216e-06,
+      "loss": 1.7895,
+      "step": 41930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.380203239933671e-06,
+      "loss": 2.124,
+      "step": 41940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.375951358476126e-06,
+      "loss": 1.7622,
+      "step": 41950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3716994770185807e-06,
+      "loss": 2.5117,
+      "step": 41960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.367447595561036e-06,
+      "loss": 1.7705,
+      "step": 41970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.363195714103491e-06,
+      "loss": 2.0807,
+      "step": 41980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3589438326459462e-06,
+      "loss": 2.247,
+      "step": 41990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.354691951188401e-06,
+      "loss": 2.1884,
+      "step": 42000
+    },
+    {
+      "epoch": 1.77,
+      "eval_loss": 1.7364351749420166,
+      "eval_runtime": 174.6812,
+      "eval_samples_per_second": 14.329,
+      "eval_steps_per_second": 7.167,
+      "step": 42000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.350440069730856e-06,
+      "loss": 2.1428,
+      "step": 42010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3461881882733113e-06,
+      "loss": 2.2284,
+      "step": 42020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.341936306815766e-06,
+      "loss": 1.9592,
+      "step": 42030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.337684425358221e-06,
+      "loss": 1.9115,
+      "step": 42040
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3334325439006764e-06,
+      "loss": 2.2041,
+      "step": 42050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3291806624431312e-06,
+      "loss": 1.9779,
+      "step": 42060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3249287809855865e-06,
+      "loss": 1.9501,
+      "step": 42070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3206768995280414e-06,
+      "loss": 1.9092,
+      "step": 42080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3164250180704963e-06,
+      "loss": 2.0939,
+      "step": 42090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.312173136612951e-06,
+      "loss": 1.8454,
+      "step": 42100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.3079212551554065e-06,
+      "loss": 1.9993,
+      "step": 42110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.303669373697862e-06,
+      "loss": 2.185,
+      "step": 42120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2994174922403163e-06,
+      "loss": 2.3047,
+      "step": 42130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2951656107827716e-06,
+      "loss": 2.1145,
+      "step": 42140
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2909137293252264e-06,
+      "loss": 1.759,
+      "step": 42150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2866618478676818e-06,
+      "loss": 2.3151,
+      "step": 42160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2824099664101366e-06,
+      "loss": 2.107,
+      "step": 42170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.2781580849525915e-06,
+      "loss": 1.9285,
+      "step": 42180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.273906203495047e-06,
+      "loss": 1.7268,
+      "step": 42190
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2696543220375017e-06,
+      "loss": 2.0042,
+      "step": 42200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.265402440579957e-06,
+      "loss": 1.8875,
+      "step": 42210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.261150559122412e-06,
+      "loss": 2.0278,
+      "step": 42220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2568986776648668e-06,
+      "loss": 2.0565,
+      "step": 42230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.252646796207322e-06,
+      "loss": 2.0557,
+      "step": 42240
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.248394914749777e-06,
+      "loss": 2.3697,
+      "step": 42250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.244143033292232e-06,
+      "loss": 2.1159,
+      "step": 42260
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2398911518346867e-06,
+      "loss": 2.259,
+      "step": 42270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.235639270377142e-06,
+      "loss": 2.234,
+      "step": 42280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2313873889195973e-06,
+      "loss": 1.9082,
+      "step": 42290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2271355074620522e-06,
+      "loss": 1.8677,
+      "step": 42300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.222883626004507e-06,
+      "loss": 1.9178,
+      "step": 42310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.218631744546962e-06,
+      "loss": 2.0711,
+      "step": 42320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2143798630894173e-06,
+      "loss": 1.9575,
+      "step": 42330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2101279816318726e-06,
+      "loss": 2.1129,
+      "step": 42340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2058761001743275e-06,
+      "loss": 2.1437,
+      "step": 42350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.2016242187167824e-06,
+      "loss": 2.0994,
+      "step": 42360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1973723372592372e-06,
+      "loss": 1.968,
+      "step": 42370
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1931204558016925e-06,
+      "loss": 2.584,
+      "step": 42380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1888685743441474e-06,
+      "loss": 2.273,
+      "step": 42390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1846166928866023e-06,
+      "loss": 1.9261,
+      "step": 42400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1803648114290576e-06,
+      "loss": 2.1038,
+      "step": 42410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1761129299715125e-06,
+      "loss": 1.7815,
+      "step": 42420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.171861048513968e-06,
+      "loss": 1.9469,
+      "step": 42430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1676091670564227e-06,
+      "loss": 1.9804,
+      "step": 42440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1633572855988776e-06,
+      "loss": 1.8273,
+      "step": 42450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.159105404141333e-06,
+      "loss": 2.06,
+      "step": 42460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1548535226837877e-06,
+      "loss": 2.397,
+      "step": 42470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.150601641226243e-06,
+      "loss": 2.0802,
+      "step": 42480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.146349759768698e-06,
+      "loss": 1.9114,
+      "step": 42490
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.142097878311153e-06,
+      "loss": 1.9195,
+      "step": 42500
+    },
+    {
+      "epoch": 1.79,
+      "eval_loss": 1.7350587844848633,
+      "eval_runtime": 175.8431,
+      "eval_samples_per_second": 14.234,
+      "eval_steps_per_second": 7.12,
+      "step": 42500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1378459968536077e-06,
+      "loss": 1.8898,
+      "step": 42510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.133594115396063e-06,
+      "loss": 1.8898,
+      "step": 42520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.129342233938518e-06,
+      "loss": 1.8507,
+      "step": 42530
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1250903524809728e-06,
+      "loss": 2.0565,
+      "step": 42540
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.120838471023428e-06,
+      "loss": 2.2139,
+      "step": 42550
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.116586589565883e-06,
+      "loss": 2.3074,
+      "step": 42560
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1123347081083383e-06,
+      "loss": 1.7307,
+      "step": 42570
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.108082826650793e-06,
+      "loss": 2.0316,
+      "step": 42580
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.103830945193248e-06,
+      "loss": 2.7731,
+      "step": 42590
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0995790637357033e-06,
+      "loss": 1.7271,
+      "step": 42600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0953271822781582e-06,
+      "loss": 2.1797,
+      "step": 42610
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0910753008206135e-06,
+      "loss": 1.8662,
+      "step": 42620
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0868234193630684e-06,
+      "loss": 2.6484,
+      "step": 42630
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0825715379055233e-06,
+      "loss": 2.1295,
+      "step": 42640
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0783196564479786e-06,
+      "loss": 1.6942,
+      "step": 42650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0740677749904335e-06,
+      "loss": 2.3459,
+      "step": 42660
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0698158935328884e-06,
+      "loss": 2.0255,
+      "step": 42670
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0655640120753432e-06,
+      "loss": 2.4654,
+      "step": 42680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0613121306177985e-06,
+      "loss": 1.888,
+      "step": 42690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.057060249160254e-06,
+      "loss": 1.9747,
+      "step": 42700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0528083677027087e-06,
+      "loss": 1.6326,
+      "step": 42710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0485564862451636e-06,
+      "loss": 1.6464,
+      "step": 42720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0443046047876185e-06,
+      "loss": 2.2843,
+      "step": 42730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.040052723330074e-06,
+      "loss": 2.2483,
+      "step": 42740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0358008418725287e-06,
+      "loss": 2.1342,
+      "step": 42750
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.031548960414984e-06,
+      "loss": 1.8196,
+      "step": 42760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.027297078957439e-06,
+      "loss": 2.1734,
+      "step": 42770
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0230451974998937e-06,
+      "loss": 1.8203,
+      "step": 42780
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.018793316042349e-06,
+      "loss": 1.8205,
+      "step": 42790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.014541434584804e-06,
+      "loss": 1.9341,
+      "step": 42800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.010289553127259e-06,
+      "loss": 1.7776,
+      "step": 42810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.006037671669714e-06,
+      "loss": 2.1631,
+      "step": 42820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.001785790212169e-06,
+      "loss": 2.3703,
+      "step": 42830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9975339087546243e-06,
+      "loss": 2.4234,
+      "step": 42840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.993282027297079e-06,
+      "loss": 2.3221,
+      "step": 42850
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.989030145839534e-06,
+      "loss": 1.8707,
+      "step": 42860
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.984778264381989e-06,
+      "loss": 1.8046,
+      "step": 42870
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9805263829244443e-06,
+      "loss": 1.9571,
+      "step": 42880
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9762745014668996e-06,
+      "loss": 2.2652,
+      "step": 42890
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9720226200093544e-06,
+      "loss": 2.3992,
+      "step": 42900
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9677707385518093e-06,
+      "loss": 1.7401,
+      "step": 42910
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.963518857094264e-06,
+      "loss": 2.2501,
+      "step": 42920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9592669756367195e-06,
+      "loss": 1.6796,
+      "step": 42930
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9550150941791744e-06,
+      "loss": 2.0322,
+      "step": 42940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9507632127216293e-06,
+      "loss": 1.8914,
+      "step": 42950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9465113312640846e-06,
+      "loss": 2.4243,
+      "step": 42960
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9422594498065395e-06,
+      "loss": 2.4589,
+      "step": 42970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9380075683489948e-06,
+      "loss": 2.1706,
+      "step": 42980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9337556868914497e-06,
+      "loss": 1.8146,
+      "step": 42990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9295038054339045e-06,
+      "loss": 2.1325,
+      "step": 43000
+    },
+    {
+      "epoch": 1.81,
+      "eval_loss": 1.7335894107818604,
+      "eval_runtime": 176.1718,
+      "eval_samples_per_second": 14.208,
+      "eval_steps_per_second": 7.107,
+      "step": 43000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.92525192397636e-06,
+      "loss": 2.1366,
+      "step": 43010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9210000425188147e-06,
+      "loss": 2.2995,
+      "step": 43020
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.91674816106127e-06,
+      "loss": 2.0707,
+      "step": 43030
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9124962796037245e-06,
+      "loss": 2.3193,
+      "step": 43040
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.90824439814618e-06,
+      "loss": 2.2661,
+      "step": 43050
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.903992516688635e-06,
+      "loss": 2.4724,
+      "step": 43060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8997406352310898e-06,
+      "loss": 2.1197,
+      "step": 43070
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.895488753773545e-06,
+      "loss": 2.4334,
+      "step": 43080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.891236872316e-06,
+      "loss": 2.0957,
+      "step": 43090
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.886984990858455e-06,
+      "loss": 2.2178,
+      "step": 43100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.88273310940091e-06,
+      "loss": 2.0521,
+      "step": 43110
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.878481227943365e-06,
+      "loss": 2.0047,
+      "step": 43120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8742293464858201e-06,
+      "loss": 2.359,
+      "step": 43130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.869977465028275e-06,
+      "loss": 2.0012,
+      "step": 43140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8657255835707303e-06,
+      "loss": 1.5969,
+      "step": 43150
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8614737021131852e-06,
+      "loss": 2.1817,
+      "step": 43160
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8572218206556403e-06,
+      "loss": 2.6578,
+      "step": 43170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8529699391980954e-06,
+      "loss": 2.5675,
+      "step": 43180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8487180577405503e-06,
+      "loss": 2.0409,
+      "step": 43190
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8444661762830054e-06,
+      "loss": 1.6609,
+      "step": 43200
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8402142948254602e-06,
+      "loss": 2.0456,
+      "step": 43210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8359624133679155e-06,
+      "loss": 1.9981,
+      "step": 43220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8317105319103706e-06,
+      "loss": 2.3557,
+      "step": 43230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8274586504528255e-06,
+      "loss": 1.8564,
+      "step": 43240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8232067689952806e-06,
+      "loss": 2.1589,
+      "step": 43250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8189548875377355e-06,
+      "loss": 2.1792,
+      "step": 43260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8147030060801906e-06,
+      "loss": 2.0653,
+      "step": 43270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8104511246226455e-06,
+      "loss": 2.2793,
+      "step": 43280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8061992431651008e-06,
+      "loss": 1.7779,
+      "step": 43290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8019473617075559e-06,
+      "loss": 2.2637,
+      "step": 43300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7976954802500107e-06,
+      "loss": 1.8934,
+      "step": 43310
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7934435987924658e-06,
+      "loss": 1.9671,
+      "step": 43320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7891917173349207e-06,
+      "loss": 1.7957,
+      "step": 43330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7849398358773758e-06,
+      "loss": 2.1045,
+      "step": 43340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.7806879544198311e-06,
+      "loss": 2.2057,
+      "step": 43350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.776436072962286e-06,
+      "loss": 2.4658,
+      "step": 43360
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.772184191504741e-06,
+      "loss": 2.2629,
+      "step": 43370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.767932310047196e-06,
+      "loss": 2.2685,
+      "step": 43380
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.763680428589651e-06,
+      "loss": 1.9794,
+      "step": 43390
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.759428547132106e-06,
+      "loss": 2.2,
+      "step": 43400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.755176665674561e-06,
+      "loss": 2.2813,
+      "step": 43410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7509247842170164e-06,
+      "loss": 2.2738,
+      "step": 43420
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7466729027594712e-06,
+      "loss": 1.9699,
+      "step": 43430
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7424210213019263e-06,
+      "loss": 1.7798,
+      "step": 43440
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7381691398443812e-06,
+      "loss": 2.2218,
+      "step": 43450
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7339172583868363e-06,
+      "loss": 2.3333,
+      "step": 43460
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7296653769292914e-06,
+      "loss": 2.2299,
+      "step": 43470
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7254134954717463e-06,
+      "loss": 2.229,
+      "step": 43480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7211616140142016e-06,
+      "loss": 2.1926,
+      "step": 43490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7169097325566565e-06,
+      "loss": 1.9533,
+      "step": 43500
+    },
+    {
+      "epoch": 1.83,
+      "eval_loss": 1.737502098083496,
+      "eval_runtime": 175.9057,
+      "eval_samples_per_second": 14.229,
+      "eval_steps_per_second": 7.117,
+      "step": 43500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7126578510991116e-06,
+      "loss": 1.6707,
+      "step": 43510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7084059696415664e-06,
+      "loss": 1.8694,
+      "step": 43520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.7041540881840215e-06,
+      "loss": 2.3288,
+      "step": 43530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6999022067264766e-06,
+      "loss": 2.1223,
+      "step": 43540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6956503252689315e-06,
+      "loss": 1.8674,
+      "step": 43550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6913984438113868e-06,
+      "loss": 2.1238,
+      "step": 43560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6871465623538417e-06,
+      "loss": 1.9917,
+      "step": 43570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6828946808962968e-06,
+      "loss": 2.1347,
+      "step": 43580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6786427994387519e-06,
+      "loss": 1.9695,
+      "step": 43590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6743909179812068e-06,
+      "loss": 2.4367,
+      "step": 43600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.6701390365236619e-06,
+      "loss": 2.082,
+      "step": 43610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6658871550661167e-06,
+      "loss": 2.1691,
+      "step": 43620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.661635273608572e-06,
+      "loss": 2.0205,
+      "step": 43630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.657383392151027e-06,
+      "loss": 1.9344,
+      "step": 43640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.653131510693482e-06,
+      "loss": 1.7473,
+      "step": 43650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6488796292359371e-06,
+      "loss": 2.46,
+      "step": 43660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.644627747778392e-06,
+      "loss": 1.9346,
+      "step": 43670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.640375866320847e-06,
+      "loss": 1.8243,
+      "step": 43680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.636123984863302e-06,
+      "loss": 2.0834,
+      "step": 43690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6318721034057573e-06,
+      "loss": 2.1517,
+      "step": 43700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6276202219482124e-06,
+      "loss": 2.3432,
+      "step": 43710
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6233683404906673e-06,
+      "loss": 2.3242,
+      "step": 43720
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6191164590331224e-06,
+      "loss": 2.154,
+      "step": 43730
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6148645775755772e-06,
+      "loss": 2.5902,
+      "step": 43740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6106126961180323e-06,
+      "loss": 2.1923,
+      "step": 43750
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6063608146604872e-06,
+      "loss": 2.0517,
+      "step": 43760
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.6021089332029425e-06,
+      "loss": 1.5443,
+      "step": 43770
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5978570517453976e-06,
+      "loss": 2.1229,
+      "step": 43780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5936051702878525e-06,
+      "loss": 2.3246,
+      "step": 43790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5893532888303076e-06,
+      "loss": 2.3142,
+      "step": 43800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5851014073727625e-06,
+      "loss": 2.4578,
+      "step": 43810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5808495259152176e-06,
+      "loss": 1.636,
+      "step": 43820
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5765976444576729e-06,
+      "loss": 1.7894,
+      "step": 43830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5723457630001277e-06,
+      "loss": 2.3135,
+      "step": 43840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5680938815425828e-06,
+      "loss": 2.2809,
+      "step": 43850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5638420000850377e-06,
+      "loss": 2.3676,
+      "step": 43860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5595901186274928e-06,
+      "loss": 2.0304,
+      "step": 43870
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5553382371699477e-06,
+      "loss": 2.2491,
+      "step": 43880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5510863557124028e-06,
+      "loss": 2.1846,
+      "step": 43890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.546834474254858e-06,
+      "loss": 1.8177,
+      "step": 43900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5425825927973128e-06,
+      "loss": 2.2895,
+      "step": 43910
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.538330711339768e-06,
+      "loss": 1.8667,
+      "step": 43920
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.534078829882223e-06,
+      "loss": 2.2535,
+      "step": 43930
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.529826948424678e-06,
+      "loss": 2.3201,
+      "step": 43940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5255750669671331e-06,
+      "loss": 1.79,
+      "step": 43950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.521323185509588e-06,
+      "loss": 2.1128,
+      "step": 43960
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5170713040520433e-06,
+      "loss": 1.9392,
+      "step": 43970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.512819422594498e-06,
+      "loss": 1.7095,
+      "step": 43980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5085675411369533e-06,
+      "loss": 1.665,
+      "step": 43990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5043156596794082e-06,
+      "loss": 2.4102,
+      "step": 44000
+    },
+    {
+      "epoch": 1.85,
+      "eval_loss": 1.737618327140808,
+      "eval_runtime": 176.0317,
+      "eval_samples_per_second": 14.219,
+      "eval_steps_per_second": 7.112,
+      "step": 44000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5000637782218633e-06,
+      "loss": 1.961,
+      "step": 44010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4958118967643184e-06,
+      "loss": 1.9114,
+      "step": 44020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4915600153067733e-06,
+      "loss": 2.2823,
+      "step": 44030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4873081338492283e-06,
+      "loss": 2.166,
+      "step": 44040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4830562523916832e-06,
+      "loss": 2.2117,
+      "step": 44050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4788043709341385e-06,
+      "loss": 2.1801,
+      "step": 44060
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4745524894765936e-06,
+      "loss": 2.0113,
+      "step": 44070
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4703006080190485e-06,
+      "loss": 1.9797,
+      "step": 44080
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4660487265615036e-06,
+      "loss": 2.0611,
+      "step": 44090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4617968451039585e-06,
+      "loss": 2.1161,
+      "step": 44100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4575449636464136e-06,
+      "loss": 1.682,
+      "step": 44110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4532930821888689e-06,
+      "loss": 2.1025,
+      "step": 44120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4490412007313238e-06,
+      "loss": 1.86,
+      "step": 44130
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4447893192737789e-06,
+      "loss": 1.8424,
+      "step": 44140
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4405374378162337e-06,
+      "loss": 1.8724,
+      "step": 44150
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4362855563586888e-06,
+      "loss": 2.3283,
+      "step": 44160
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4320336749011437e-06,
+      "loss": 1.9669,
+      "step": 44170
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4277817934435988e-06,
+      "loss": 2.5114,
+      "step": 44180
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4235299119860541e-06,
+      "loss": 2.3339,
+      "step": 44190
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.419278030528509e-06,
+      "loss": 2.0356,
+      "step": 44200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.415026149070964e-06,
+      "loss": 2.1621,
+      "step": 44210
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.410774267613419e-06,
+      "loss": 1.8599,
+      "step": 44220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.406522386155874e-06,
+      "loss": 1.8576,
+      "step": 44230
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4022705046983292e-06,
+      "loss": 2.0705,
+      "step": 44240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.398018623240784e-06,
+      "loss": 1.9939,
+      "step": 44250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3937667417832394e-06,
+      "loss": 2.3555,
+      "step": 44260
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3895148603256942e-06,
+      "loss": 2.4827,
+      "step": 44270
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3852629788681493e-06,
+      "loss": 2.4176,
+      "step": 44280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3810110974106042e-06,
+      "loss": 2.4387,
+      "step": 44290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3767592159530593e-06,
+      "loss": 2.0681,
+      "step": 44300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3725073344955144e-06,
+      "loss": 1.8416,
+      "step": 44310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.3682554530379693e-06,
+      "loss": 2.4489,
+      "step": 44320
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3640035715804246e-06,
+      "loss": 2.2564,
+      "step": 44330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3597516901228795e-06,
+      "loss": 1.8832,
+      "step": 44340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3554998086653346e-06,
+      "loss": 2.1685,
+      "step": 44350
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3512479272077897e-06,
+      "loss": 2.0292,
+      "step": 44360
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3469960457502445e-06,
+      "loss": 1.9095,
+      "step": 44370
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3427441642926996e-06,
+      "loss": 2.0641,
+      "step": 44380
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3384922828351545e-06,
+      "loss": 2.5147,
+      "step": 44390
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3342404013776098e-06,
+      "loss": 2.2525,
+      "step": 44400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3299885199200647e-06,
+      "loss": 2.2836,
+      "step": 44410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3257366384625198e-06,
+      "loss": 2.2603,
+      "step": 44420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3214847570049749e-06,
+      "loss": 1.9764,
+      "step": 44430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3172328755474298e-06,
+      "loss": 2.1242,
+      "step": 44440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3129809940898849e-06,
+      "loss": 2.5716,
+      "step": 44450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3087291126323397e-06,
+      "loss": 2.1799,
+      "step": 44460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.304477231174795e-06,
+      "loss": 1.9316,
+      "step": 44470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.3002253497172501e-06,
+      "loss": 2.0683,
+      "step": 44480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.295973468259705e-06,
+      "loss": 2.4333,
+      "step": 44490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2917215868021601e-06,
+      "loss": 2.1635,
+      "step": 44500
+    },
+    {
+      "epoch": 1.87,
+      "eval_loss": 1.731579303741455,
+      "eval_runtime": 175.8675,
+      "eval_samples_per_second": 14.232,
+      "eval_steps_per_second": 7.119,
+      "step": 44500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.287469705344615e-06,
+      "loss": 2.051,
+      "step": 44510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.28321782388707e-06,
+      "loss": 2.4778,
+      "step": 44520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.278965942429525e-06,
+      "loss": 2.21,
+      "step": 44530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2747140609719803e-06,
+      "loss": 2.3338,
+      "step": 44540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2704621795144354e-06,
+      "loss": 1.7422,
+      "step": 44550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2662102980568903e-06,
+      "loss": 1.9787,
+      "step": 44560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2619584165993453e-06,
+      "loss": 1.9784,
+      "step": 44570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2577065351418002e-06,
+      "loss": 2.1287,
+      "step": 44580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2534546536842553e-06,
+      "loss": 2.0922,
+      "step": 44590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2492027722267104e-06,
+      "loss": 2.2547,
+      "step": 44600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2449508907691655e-06,
+      "loss": 1.8577,
+      "step": 44610
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2406990093116204e-06,
+      "loss": 1.823,
+      "step": 44620
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2364471278540755e-06,
+      "loss": 1.8993,
+      "step": 44630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2321952463965306e-06,
+      "loss": 1.6693,
+      "step": 44640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2279433649389857e-06,
+      "loss": 1.5397,
+      "step": 44650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2236914834814406e-06,
+      "loss": 2.1911,
+      "step": 44660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2194396020238957e-06,
+      "loss": 2.1,
+      "step": 44670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2151877205663507e-06,
+      "loss": 1.9287,
+      "step": 44680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2109358391088058e-06,
+      "loss": 2.3998,
+      "step": 44690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2066839576512607e-06,
+      "loss": 2.2794,
+      "step": 44700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2024320761937158e-06,
+      "loss": 2.3223,
+      "step": 44710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.198180194736171e-06,
+      "loss": 2.4034,
+      "step": 44720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1939283132786258e-06,
+      "loss": 2.3119,
+      "step": 44730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1896764318210809e-06,
+      "loss": 2.0402,
+      "step": 44740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.185424550363536e-06,
+      "loss": 1.5494,
+      "step": 44750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.181172668905991e-06,
+      "loss": 1.6709,
+      "step": 44760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.176920787448446e-06,
+      "loss": 1.8675,
+      "step": 44770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.172668905990901e-06,
+      "loss": 2.3726,
+      "step": 44780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1684170245333561e-06,
+      "loss": 2.1301,
+      "step": 44790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.164165143075811e-06,
+      "loss": 2.1072,
+      "step": 44800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1599132616182663e-06,
+      "loss": 1.9413,
+      "step": 44810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1556613801607212e-06,
+      "loss": 1.8215,
+      "step": 44820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1514094987031763e-06,
+      "loss": 2.2847,
+      "step": 44830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1471576172456312e-06,
+      "loss": 1.9615,
+      "step": 44840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1429057357880863e-06,
+      "loss": 2.334,
+      "step": 44850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1386538543305414e-06,
+      "loss": 2.1077,
+      "step": 44860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1344019728729965e-06,
+      "loss": 1.8033,
+      "step": 44870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1301500914154516e-06,
+      "loss": 2.0681,
+      "step": 44880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1258982099579064e-06,
+      "loss": 2.0213,
+      "step": 44890
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1216463285003615e-06,
+      "loss": 1.8618,
+      "step": 44900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1173944470428164e-06,
+      "loss": 1.9152,
+      "step": 44910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1131425655852715e-06,
+      "loss": 2.1678,
+      "step": 44920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1088906841277266e-06,
+      "loss": 2.3216,
+      "step": 44930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1046388026701817e-06,
+      "loss": 2.428,
+      "step": 44940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1003869212126366e-06,
+      "loss": 2.1514,
+      "step": 44950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0961350397550917e-06,
+      "loss": 1.9048,
+      "step": 44960
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0918831582975468e-06,
+      "loss": 2.1308,
+      "step": 44970
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0876312768400016e-06,
+      "loss": 1.7374,
+      "step": 44980
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.083379395382457e-06,
+      "loss": 2.0971,
+      "step": 44990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0791275139249118e-06,
+      "loss": 2.0447,
+      "step": 45000
+    },
+    {
+      "epoch": 1.89,
+      "eval_loss": 1.7354670763015747,
+      "eval_runtime": 175.9119,
+      "eval_samples_per_second": 14.229,
+      "eval_steps_per_second": 7.117,
+      "step": 45000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.074875632467367e-06,
+      "loss": 1.6593,
+      "step": 45010
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.0706237510098218e-06,
+      "loss": 1.9934,
+      "step": 45020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.066371869552277e-06,
+      "loss": 2.028,
+      "step": 45030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.062119988094732e-06,
+      "loss": 1.8805,
+      "step": 45040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.057868106637187e-06,
+      "loss": 2.2899,
+      "step": 45050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0536162251796422e-06,
+      "loss": 2.0075,
+      "step": 45060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.049364343722097e-06,
+      "loss": 1.8379,
+      "step": 45070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0451124622645522e-06,
+      "loss": 2.3873,
+      "step": 45080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.040860580807007e-06,
+      "loss": 2.2744,
+      "step": 45090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0366086993494623e-06,
+      "loss": 1.6716,
+      "step": 45100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0323568178919172e-06,
+      "loss": 2.1744,
+      "step": 45110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0281049364343723e-06,
+      "loss": 2.097,
+      "step": 45120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0238530549768274e-06,
+      "loss": 2.1333,
+      "step": 45130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0196011735192823e-06,
+      "loss": 1.8945,
+      "step": 45140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0153492920617374e-06,
+      "loss": 2.2689,
+      "step": 45150
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0110974106041925e-06,
+      "loss": 1.9336,
+      "step": 45160
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0068455291466476e-06,
+      "loss": 2.1484,
+      "step": 45170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0025936476891025e-06,
+      "loss": 1.8955,
+      "step": 45180
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.983417662315576e-07,
+      "loss": 1.9074,
+      "step": 45190
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.940898847740127e-07,
+      "loss": 1.9534,
+      "step": 45200
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.898380033164675e-07,
+      "loss": 2.0347,
+      "step": 45210
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.855861218589226e-07,
+      "loss": 1.9925,
+      "step": 45220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.813342404013777e-07,
+      "loss": 1.9415,
+      "step": 45230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.770823589438328e-07,
+      "loss": 2.0953,
+      "step": 45240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.728304774862877e-07,
+      "loss": 1.947,
+      "step": 45250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.685785960287428e-07,
+      "loss": 1.7225,
+      "step": 45260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.643267145711979e-07,
+      "loss": 2.1531,
+      "step": 45270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.60074833113653e-07,
+      "loss": 2.3571,
+      "step": 45280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.558229516561079e-07,
+      "loss": 1.992,
+      "step": 45290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.51571070198563e-07,
+      "loss": 1.9213,
+      "step": 45300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.473191887410179e-07,
+      "loss": 1.6985,
+      "step": 45310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.43067307283473e-07,
+      "loss": 2.0781,
+      "step": 45320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.38815425825928e-07,
+      "loss": 2.4186,
+      "step": 45330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.345635443683831e-07,
+      "loss": 1.806,
+      "step": 45340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.303116629108382e-07,
+      "loss": 2.1777,
+      "step": 45350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.260597814532932e-07,
+      "loss": 2.0575,
+      "step": 45360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.218078999957482e-07,
+      "loss": 2.0456,
+      "step": 45370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.175560185382032e-07,
+      "loss": 2.2842,
+      "step": 45380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.133041370806582e-07,
+      "loss": 2.0703,
+      "step": 45390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.090522556231134e-07,
+      "loss": 2.0194,
+      "step": 45400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.048003741655683e-07,
+      "loss": 2.5022,
+      "step": 45410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.005484927080233e-07,
+      "loss": 1.9706,
+      "step": 45420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.962966112504784e-07,
+      "loss": 2.0497,
+      "step": 45430
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.920447297929334e-07,
+      "loss": 2.2135,
+      "step": 45440
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.877928483353884e-07,
+      "loss": 2.0119,
+      "step": 45450
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.835409668778436e-07,
+      "loss": 1.981,
+      "step": 45460
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.792890854202986e-07,
+      "loss": 2.1487,
+      "step": 45470
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.750372039627536e-07,
+      "loss": 2.5634,
+      "step": 45480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.707853225052086e-07,
+      "loss": 2.3092,
+      "step": 45490
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.665334410476637e-07,
+      "loss": 2.1664,
+      "step": 45500
+    },
+    {
+      "epoch": 1.91,
+      "eval_loss": 1.7321257591247559,
+      "eval_runtime": 174.5157,
+      "eval_samples_per_second": 14.343,
+      "eval_steps_per_second": 7.174,
+      "step": 45500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 8.622815595901186e-07,
+      "loss": 1.8467,
+      "step": 45510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.580296781325737e-07,
+      "loss": 1.9513,
+      "step": 45520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.537777966750288e-07,
+      "loss": 1.8185,
+      "step": 45530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.495259152174838e-07,
+      "loss": 2.3288,
+      "step": 45540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.452740337599388e-07,
+      "loss": 1.966,
+      "step": 45550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.410221523023938e-07,
+      "loss": 2.037,
+      "step": 45560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.367702708448489e-07,
+      "loss": 1.9471,
+      "step": 45570
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.32518389387304e-07,
+      "loss": 2.0111,
+      "step": 45580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.28266507929759e-07,
+      "loss": 1.9612,
+      "step": 45590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.240146264722141e-07,
+      "loss": 1.5161,
+      "step": 45600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.197627450146691e-07,
+      "loss": 1.9682,
+      "step": 45610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.15510863557124e-07,
+      "loss": 2.0455,
+      "step": 45620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.11258982099579e-07,
+      "loss": 1.9825,
+      "step": 45630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.070071006420342e-07,
+      "loss": 1.9404,
+      "step": 45640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.027552191844892e-07,
+      "loss": 1.8732,
+      "step": 45650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.985033377269442e-07,
+      "loss": 2.3782,
+      "step": 45660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.942514562693993e-07,
+      "loss": 1.8175,
+      "step": 45670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.899995748118543e-07,
+      "loss": 2.3118,
+      "step": 45680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.857476933543093e-07,
+      "loss": 2.1101,
+      "step": 45690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.814958118967645e-07,
+      "loss": 1.8881,
+      "step": 45700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.772439304392195e-07,
+      "loss": 1.8899,
+      "step": 45710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.729920489816745e-07,
+      "loss": 2.336,
+      "step": 45720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.687401675241294e-07,
+      "loss": 2.294,
+      "step": 45730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.644882860665845e-07,
+      "loss": 2.6725,
+      "step": 45740
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.602364046090395e-07,
+      "loss": 2.3393,
+      "step": 45750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.559845231514946e-07,
+      "loss": 2.2539,
+      "step": 45760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.517326416939497e-07,
+      "loss": 1.9912,
+      "step": 45770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.474807602364047e-07,
+      "loss": 2.3202,
+      "step": 45780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.432288787788597e-07,
+      "loss": 1.7479,
+      "step": 45790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.389769973213147e-07,
+      "loss": 2.1202,
+      "step": 45800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.347251158637697e-07,
+      "loss": 1.2499,
+      "step": 45810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.304732344062249e-07,
+      "loss": 1.9322,
+      "step": 45820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.262213529486798e-07,
+      "loss": 2.1149,
+      "step": 45830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.219694714911348e-07,
+      "loss": 1.8979,
+      "step": 45840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.177175900335899e-07,
+      "loss": 2.0963,
+      "step": 45850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.134657085760449e-07,
+      "loss": 2.3248,
+      "step": 45860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.092138271184999e-07,
+      "loss": 2.2861,
+      "step": 45870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.049619456609551e-07,
+      "loss": 2.2305,
+      "step": 45880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.007100642034101e-07,
+      "loss": 1.9268,
+      "step": 45890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.964581827458651e-07,
+      "loss": 2.0687,
+      "step": 45900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.922063012883201e-07,
+      "loss": 2.2125,
+      "step": 45910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.879544198307752e-07,
+      "loss": 2.8913,
+      "step": 45920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.837025383732303e-07,
+      "loss": 1.9888,
+      "step": 45930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.794506569156852e-07,
+      "loss": 2.1703,
+      "step": 45940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.751987754581403e-07,
+      "loss": 2.3161,
+      "step": 45950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.709468940005953e-07,
+      "loss": 2.4341,
+      "step": 45960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.666950125430503e-07,
+      "loss": 2.1793,
+      "step": 45970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.624431310855053e-07,
+      "loss": 2.4448,
+      "step": 45980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.581912496279605e-07,
+      "loss": 2.4227,
+      "step": 45990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.539393681704155e-07,
+      "loss": 2.159,
+      "step": 46000
+    },
+    {
+      "epoch": 1.94,
+      "eval_loss": 1.7307124137878418,
+      "eval_runtime": 176.3216,
+      "eval_samples_per_second": 14.196,
+      "eval_steps_per_second": 7.101,
+      "step": 46000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.496874867128705e-07,
+      "loss": 2.0888,
+      "step": 46010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.454356052553256e-07,
+      "loss": 2.3548,
+      "step": 46020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.411837237977806e-07,
+      "loss": 2.0837,
+      "step": 46030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.369318423402355e-07,
+      "loss": 2.129,
+      "step": 46040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.326799608826907e-07,
+      "loss": 2.1136,
+      "step": 46050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.284280794251457e-07,
+      "loss": 1.82,
+      "step": 46060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.241761979676007e-07,
+      "loss": 2.4332,
+      "step": 46070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.199243165100557e-07,
+      "loss": 2.4913,
+      "step": 46080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.156724350525108e-07,
+      "loss": 2.1268,
+      "step": 46090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.114205535949658e-07,
+      "loss": 1.8976,
+      "step": 46100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.071686721374209e-07,
+      "loss": 2.7105,
+      "step": 46110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.029167906798759e-07,
+      "loss": 1.9377,
+      "step": 46120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.98664909222331e-07,
+      "loss": 1.6299,
+      "step": 46130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.94413027764786e-07,
+      "loss": 2.3748,
+      "step": 46140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901611463072409e-07,
+      "loss": 1.6503,
+      "step": 46150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85909264849696e-07,
+      "loss": 1.9663,
+      "step": 46160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.81657383392151e-07,
+      "loss": 2.0319,
+      "step": 46170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.774055019346061e-07,
+      "loss": 1.6007,
+      "step": 46180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.731536204770612e-07,
+      "loss": 1.4921,
+      "step": 46190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.689017390195162e-07,
+      "loss": 2.2399,
+      "step": 46200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.646498575619712e-07,
+      "loss": 2.7385,
+      "step": 46210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.603979761044263e-07,
+      "loss": 2.3728,
+      "step": 46220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.561460946468813e-07,
+      "loss": 2.0037,
+      "step": 46230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.518942131893363e-07,
+      "loss": 1.7057,
+      "step": 46240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.476423317317913e-07,
+      "loss": 2.0428,
+      "step": 46250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.433904502742464e-07,
+      "loss": 2.0758,
+      "step": 46260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.391385688167014e-07,
+      "loss": 1.8275,
+      "step": 46270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.348866873591565e-07,
+      "loss": 1.7757,
+      "step": 46280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.306348059016115e-07,
+      "loss": 1.9977,
+      "step": 46290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.263829244440665e-07,
+      "loss": 1.8494,
+      "step": 46300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.221310429865216e-07,
+      "loss": 2.1142,
+      "step": 46310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.178791615289766e-07,
+      "loss": 2.307,
+      "step": 46320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.136272800714316e-07,
+      "loss": 2.566,
+      "step": 46330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.093753986138867e-07,
+      "loss": 2.0823,
+      "step": 46340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.051235171563418e-07,
+      "loss": 1.5522,
+      "step": 46350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.008716356987967e-07,
+      "loss": 1.9832,
+      "step": 46360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.966197542412518e-07,
+      "loss": 1.8085,
+      "step": 46370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.923678727837068e-07,
+      "loss": 2.0295,
+      "step": 46380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.881159913261619e-07,
+      "loss": 2.2643,
+      "step": 46390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.838641098686169e-07,
+      "loss": 1.7472,
+      "step": 46400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.796122284110719e-07,
+      "loss": 2.0468,
+      "step": 46410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.75360346953527e-07,
+      "loss": 2.3911,
+      "step": 46420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.7110846549598203e-07,
+      "loss": 2.1216,
+      "step": 46430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.66856584038437e-07,
+      "loss": 2.4274,
+      "step": 46440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.626047025808921e-07,
+      "loss": 1.8198,
+      "step": 46450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.5835282112334715e-07,
+      "loss": 1.9714,
+      "step": 46460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.5410093966580214e-07,
+      "loss": 1.5653,
+      "step": 46470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.4984905820825723e-07,
+      "loss": 1.8752,
+      "step": 46480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.455971767507122e-07,
+      "loss": 1.9234,
+      "step": 46490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.4134529529316726e-07,
+      "loss": 1.8075,
+      "step": 46500
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 1.730423927307129,
+      "eval_runtime": 175.8476,
+      "eval_samples_per_second": 14.234,
+      "eval_steps_per_second": 7.12,
+      "step": 46500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3709341383562235e-07,
+      "loss": 1.9106,
+      "step": 46510
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3284153237807734e-07,
+      "loss": 2.4398,
+      "step": 46520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.285896509205324e-07,
+      "loss": 2.4179,
+      "step": 46530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.243377694629874e-07,
+      "loss": 1.7337,
+      "step": 46540
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.2008588800544246e-07,
+      "loss": 1.85,
+      "step": 46550
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.1583400654789745e-07,
+      "loss": 2.0444,
+      "step": 46560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.1158212509035254e-07,
+      "loss": 2.1128,
+      "step": 46570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0733024363280753e-07,
+      "loss": 1.9479,
+      "step": 46580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0307836217526257e-07,
+      "loss": 1.8774,
+      "step": 46590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.9882648071771767e-07,
+      "loss": 2.3917,
+      "step": 46600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.9457459926017265e-07,
+      "loss": 2.0014,
+      "step": 46610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.903227178026277e-07,
+      "loss": 2.0376,
+      "step": 46620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8607083634508274e-07,
+      "loss": 1.7728,
+      "step": 46630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.818189548875378e-07,
+      "loss": 2.0113,
+      "step": 46640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7756707342999276e-07,
+      "loss": 1.9953,
+      "step": 46650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7331519197244786e-07,
+      "loss": 2.2275,
+      "step": 46660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.690633105149029e-07,
+      "loss": 1.6948,
+      "step": 46670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.648114290573579e-07,
+      "loss": 2.005,
+      "step": 46680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.60559547599813e-07,
+      "loss": 1.8244,
+      "step": 46690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.5630766614226797e-07,
+      "loss": 2.1882,
+      "step": 46700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.52055784684723e-07,
+      "loss": 1.4587,
+      "step": 46710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.478039032271781e-07,
+      "loss": 2.0443,
+      "step": 46720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.435520217696331e-07,
+      "loss": 2.1665,
+      "step": 46730
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.3930014031208813e-07,
+      "loss": 1.856,
+      "step": 46740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.3504825885454317e-07,
+      "loss": 2.052,
+      "step": 46750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.307963773969982e-07,
+      "loss": 2.127,
+      "step": 46760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.265444959394532e-07,
+      "loss": 1.7498,
+      "step": 46770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.222926144819083e-07,
+      "loss": 2.1656,
+      "step": 46780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.180407330243633e-07,
+      "loss": 2.2154,
+      "step": 46790
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.137888515668183e-07,
+      "loss": 2.3625,
+      "step": 46800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0953697010927336e-07,
+      "loss": 1.5446,
+      "step": 46810
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.052850886517284e-07,
+      "loss": 2.333,
+      "step": 46820
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0103320719418344e-07,
+      "loss": 2.1273,
+      "step": 46830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.967813257366385e-07,
+      "loss": 2.0625,
+      "step": 46840
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.925294442790935e-07,
+      "loss": 1.7183,
+      "step": 46850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8827756282154857e-07,
+      "loss": 1.9478,
+      "step": 46860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.840256813640036e-07,
+      "loss": 1.7617,
+      "step": 46870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7977379990645865e-07,
+      "loss": 1.9488,
+      "step": 46880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.755219184489137e-07,
+      "loss": 2.2433,
+      "step": 46890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.712700369913687e-07,
+      "loss": 1.9074,
+      "step": 46900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.670181555338237e-07,
+      "loss": 1.9099,
+      "step": 46910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.6276627407627876e-07,
+      "loss": 1.9294,
+      "step": 46920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.585143926187338e-07,
+      "loss": 1.9785,
+      "step": 46930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5426251116118884e-07,
+      "loss": 2.2433,
+      "step": 46940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.500106297036439e-07,
+      "loss": 2.3595,
+      "step": 46950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.457587482460989e-07,
+      "loss": 2.0593,
+      "step": 46960
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4150686678855396e-07,
+      "loss": 1.6623,
+      "step": 46970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.37254985331009e-07,
+      "loss": 2.1463,
+      "step": 46980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3300310387346402e-07,
+      "loss": 1.8958,
+      "step": 46990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.2875122241591906e-07,
+      "loss": 1.7826,
+      "step": 47000
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 1.7335072755813599,
+      "eval_runtime": 175.9239,
+      "eval_samples_per_second": 14.228,
+      "eval_steps_per_second": 7.117,
+      "step": 47000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.244993409583741e-07,
+      "loss": 1.9725,
+      "step": 47010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.202474595008291e-07,
+      "loss": 2.3016,
+      "step": 47020
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1599557804328415e-07,
+      "loss": 2.0188,
+      "step": 47030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.1174369658573922e-07,
+      "loss": 1.9837,
+      "step": 47040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0749181512819426e-07,
+      "loss": 2.1594,
+      "step": 47050
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.0323993367064928e-07,
+      "loss": 2.0787,
+      "step": 47060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9898805221310432e-07,
+      "loss": 2.0541,
+      "step": 47070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9473617075555936e-07,
+      "loss": 1.8778,
+      "step": 47080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9048428929801437e-07,
+      "loss": 2.6506,
+      "step": 47090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8623240784046944e-07,
+      "loss": 2.3363,
+      "step": 47100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8198052638292448e-07,
+      "loss": 1.9014,
+      "step": 47110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.777286449253795e-07,
+      "loss": 1.8956,
+      "step": 47120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7347676346783453e-07,
+      "loss": 2.1171,
+      "step": 47130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6922488201028957e-07,
+      "loss": 2.0703,
+      "step": 47140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.649730005527446e-07,
+      "loss": 1.8094,
+      "step": 47150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6072111909519963e-07,
+      "loss": 2.2072,
+      "step": 47160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.564692376376547e-07,
+      "loss": 2.6089,
+      "step": 47170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.522173561801097e-07,
+      "loss": 2.0519,
+      "step": 47180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4796547472256475e-07,
+      "loss": 2.1111,
+      "step": 47190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4371359326501977e-07,
+      "loss": 1.8261,
+      "step": 47200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3946171180747483e-07,
+      "loss": 1.8875,
+      "step": 47210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3520983034992985e-07,
+      "loss": 2.3299,
+      "step": 47220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.309579488923849e-07,
+      "loss": 1.7025,
+      "step": 47230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2670606743483993e-07,
+      "loss": 2.2977,
+      "step": 47240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2245418597729497e-07,
+      "loss": 2.1292,
+      "step": 47250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1820230451975e-07,
+      "loss": 2.1523,
+      "step": 47260
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1395042306220504e-07,
+      "loss": 1.7311,
+      "step": 47270
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0969854160466007e-07,
+      "loss": 2.1331,
+      "step": 47280
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.054466601471151e-07,
+      "loss": 1.7428,
+      "step": 47290
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0119477868957015e-07,
+      "loss": 2.2247,
+      "step": 47300
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.694289723202517e-08,
+      "loss": 2.0562,
+      "step": 47310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.26910157744802e-08,
+      "loss": 1.9015,
+      "step": 47320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.843913431693526e-08,
+      "loss": 2.1303,
+      "step": 47330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.418725285939028e-08,
+      "loss": 1.8816,
+      "step": 47340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.993537140184531e-08,
+      "loss": 2.2489,
+      "step": 47350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.568348994430036e-08,
+      "loss": 2.5636,
+      "step": 47360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.143160848675539e-08,
+      "loss": 2.5362,
+      "step": 47370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.717972702921043e-08,
+      "loss": 2.2663,
+      "step": 47380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.292784557166547e-08,
+      "loss": 2.4488,
+      "step": 47390
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.86759641141205e-08,
+      "loss": 1.5922,
+      "step": 47400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4424082656575536e-08,
+      "loss": 2.1635,
+      "step": 47410
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.0172201199030576e-08,
+      "loss": 1.5336,
+      "step": 47420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.5920319741485604e-08,
+      "loss": 1.998,
+      "step": 47430
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.1668438283940644e-08,
+      "loss": 2.2229,
+      "step": 47440
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7416556826395685e-08,
+      "loss": 1.8672,
+      "step": 47450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.316467536885072e-08,
+      "loss": 1.8159,
+      "step": 47460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.8912793911305757e-08,
+      "loss": 1.5926,
+      "step": 47470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.466091245376079e-08,
+      "loss": 1.6906,
+      "step": 47480
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0409030996215825e-08,
+      "loss": 2.1867,
+      "step": 47490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6157149538670862e-08,
+      "loss": 1.9429,
+      "step": 47500
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.7336406707763672,
+      "eval_runtime": 175.8157,
+      "eval_samples_per_second": 14.236,
+      "eval_steps_per_second": 7.121,
+      "step": 47500
+    }
+  ],
+  "max_steps": 47538,
+  "num_train_epochs": 2,
+  "total_flos": 3.021220380205056e+16,
+  "trial_name": null,
+  "trial_params": null
+}