diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7533 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999842829076621,
+  "eval_steps": 1590,
+  "global_step": 15906,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0009430255402750491,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.001,
+      "loss": 5.5551,
+      "step": 15
+    },
+    {
+      "epoch": 0.0018860510805500982,
+      "grad_norm": 0.0791015625,
+      "learning_rate": 0.001,
+      "loss": 3.5038,
+      "step": 30
+    },
+    {
+      "epoch": 0.002829076620825147,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.001,
+      "loss": 3.5068,
+      "step": 45
+    },
+    {
+      "epoch": 0.0037721021611001964,
+      "grad_norm": 0.0693359375,
+      "learning_rate": 0.001,
+      "loss": 3.4288,
+      "step": 60
+    },
+    {
+      "epoch": 0.004715127701375246,
+      "grad_norm": 0.12255859375,
+      "learning_rate": 0.001,
+      "loss": 3.3071,
+      "step": 75
+    },
+    {
+      "epoch": 0.005658153241650294,
+      "grad_norm": 0.091796875,
+      "learning_rate": 0.001,
+      "loss": 3.2653,
+      "step": 90
+    },
+    {
+      "epoch": 0.006601178781925344,
+      "grad_norm": 0.1318359375,
+      "learning_rate": 0.001,
+      "loss": 3.1297,
+      "step": 105
+    },
+    {
+      "epoch": 0.007544204322200393,
+      "grad_norm": 0.12451171875,
+      "learning_rate": 0.001,
+      "loss": 3.0482,
+      "step": 120
+    },
+    {
+      "epoch": 0.008487229862475442,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.001,
+      "loss": 2.9037,
+      "step": 135
+    },
+    {
+      "epoch": 0.009430255402750491,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 0.001,
+      "loss": 2.8178,
+      "step": 150
+    },
+    {
+      "epoch": 0.01037328094302554,
+      "grad_norm": 0.111328125,
+      "learning_rate": 0.001,
+      "loss": 2.687,
+      "step": 165
+    },
+    {
+      "epoch": 0.011316306483300589,
+      "grad_norm": 0.1640625,
+      "learning_rate": 0.001,
+      "loss": 2.6247,
+      "step": 180
+    },
+    {
+      "epoch": 0.01225933202357564,
+      "grad_norm": 0.1298828125,
+      "learning_rate": 0.001,
+      "loss": 2.5556,
+      "step": 195
+    },
+    {
+      "epoch": 0.013202357563850688,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.001,
+      "loss": 2.4524,
+      "step": 210
+    },
+    {
+      "epoch": 0.014145383104125737,
+      "grad_norm": 0.1083984375,
+      "learning_rate": 0.001,
+      "loss": 2.4904,
+      "step": 225
+    },
+    {
+      "epoch": 0.015088408644400786,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 0.001,
+      "loss": 2.4211,
+      "step": 240
+    },
+    {
+      "epoch": 0.016031434184675834,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.001,
+      "loss": 2.419,
+      "step": 255
+    },
+    {
+      "epoch": 0.016974459724950885,
+      "grad_norm": 0.130859375,
+      "learning_rate": 0.001,
+      "loss": 2.3542,
+      "step": 270
+    },
+    {
+      "epoch": 0.017917485265225932,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.001,
+      "loss": 2.2893,
+      "step": 285
+    },
+    {
+      "epoch": 0.018860510805500982,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 0.001,
+      "loss": 2.2671,
+      "step": 300
+    },
+    {
+      "epoch": 0.019803536345776033,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.001,
+      "loss": 2.2644,
+      "step": 315
+    },
+    {
+      "epoch": 0.02074656188605108,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.001,
+      "loss": 2.2669,
+      "step": 330
+    },
+    {
+      "epoch": 0.02168958742632613,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.001,
+      "loss": 2.2009,
+      "step": 345
+    },
+    {
+      "epoch": 0.022632612966601177,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.001,
+      "loss": 2.1569,
+      "step": 360
+    },
+    {
+      "epoch": 0.023575638506876228,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.001,
+      "loss": 2.0607,
+      "step": 375
+    },
+    {
+      "epoch": 0.02451866404715128,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.001,
+      "loss": 2.1118,
+      "step": 390
+    },
+    {
+      "epoch": 0.025461689587426325,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 2.0465,
+      "step": 405
+    },
+    {
+      "epoch": 0.026404715127701376,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.001,
+      "loss": 2.0682,
+      "step": 420
+    },
+    {
+      "epoch": 0.027347740667976423,
+      "grad_norm": 0.158203125,
+      "learning_rate": 0.001,
+      "loss": 2.014,
+      "step": 435
+    },
+    {
+      "epoch": 0.028290766208251474,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.001,
+      "loss": 2.0251,
+      "step": 450
+    },
+    {
+      "epoch": 0.029233791748526524,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.001,
+      "loss": 1.991,
+      "step": 465
+    },
+    {
+      "epoch": 0.03017681728880157,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 0.001,
+      "loss": 1.9579,
+      "step": 480
+    },
+    {
+      "epoch": 0.03111984282907662,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 0.001,
+      "loss": 1.9253,
+      "step": 495
+    },
+    {
+      "epoch": 0.03206286836935167,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.001,
+      "loss": 1.9019,
+      "step": 510
+    },
+    {
+      "epoch": 0.033005893909626716,
+      "grad_norm": 0.15234375,
+      "learning_rate": 0.001,
+      "loss": 1.9208,
+      "step": 525
+    },
+    {
+      "epoch": 0.03394891944990177,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 1.9165,
+      "step": 540
+    },
+    {
+      "epoch": 0.03489194499017682,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.001,
+      "loss": 1.8541,
+      "step": 555
+    },
+    {
+      "epoch": 0.035834970530451864,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 1.8854,
+      "step": 570
+    },
+    {
+      "epoch": 0.03677799607072692,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.001,
+      "loss": 1.8651,
+      "step": 585
+    },
+    {
+      "epoch": 0.037721021611001965,
+      "grad_norm": 0.2734375,
+      "learning_rate": 0.001,
+      "loss": 1.8392,
+      "step": 600
+    },
+    {
+      "epoch": 0.03866404715127701,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.001,
+      "loss": 1.843,
+      "step": 615
+    },
+    {
+      "epoch": 0.039607072691552066,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.001,
+      "loss": 1.7958,
+      "step": 630
+    },
+    {
+      "epoch": 0.04055009823182711,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.001,
+      "loss": 1.7849,
+      "step": 645
+    },
+    {
+      "epoch": 0.04149312377210216,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.001,
+      "loss": 1.7397,
+      "step": 660
+    },
+    {
+      "epoch": 0.04243614931237721,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.001,
+      "loss": 1.7396,
+      "step": 675
+    },
+    {
+      "epoch": 0.04337917485265226,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.001,
+      "loss": 1.7219,
+      "step": 690
+    },
+    {
+      "epoch": 0.04432220039292731,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.001,
+      "loss": 1.7536,
+      "step": 705
+    },
+    {
+      "epoch": 0.045265225933202355,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 0.001,
+      "loss": 1.697,
+      "step": 720
+    },
+    {
+      "epoch": 0.04620825147347741,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.001,
+      "loss": 1.6725,
+      "step": 735
+    },
+    {
+      "epoch": 0.047151277013752456,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.001,
+      "loss": 1.691,
+      "step": 750
+    },
+    {
+      "epoch": 0.0480943025540275,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 1.6721,
+      "step": 765
+    },
+    {
+      "epoch": 0.04903732809430256,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.001,
+      "loss": 1.7221,
+      "step": 780
+    },
+    {
+      "epoch": 0.049980353634577604,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.001,
+      "loss": 1.6609,
+      "step": 795
+    },
+    {
+      "epoch": 0.05092337917485265,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.001,
+      "loss": 1.6805,
+      "step": 810
+    },
+    {
+      "epoch": 0.0518664047151277,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.001,
+      "loss": 1.6157,
+      "step": 825
+    },
+    {
+      "epoch": 0.05280943025540275,
+      "grad_norm": 0.19921875,
+      "learning_rate": 0.001,
+      "loss": 1.5996,
+      "step": 840
+    },
+    {
+      "epoch": 0.0537524557956778,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.001,
+      "loss": 1.5686,
+      "step": 855
+    },
+    {
+      "epoch": 0.054695481335952846,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.001,
+      "loss": 1.6021,
+      "step": 870
+    },
+    {
+      "epoch": 0.0556385068762279,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001,
+      "loss": 1.6159,
+      "step": 885
+    },
+    {
+      "epoch": 0.05658153241650295,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.001,
+      "loss": 1.5456,
+      "step": 900
+    },
+    {
+      "epoch": 0.057524557956777994,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.001,
+      "loss": 1.5764,
+      "step": 915
+    },
+    {
+      "epoch": 0.05846758349705305,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.001,
+      "loss": 1.5426,
+      "step": 930
+    },
+    {
+      "epoch": 0.059410609037328095,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.001,
+      "loss": 1.5535,
+      "step": 945
+    },
+    {
+      "epoch": 0.06035363457760314,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.001,
+      "loss": 1.505,
+      "step": 960
+    },
+    {
+      "epoch": 0.06129666011787819,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.001,
+      "loss": 1.5328,
+      "step": 975
+    },
+    {
+      "epoch": 0.06223968565815324,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.001,
+      "loss": 1.5274,
+      "step": 990
+    },
+    {
+      "epoch": 0.06318271119842829,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.001,
+      "loss": 1.5246,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06412573673870334,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 1.4633,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06506876227897838,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.001,
+      "loss": 1.487,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06601178781925343,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.001,
+      "loss": 1.4582,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06695481335952849,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.001,
+      "loss": 1.4586,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06789783889980354,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 1.4322,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06884086444007859,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.001,
+      "loss": 1.47,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06978388998035363,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.001,
+      "loss": 1.4215,
+      "step": 1110
+    },
+    {
+      "epoch": 0.07072691552062868,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 1.4569,
+      "step": 1125
+    },
+    {
+      "epoch": 0.07166994106090373,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.001,
+      "loss": 1.4428,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07261296660117879,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.001,
+      "loss": 1.3861,
+      "step": 1155
+    },
+    {
+      "epoch": 0.07355599214145384,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.001,
+      "loss": 1.4478,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07449901768172888,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.001,
+      "loss": 1.406,
+      "step": 1185
+    },
+    {
+      "epoch": 0.07544204322200393,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.001,
+      "loss": 1.3944,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07638506876227898,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 1.3884,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07732809430255402,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.001,
+      "loss": 1.38,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07827111984282907,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 1.3446,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07921414538310413,
+      "grad_norm": 0.28515625,
+      "learning_rate": 0.001,
+      "loss": 1.351,
+      "step": 1260
+    },
+    {
+      "epoch": 0.08015717092337918,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001,
+      "loss": 1.352,
+      "step": 1275
+    },
+    {
+      "epoch": 0.08110019646365423,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.001,
+      "loss": 1.3378,
+      "step": 1290
+    },
+    {
+      "epoch": 0.08204322200392927,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.001,
+      "loss": 1.3056,
+      "step": 1305
+    },
+    {
+      "epoch": 0.08298624754420432,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.001,
+      "loss": 1.3099,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08392927308447937,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 1.3364,
+      "step": 1335
+    },
+    {
+      "epoch": 0.08487229862475441,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.001,
+      "loss": 1.2865,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08581532416502947,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.001,
+      "loss": 1.3022,
+      "step": 1365
+    },
+    {
+      "epoch": 0.08675834970530452,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.001,
+      "loss": 1.2641,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08770137524557957,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 1.291,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08864440078585462,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.001,
+      "loss": 1.2947,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08958742632612966,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.001,
+      "loss": 1.2626,
+      "step": 1425
+    },
+    {
+      "epoch": 0.09053045186640471,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 1.2719,
+      "step": 1440
+    },
+    {
+      "epoch": 0.09147347740667977,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 1.2817,
+      "step": 1455
+    },
+    {
+      "epoch": 0.09241650294695482,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.001,
+      "loss": 1.2678,
+      "step": 1470
+    },
+    {
+      "epoch": 0.09335952848722986,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.001,
+      "loss": 1.2336,
+      "step": 1485
+    },
+    {
+      "epoch": 0.09430255402750491,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.001,
+      "loss": 1.2415,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09524557956777996,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.001,
+      "loss": 1.2478,
+      "step": 1515
+    },
+    {
+      "epoch": 0.096188605108055,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.001,
+      "loss": 1.2475,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09713163064833005,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.001,
+      "loss": 1.2128,
+      "step": 1545
+    },
+    {
+      "epoch": 0.09807465618860511,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.001,
+      "loss": 1.2292,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09901768172888016,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.001,
+      "loss": 1.2015,
+      "step": 1575
+    },
+    {
+      "epoch": 0.09996070726915521,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 1.2088,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09996070726915521,
+      "eval_loss": 1.5537890195846558,
+      "eval_runtime": 9.6819,
+      "eval_samples_per_second": 103.285,
+      "eval_steps_per_second": 1.446,
+      "step": 1590
+    },
+    {
+      "epoch": 0.10090373280943025,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.001,
+      "loss": 1.2156,
+      "step": 1605
+    },
+    {
+      "epoch": 0.1018467583497053,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.001,
+      "loss": 1.2115,
+      "step": 1620
+    },
+    {
+      "epoch": 0.10278978388998035,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001,
+      "loss": 1.2202,
+      "step": 1635
+    },
+    {
+      "epoch": 0.1037328094302554,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.001,
+      "loss": 1.2208,
+      "step": 1650
+    },
+    {
+      "epoch": 0.10467583497053046,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 1.1911,
+      "step": 1665
+    },
+    {
+      "epoch": 0.1056188605108055,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 1.2102,
+      "step": 1680
+    },
+    {
+      "epoch": 0.10656188605108055,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.001,
+      "loss": 1.1984,
+      "step": 1695
+    },
+    {
+      "epoch": 0.1075049115913556,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.001,
+      "loss": 1.2012,
+      "step": 1710
+    },
+    {
+      "epoch": 0.10844793713163065,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.001,
+      "loss": 1.1869,
+      "step": 1725
+    },
+    {
+      "epoch": 0.10939096267190569,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.001,
+      "loss": 1.1948,
+      "step": 1740
+    },
+    {
+      "epoch": 0.11033398821218075,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.001,
+      "loss": 1.1783,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1112770137524558,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.001,
+      "loss": 1.1893,
+      "step": 1770
+    },
+    {
+      "epoch": 0.11222003929273085,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.001,
+      "loss": 1.1495,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1131630648330059,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 1.175,
+      "step": 1800
+    },
+    {
+      "epoch": 0.11410609037328094,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001,
+      "loss": 1.1588,
+      "step": 1815
+    },
+    {
+      "epoch": 0.11504911591355599,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.001,
+      "loss": 1.1376,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11599214145383104,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 1.1511,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1169351669941061,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 1.1645,
+      "step": 1860
+    },
+    {
+      "epoch": 0.11787819253438114,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 1.1619,
+      "step": 1875
+    },
+    {
+      "epoch": 0.11882121807465619,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 1.1304,
+      "step": 1890
+    },
+    {
+      "epoch": 0.11976424361493124,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 1.1361,
+      "step": 1905
+    },
+    {
+      "epoch": 0.12070726915520628,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.001,
+      "loss": 1.1151,
+      "step": 1920
+    },
+    {
+      "epoch": 0.12165029469548133,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 1.1299,
+      "step": 1935
+    },
+    {
+      "epoch": 0.12259332023575638,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.001,
+      "loss": 1.1334,
+      "step": 1950
+    },
+    {
+      "epoch": 0.12353634577603144,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 1.112,
+      "step": 1965
+    },
+    {
+      "epoch": 0.12447937131630649,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001,
+      "loss": 1.1034,
+      "step": 1980
+    },
+    {
+      "epoch": 0.12542239685658152,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.001,
+      "loss": 1.12,
+      "step": 1995
+    },
+    {
+      "epoch": 0.12636542239685658,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.001,
+      "loss": 1.0996,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12730844793713164,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.001,
+      "loss": 1.1141,
+      "step": 2025
+    },
+    {
+      "epoch": 0.12825147347740667,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001,
+      "loss": 1.1112,
+      "step": 2040
+    },
+    {
+      "epoch": 0.12919449901768174,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 1.1229,
+      "step": 2055
+    },
+    {
+      "epoch": 0.13013752455795677,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.001,
+      "loss": 1.074,
+      "step": 2070
+    },
+    {
+      "epoch": 0.13108055009823183,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.001,
+      "loss": 1.1199,
+      "step": 2085
+    },
+    {
+      "epoch": 0.13202357563850686,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.001,
+      "loss": 1.097,
+      "step": 2100
+    },
+    {
+      "epoch": 0.13296660117878192,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.001,
+      "loss": 1.0832,
+      "step": 2115
+    },
+    {
+      "epoch": 0.13390962671905698,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.001,
+      "loss": 1.0887,
+      "step": 2130
+    },
+    {
+      "epoch": 0.13485265225933202,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.001,
+      "loss": 1.066,
+      "step": 2145
+    },
+    {
+      "epoch": 0.13579567779960708,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 1.0979,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1367387033398821,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.001,
+      "loss": 1.101,
+      "step": 2175
+    },
+    {
+      "epoch": 0.13768172888015717,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.001,
+      "loss": 1.0761,
+      "step": 2190
+    },
+    {
+      "epoch": 0.13862475442043223,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 1.0845,
+      "step": 2205
+    },
+    {
+      "epoch": 0.13956777996070727,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 1.0938,
+      "step": 2220
+    },
+    {
+      "epoch": 0.14051080550098233,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.001,
+      "loss": 1.0659,
+      "step": 2235
+    },
+    {
+      "epoch": 0.14145383104125736,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.001,
+      "loss": 1.0683,
+      "step": 2250
+    },
+    {
+      "epoch": 0.14239685658153242,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001,
+      "loss": 1.0777,
+      "step": 2265
+    },
+    {
+      "epoch": 0.14333988212180745,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 1.0741,
+      "step": 2280
+    },
+    {
+      "epoch": 0.14428290766208252,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 1.0533,
+      "step": 2295
+    },
+    {
+      "epoch": 0.14522593320235758,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 1.0655,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1461689587426326,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.001,
+      "loss": 1.0541,
+      "step": 2325
+    },
+    {
+      "epoch": 0.14711198428290767,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 1.0506,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1480550098231827,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 1.0596,
+      "step": 2355
+    },
+    {
+      "epoch": 0.14899803536345776,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 1.0586,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1499410609037328,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.001,
+      "loss": 1.0466,
+      "step": 2385
+    },
+    {
+      "epoch": 0.15088408644400786,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001,
+      "loss": 1.0485,
+      "step": 2400
+    },
+    {
+      "epoch": 0.15182711198428292,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 1.011,
+      "step": 2415
+    },
+    {
+      "epoch": 0.15277013752455795,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 1.0434,
+      "step": 2430
+    },
+    {
+      "epoch": 0.153713163064833,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 1.0353,
+      "step": 2445
+    },
+    {
+      "epoch": 0.15465618860510805,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.001,
+      "loss": 1.0222,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1555992141453831,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 1.0403,
+      "step": 2475
+    },
+    {
+      "epoch": 0.15654223968565814,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.001,
+      "loss": 1.0397,
+      "step": 2490
+    },
+    {
+      "epoch": 0.1574852652259332,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 1.0382,
+      "step": 2505
+    },
+    {
+      "epoch": 0.15842829076620826,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 1.0336,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1593713163064833,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001,
+      "loss": 1.0083,
+      "step": 2535
+    },
+    {
+      "epoch": 0.16031434184675836,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 1.0236,
+      "step": 2550
+    },
+    {
+      "epoch": 0.1612573673870334,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.001,
+      "loss": 1.0245,
+      "step": 2565
+    },
+    {
+      "epoch": 0.16220039292730845,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 1.026,
+      "step": 2580
+    },
+    {
+      "epoch": 0.16314341846758348,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 1.0276,
+      "step": 2595
+    },
+    {
+      "epoch": 0.16408644400785855,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.9937,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1650294695481336,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 1.0249,
+      "step": 2625
+    },
+    {
+      "epoch": 0.16597249508840864,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 1.0096,
+      "step": 2640
+    },
+    {
+      "epoch": 0.1669155206286837,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 1.0195,
+      "step": 2655
+    },
+    {
+      "epoch": 0.16785854616895873,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 1.018,
+      "step": 2670
+    },
+    {
+      "epoch": 0.1688015717092338,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 1.0289,
+      "step": 2685
+    },
+    {
+      "epoch": 0.16974459724950883,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.001,
+      "loss": 0.9931,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1706876227897839,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.001,
+      "loss": 1.0101,
+      "step": 2715
+    },
+    {
+      "epoch": 0.17163064833005895,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 1.0159,
+      "step": 2730
+    },
+    {
+      "epoch": 0.17257367387033398,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 1.0094,
+      "step": 2745
+    },
+    {
+      "epoch": 0.17351669941060904,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 1.0081,
+      "step": 2760
+    },
+    {
+      "epoch": 0.17445972495088408,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001,
+      "loss": 0.9958,
+      "step": 2775
+    },
+    {
+      "epoch": 0.17540275049115914,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.9909,
+      "step": 2790
+    },
+    {
+      "epoch": 0.1763457760314342,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.9854,
+      "step": 2805
+    },
+    {
+      "epoch": 0.17728880157170923,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.9858,
+      "step": 2820
+    },
+    {
+      "epoch": 0.1782318271119843,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.001,
+      "loss": 0.9825,
+      "step": 2835
+    },
+    {
+      "epoch": 0.17917485265225933,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.001,
+      "loss": 1.0153,
+      "step": 2850
+    },
+    {
+      "epoch": 0.1801178781925344,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.9984,
+      "step": 2865
+    },
+    {
+      "epoch": 0.18106090373280942,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.9832,
+      "step": 2880
+    },
+    {
+      "epoch": 0.18200392927308448,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.9843,
+      "step": 2895
+    },
+    {
+      "epoch": 0.18294695481335954,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.001,
+      "loss": 0.9774,
+      "step": 2910
+    },
+    {
+      "epoch": 0.18388998035363457,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.9824,
+      "step": 2925
+    },
+    {
+      "epoch": 0.18483300589390964,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001,
+      "loss": 0.9884,
+      "step": 2940
+    },
+    {
+      "epoch": 0.18577603143418467,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001,
+      "loss": 0.9684,
+      "step": 2955
+    },
+    {
+      "epoch": 0.18671905697445973,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001,
+      "loss": 0.9746,
+      "step": 2970
+    },
+    {
+      "epoch": 0.18766208251473476,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.9831,
+      "step": 2985
+    },
+    {
+      "epoch": 0.18860510805500982,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001,
+      "loss": 0.9868,
+      "step": 3000
+    },
+    {
+      "epoch": 0.18954813359528488,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.9687,
+      "step": 3015
+    },
+    {
+      "epoch": 0.19049115913555992,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.9759,
+      "step": 3030
+    },
+    {
+      "epoch": 0.19143418467583498,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.9755,
+      "step": 3045
+    },
+    {
+      "epoch": 0.19237721021611,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.9784,
+      "step": 3060
+    },
+    {
+      "epoch": 0.19332023575638507,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.9691,
+      "step": 3075
+    },
+    {
+      "epoch": 0.1942632612966601,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.9851,
+      "step": 3090
+    },
+    {
+      "epoch": 0.19520628683693517,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.9695,
+      "step": 3105
+    },
+    {
+      "epoch": 0.19614931237721023,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.993,
+      "step": 3120
+    },
+    {
+      "epoch": 0.19709233791748526,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.001,
+      "loss": 0.9625,
+      "step": 3135
+    },
+    {
+      "epoch": 0.19803536345776032,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.001,
+      "loss": 0.9655,
+      "step": 3150
+    },
+    {
+      "epoch": 0.19897838899803535,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.9606,
+      "step": 3165
+    },
+    {
+      "epoch": 0.19992141453831042,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.9608,
+      "step": 3180
+    },
+    {
+      "epoch": 0.19992141453831042,
+      "eval_loss": 1.169226050376892,
+      "eval_runtime": 9.7503,
+      "eval_samples_per_second": 102.561,
+      "eval_steps_per_second": 1.436,
+      "step": 3180
+    },
+    {
+      "epoch": 0.20086444007858545,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001,
+      "loss": 0.9741,
+      "step": 3195
+    },
+    {
+      "epoch": 0.2018074656188605,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.9608,
+      "step": 3210
+    },
+    {
+      "epoch": 0.20275049115913557,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.001,
+      "loss": 0.9464,
+      "step": 3225
+    },
+    {
+      "epoch": 0.2036935166994106,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001,
+      "loss": 0.9683,
+      "step": 3240
+    },
+    {
+      "epoch": 0.20463654223968566,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.9308,
+      "step": 3255
+    },
+    {
+      "epoch": 0.2055795677799607,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.001,
+      "loss": 0.9541,
+      "step": 3270
+    },
+    {
+      "epoch": 0.20652259332023576,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.9452,
+      "step": 3285
+    },
+    {
+      "epoch": 0.2074656188605108,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.9673,
+      "step": 3300
+    },
+    {
+      "epoch": 0.20840864440078585,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.9508,
+      "step": 3315
+    },
+    {
+      "epoch": 0.2093516699410609,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.955,
+      "step": 3330
+    },
+    {
+      "epoch": 0.21029469548133595,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.9499,
+      "step": 3345
+    },
+    {
+      "epoch": 0.211237721021611,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.9441,
+      "step": 3360
+    },
+    {
+      "epoch": 0.21218074656188604,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.9476,
+      "step": 3375
+    },
+    {
+      "epoch": 0.2131237721021611,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.001,
+      "loss": 0.9506,
+      "step": 3390
+    },
+    {
+      "epoch": 0.21406679764243616,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.9546,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2150098231827112,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.9488,
+      "step": 3420
+    },
+    {
+      "epoch": 0.21595284872298626,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 0.9473,
+      "step": 3435
+    },
+    {
+      "epoch": 0.2168958742632613,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.001,
+      "loss": 0.9491,
+      "step": 3450
+    },
+    {
+      "epoch": 0.21783889980353635,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001,
+      "loss": 0.9304,
+      "step": 3465
+    },
+    {
+      "epoch": 0.21878192534381138,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.001,
+      "loss": 0.9482,
+      "step": 3480
+    },
+    {
+      "epoch": 0.21972495088408645,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.9418,
+      "step": 3495
+    },
+    {
+      "epoch": 0.2206679764243615,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001,
+      "loss": 0.9226,
+      "step": 3510
+    },
+    {
+      "epoch": 0.22161100196463654,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.001,
+      "loss": 0.9427,
+      "step": 3525
+    },
+    {
+      "epoch": 0.2225540275049116,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.001,
+      "loss": 0.9261,
+      "step": 3540
+    },
+    {
+      "epoch": 0.22349705304518663,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001,
+      "loss": 0.9418,
+      "step": 3555
+    },
+    {
+      "epoch": 0.2244400785854617,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.9382,
+      "step": 3570
+    },
+    {
+      "epoch": 0.22538310412573673,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.001,
+      "loss": 0.9353,
+      "step": 3585
+    },
+    {
+      "epoch": 0.2263261296660118,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001,
+      "loss": 0.9138,
+      "step": 3600
+    },
+    {
+      "epoch": 0.22726915520628685,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.9033,
+      "step": 3615
+    },
+    {
+      "epoch": 0.22821218074656188,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.9337,
+      "step": 3630
+    },
+    {
+      "epoch": 0.22915520628683694,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.9188,
+      "step": 3645
+    },
+    {
+      "epoch": 0.23009823182711198,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.9407,
+      "step": 3660
+    },
+    {
+      "epoch": 0.23104125736738704,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.001,
+      "loss": 0.9068,
+      "step": 3675
+    },
+    {
+      "epoch": 0.23198428290766207,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.9079,
+      "step": 3690
+    },
+    {
+      "epoch": 0.23292730844793713,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 0.9095,
+      "step": 3705
+    },
+    {
+      "epoch": 0.2338703339882122,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.9148,
+      "step": 3720
+    },
+    {
+      "epoch": 0.23481335952848723,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.9044,
+      "step": 3735
+    },
+    {
+      "epoch": 0.2357563850687623,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.9401,
+      "step": 3750
+    },
+    {
+      "epoch": 0.23669941060903732,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.9228,
+      "step": 3765
+    },
+    {
+      "epoch": 0.23764243614931238,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001,
+      "loss": 0.9071,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2385854616895874,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001,
+      "loss": 0.92,
+      "step": 3795
+    },
+    {
+      "epoch": 0.23952848722986247,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.9323,
+      "step": 3810
+    },
+    {
+      "epoch": 0.24047151277013754,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.001,
+      "loss": 0.9013,
+      "step": 3825
+    },
+    {
+      "epoch": 0.24141453831041257,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.9045,
+      "step": 3840
+    },
+    {
+      "epoch": 0.24235756385068763,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.9049,
+      "step": 3855
+    },
+    {
+      "epoch": 0.24330058939096266,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 0.8902,
+      "step": 3870
+    },
+    {
+      "epoch": 0.24424361493123772,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.911,
+      "step": 3885
+    },
+    {
+      "epoch": 0.24518664047151276,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.9092,
+      "step": 3900
+    },
+    {
+      "epoch": 0.24612966601178782,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 0.894,
+      "step": 3915
+    },
+    {
+      "epoch": 0.24707269155206288,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.9096,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2480157170923379,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.9147,
+      "step": 3945
+    },
+    {
+      "epoch": 0.24895874263261297,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.001,
+      "loss": 0.9088,
+      "step": 3960
+    },
+    {
+      "epoch": 0.249901768172888,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.9116,
+      "step": 3975
+    },
+    {
+      "epoch": 0.25084479371316304,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.001,
+      "loss": 0.901,
+      "step": 3990
+    },
+    {
+      "epoch": 0.2517878192534381,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.9013,
+      "step": 4005
+    },
+    {
+      "epoch": 0.25273084479371316,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.001,
+      "loss": 0.903,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2536738703339882,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.8916,
+      "step": 4035
+    },
+    {
+      "epoch": 0.2546168958742633,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001,
+      "loss": 0.897,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2555599214145383,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001,
+      "loss": 0.9015,
+      "step": 4065
+    },
+    {
+      "epoch": 0.25650294695481335,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.001,
+      "loss": 0.897,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2574459724950884,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.8936,
+      "step": 4095
+    },
+    {
+      "epoch": 0.25838899803536347,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 0.9048,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2593320235756385,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.8973,
+      "step": 4125
+    },
+    {
+      "epoch": 0.26027504911591354,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.001,
+      "loss": 0.9053,
+      "step": 4140
+    },
+    {
+      "epoch": 0.2612180746561886,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.9121,
+      "step": 4155
+    },
+    {
+      "epoch": 0.26216110019646366,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.001,
+      "loss": 0.89,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2631041257367387,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.9025,
+      "step": 4185
+    },
+    {
+      "epoch": 0.2640471512770137,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.001,
+      "loss": 0.899,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2649901768172888,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.8793,
+      "step": 4215
+    },
+    {
+      "epoch": 0.26593320235756385,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.8964,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2668762278978389,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.896,
+      "step": 4245
+    },
+    {
+      "epoch": 0.26781925343811397,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001,
+      "loss": 0.886,
+      "step": 4260
+    },
+    {
+      "epoch": 0.268762278978389,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001,
+      "loss": 0.8861,
+      "step": 4275
+    },
+    {
+      "epoch": 0.26970530451866404,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.8864,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2706483300589391,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001,
+      "loss": 0.8834,
+      "step": 4305
+    },
+    {
+      "epoch": 0.27159135559921416,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.8859,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2725343811394892,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.8953,
+      "step": 4335
+    },
+    {
+      "epoch": 0.2734774066797642,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.8928,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2744204322200393,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001,
+      "loss": 0.8821,
+      "step": 4365
+    },
+    {
+      "epoch": 0.27536345776031435,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.8872,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2763064833005894,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.8753,
+      "step": 4395
+    },
+    {
+      "epoch": 0.27724950884086447,
+      "grad_norm": 0.80859375,
+      "learning_rate": 0.001,
+      "loss": 0.9047,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2781925343811395,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.8876,
+      "step": 4425
+    },
+    {
+      "epoch": 0.27913555992141453,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.864,
+      "step": 4440
+    },
+    {
+      "epoch": 0.28007858546168957,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.001,
+      "loss": 0.8863,
+      "step": 4455
+    },
+    {
+      "epoch": 0.28102161100196466,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.9028,
+      "step": 4470
+    },
+    {
+      "epoch": 0.2819646365422397,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.8684,
+      "step": 4485
+    },
+    {
+      "epoch": 0.2829076620825147,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.8808,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2838506876227898,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.8736,
+      "step": 4515
+    },
+    {
+      "epoch": 0.28479371316306484,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.001,
+      "loss": 0.8729,
+      "step": 4530
+    },
+    {
+      "epoch": 0.2857367387033399,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.8807,
+      "step": 4545
+    },
+    {
+      "epoch": 0.2866797642436149,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.8716,
+      "step": 4560
+    },
+    {
+      "epoch": 0.28762278978389,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.8754,
+      "step": 4575
+    },
+    {
+      "epoch": 0.28856581532416503,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.866,
+      "step": 4590
+    },
+    {
+      "epoch": 0.28950884086444006,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.8661,
+      "step": 4605
+    },
+    {
+      "epoch": 0.29045186640471515,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.8797,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2913948919449902,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.8523,
+      "step": 4635
+    },
+    {
+      "epoch": 0.2923379174852652,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.8774,
+      "step": 4650
+    },
+    {
+      "epoch": 0.29328094302554025,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.8785,
+      "step": 4665
+    },
+    {
+      "epoch": 0.29422396856581534,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.001,
+      "loss": 0.8648,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2951669941060904,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001,
+      "loss": 0.8676,
+      "step": 4695
+    },
+    {
+      "epoch": 0.2961100196463654,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.8557,
+      "step": 4710
+    },
+    {
+      "epoch": 0.2970530451866405,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001,
+      "loss": 0.8694,
+      "step": 4725
+    },
+    {
+      "epoch": 0.29799607072691553,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.8459,
+      "step": 4740
+    },
+    {
+      "epoch": 0.29893909626719056,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.001,
+      "loss": 0.8551,
+      "step": 4755
+    },
+    {
+      "epoch": 0.2998821218074656,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.8717,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2998821218074656,
+      "eval_loss": 1.035895824432373,
+      "eval_runtime": 9.7687,
+      "eval_samples_per_second": 102.368,
+      "eval_steps_per_second": 1.433,
+      "step": 4770
+    },
+    {
+      "epoch": 0.3008251473477407,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.8668,
+      "step": 4785
+    },
+    {
+      "epoch": 0.3017681728880157,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001,
+      "loss": 0.8674,
+      "step": 4800
+    },
+    {
+      "epoch": 0.30271119842829075,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.8886,
+      "step": 4815
+    },
+    {
+      "epoch": 0.30365422396856584,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.854,
+      "step": 4830
+    },
+    {
+      "epoch": 0.3045972495088409,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.8513,
+      "step": 4845
+    },
+    {
+      "epoch": 0.3055402750491159,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.8574,
+      "step": 4860
+    },
+    {
+      "epoch": 0.30648330058939094,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.8437,
+      "step": 4875
+    },
+    {
+      "epoch": 0.307426326129666,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.8604,
+      "step": 4890
+    },
+    {
+      "epoch": 0.30836935166994106,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.8544,
+      "step": 4905
+    },
+    {
+      "epoch": 0.3093123772102161,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.8607,
+      "step": 4920
+    },
+    {
+      "epoch": 0.3102554027504912,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.8454,
+      "step": 4935
+    },
+    {
+      "epoch": 0.3111984282907662,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.001,
+      "loss": 0.8575,
+      "step": 4950
+    },
+    {
+      "epoch": 0.31214145383104125,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.001,
+      "loss": 0.8401,
+      "step": 4965
+    },
+    {
+      "epoch": 0.3130844793713163,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 0.8592,
+      "step": 4980
+    },
+    {
+      "epoch": 0.31402750491159137,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.8376,
+      "step": 4995
+    },
+    {
+      "epoch": 0.3149705304518664,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.853,
+      "step": 5010
+    },
+    {
+      "epoch": 0.31591355599214144,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001,
+      "loss": 0.8659,
+      "step": 5025
+    },
+    {
+      "epoch": 0.3168565815324165,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.8733,
+      "step": 5040
+    },
+    {
+      "epoch": 0.31779960707269156,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001,
+      "loss": 0.8541,
+      "step": 5055
+    },
+    {
+      "epoch": 0.3187426326129666,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001,
+      "loss": 0.8474,
+      "step": 5070
+    },
+    {
+      "epoch": 0.3196856581532416,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.8421,
+      "step": 5085
+    },
+    {
+      "epoch": 0.3206286836935167,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.8501,
+      "step": 5100
+    },
+    {
+      "epoch": 0.32157170923379175,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.001,
+      "loss": 0.8596,
+      "step": 5115
+    },
+    {
+      "epoch": 0.3225147347740668,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.8421,
+      "step": 5130
+    },
+    {
+      "epoch": 0.32345776031434187,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.8732,
+      "step": 5145
+    },
+    {
+      "epoch": 0.3244007858546169,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.8549,
+      "step": 5160
+    },
+    {
+      "epoch": 0.32534381139489194,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001,
+      "loss": 0.8468,
+      "step": 5175
+    },
+    {
+      "epoch": 0.32628683693516697,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.8419,
+      "step": 5190
+    },
+    {
+      "epoch": 0.32722986247544206,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.001,
+      "loss": 0.8531,
+      "step": 5205
+    },
+    {
+      "epoch": 0.3281728880157171,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.848,
+      "step": 5220
+    },
+    {
+      "epoch": 0.3291159135559921,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.001,
+      "loss": 0.8367,
+      "step": 5235
+    },
+    {
+      "epoch": 0.3300589390962672,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.001,
+      "loss": 0.8405,
+      "step": 5250
+    },
+    {
+      "epoch": 0.33100196463654225,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.8567,
+      "step": 5265
+    },
+    {
+      "epoch": 0.3319449901768173,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001,
+      "loss": 0.8572,
+      "step": 5280
+    },
+    {
+      "epoch": 0.3328880157170923,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.001,
+      "loss": 0.8505,
+      "step": 5295
+    },
+    {
+      "epoch": 0.3338310412573674,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.8398,
+      "step": 5310
+    },
+    {
+      "epoch": 0.33477406679764243,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.8475,
+      "step": 5325
+    },
+    {
+      "epoch": 0.33571709233791747,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001,
+      "loss": 0.8267,
+      "step": 5340
+    },
+    {
+      "epoch": 0.33666011787819256,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001,
+      "loss": 0.8442,
+      "step": 5355
+    },
+    {
+      "epoch": 0.3376031434184676,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.8605,
+      "step": 5370
+    },
+    {
+      "epoch": 0.3385461689587426,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 0.8458,
+      "step": 5385
+    },
+    {
+      "epoch": 0.33948919449901765,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.8474,
+      "step": 5400
+    },
+    {
+      "epoch": 0.34043222003929274,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.8507,
+      "step": 5415
+    },
+    {
+      "epoch": 0.3413752455795678,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.8449,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3423182711198428,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.8456,
+      "step": 5445
+    },
+    {
+      "epoch": 0.3432612966601179,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.834,
+      "step": 5460
+    },
+    {
+      "epoch": 0.34420432220039293,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.8382,
+      "step": 5475
+    },
+    {
+      "epoch": 0.34514734774066796,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001,
+      "loss": 0.8162,
+      "step": 5490
+    },
+    {
+      "epoch": 0.346090373280943,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.8331,
+      "step": 5505
+    },
+    {
+      "epoch": 0.3470333988212181,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.8461,
+      "step": 5520
+    },
+    {
+      "epoch": 0.3479764243614931,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.8277,
+      "step": 5535
+    },
+    {
+      "epoch": 0.34891944990176815,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.8261,
+      "step": 5550
+    },
+    {
+      "epoch": 0.34986247544204324,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.8368,
+      "step": 5565
+    },
+    {
+      "epoch": 0.3508055009823183,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.829,
+      "step": 5580
+    },
+    {
+      "epoch": 0.3517485265225933,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.8356,
+      "step": 5595
+    },
+    {
+      "epoch": 0.3526915520628684,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.8404,
+      "step": 5610
+    },
+    {
+      "epoch": 0.35363457760314343,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001,
+      "loss": 0.8221,
+      "step": 5625
+    },
+    {
+      "epoch": 0.35457760314341846,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.8336,
+      "step": 5640
+    },
+    {
+      "epoch": 0.3555206286836935,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.8118,
+      "step": 5655
+    },
+    {
+      "epoch": 0.3564636542239686,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 0.8288,
+      "step": 5670
+    },
+    {
+      "epoch": 0.3574066797642436,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.8376,
+      "step": 5685
+    },
+    {
+      "epoch": 0.35834970530451865,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.8426,
+      "step": 5700
+    },
+    {
+      "epoch": 0.35929273084479374,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.8437,
+      "step": 5715
+    },
+    {
+      "epoch": 0.3602357563850688,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.8469,
+      "step": 5730
+    },
+    {
+      "epoch": 0.3611787819253438,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.8274,
+      "step": 5745
+    },
+    {
+      "epoch": 0.36212180746561884,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.8306,
+      "step": 5760
+    },
+    {
+      "epoch": 0.3630648330058939,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.8315,
+      "step": 5775
+    },
+    {
+      "epoch": 0.36400785854616896,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001,
+      "loss": 0.8379,
+      "step": 5790
+    },
+    {
+      "epoch": 0.364950884086444,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.8342,
+      "step": 5805
+    },
+    {
+      "epoch": 0.3658939096267191,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 0.8374,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3668369351669941,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.8103,
+      "step": 5835
+    },
+    {
+      "epoch": 0.36777996070726915,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.8053,
+      "step": 5850
+    },
+    {
+      "epoch": 0.3687229862475442,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.8248,
+      "step": 5865
+    },
+    {
+      "epoch": 0.36966601178781927,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.8118,
+      "step": 5880
+    },
+    {
+      "epoch": 0.3706090373280943,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.8289,
+      "step": 5895
+    },
+    {
+      "epoch": 0.37155206286836934,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.8295,
+      "step": 5910
+    },
+    {
+      "epoch": 0.3724950884086444,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.8158,
+      "step": 5925
+    },
+    {
+      "epoch": 0.37343811394891946,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.001,
+      "loss": 0.8235,
+      "step": 5940
+    },
+    {
+      "epoch": 0.3743811394891945,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.8148,
+      "step": 5955
+    },
+    {
+      "epoch": 0.3753241650294695,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.8161,
+      "step": 5970
+    },
+    {
+      "epoch": 0.3762671905697446,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.812,
+      "step": 5985
+    },
+    {
+      "epoch": 0.37721021611001965,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.8154,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3781532416502947,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.8248,
+      "step": 6015
+    },
+    {
+      "epoch": 0.37909626719056977,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001,
+      "loss": 0.8104,
+      "step": 6030
+    },
+    {
+      "epoch": 0.3800392927308448,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.8228,
+      "step": 6045
+    },
+    {
+      "epoch": 0.38098231827111984,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.8392,
+      "step": 6060
+    },
+    {
+      "epoch": 0.38192534381139487,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.8352,
+      "step": 6075
+    },
+    {
+      "epoch": 0.38286836935166996,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.8271,
+      "step": 6090
+    },
+    {
+      "epoch": 0.383811394891945,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.8122,
+      "step": 6105
+    },
+    {
+      "epoch": 0.38475442043222,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.8221,
+      "step": 6120
+    },
+    {
+      "epoch": 0.3856974459724951,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.8354,
+      "step": 6135
+    },
+    {
+      "epoch": 0.38664047151277015,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.8277,
+      "step": 6150
+    },
+    {
+      "epoch": 0.3875834970530452,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.8263,
+      "step": 6165
+    },
+    {
+      "epoch": 0.3885265225933202,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.8122,
+      "step": 6180
+    },
+    {
+      "epoch": 0.3894695481335953,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001,
+      "loss": 0.8296,
+      "step": 6195
+    },
+    {
+      "epoch": 0.39041257367387033,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001,
+      "loss": 0.8171,
+      "step": 6210
+    },
+    {
+      "epoch": 0.39135559921414537,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.8127,
+      "step": 6225
+    },
+    {
+      "epoch": 0.39229862475442046,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.806,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3932416502946955,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001,
+      "loss": 0.8157,
+      "step": 6255
+    },
+    {
+      "epoch": 0.3941846758349705,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.826,
+      "step": 6270
+    },
+    {
+      "epoch": 0.39512770137524555,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001,
+      "loss": 0.8208,
+      "step": 6285
+    },
+    {
+      "epoch": 0.39607072691552064,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.8041,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3970137524557957,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.8254,
+      "step": 6315
+    },
+    {
+      "epoch": 0.3979567779960707,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.8332,
+      "step": 6330
+    },
+    {
+      "epoch": 0.3988998035363458,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 0.8143,
+      "step": 6345
+    },
+    {
+      "epoch": 0.39984282907662083,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.8087,
+      "step": 6360
+    },
+    {
+      "epoch": 0.39984282907662083,
+      "eval_loss": 0.9629083871841431,
+      "eval_runtime": 9.6716,
+      "eval_samples_per_second": 103.395,
+      "eval_steps_per_second": 1.448,
+      "step": 6360
+    },
+    {
+      "epoch": 0.40078585461689586,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.8169,
+      "step": 6375
+    },
+    {
+      "epoch": 0.4017288801571709,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.8229,
+      "step": 6390
+    },
+    {
+      "epoch": 0.402671905697446,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.8108,
+      "step": 6405
+    },
+    {
+      "epoch": 0.403614931237721,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.814,
+      "step": 6420
+    },
+    {
+      "epoch": 0.40455795677799605,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.8077,
+      "step": 6435
+    },
+    {
+      "epoch": 0.40550098231827114,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.8103,
+      "step": 6450
+    },
+    {
+      "epoch": 0.4064440078585462,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001,
+      "loss": 0.7904,
+      "step": 6465
+    },
+    {
+      "epoch": 0.4073870333988212,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.8006,
+      "step": 6480
+    },
+    {
+      "epoch": 0.40833005893909624,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.8112,
+      "step": 6495
+    },
+    {
+      "epoch": 0.40927308447937133,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 0.7984,
+      "step": 6510
+    },
+    {
+      "epoch": 0.41021611001964636,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7883,
+      "step": 6525
+    },
+    {
+      "epoch": 0.4111591355599214,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.001,
+      "loss": 0.8196,
+      "step": 6540
+    },
+    {
+      "epoch": 0.4121021611001965,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.8274,
+      "step": 6555
+    },
+    {
+      "epoch": 0.4130451866404715,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.001,
+      "loss": 0.7942,
+      "step": 6570
+    },
+    {
+      "epoch": 0.41398821218074655,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7965,
+      "step": 6585
+    },
+    {
+      "epoch": 0.4149312377210216,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.001,
+      "loss": 0.7944,
+      "step": 6600
+    },
+    {
+      "epoch": 0.4158742632612967,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.8055,
+      "step": 6615
+    },
+    {
+      "epoch": 0.4168172888015717,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.8083,
+      "step": 6630
+    },
+    {
+      "epoch": 0.41776031434184674,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.8151,
+      "step": 6645
+    },
+    {
+      "epoch": 0.4187033398821218,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.8093,
+      "step": 6660
+    },
+    {
+      "epoch": 0.41964636542239686,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 0.807,
+      "step": 6675
+    },
+    {
+      "epoch": 0.4205893909626719,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7884,
+      "step": 6690
+    },
+    {
+      "epoch": 0.4215324165029469,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7958,
+      "step": 6705
+    },
+    {
+      "epoch": 0.422475442043222,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001,
+      "loss": 0.8029,
+      "step": 6720
+    },
+    {
+      "epoch": 0.42341846758349705,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001,
+      "loss": 0.804,
+      "step": 6735
+    },
+    {
+      "epoch": 0.4243614931237721,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001,
+      "loss": 0.8235,
+      "step": 6750
+    },
+    {
+      "epoch": 0.42530451866404717,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.8105,
+      "step": 6765
+    },
+    {
+      "epoch": 0.4262475442043222,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.8028,
+      "step": 6780
+    },
+    {
+      "epoch": 0.42719056974459724,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.8017,
+      "step": 6795
+    },
+    {
+      "epoch": 0.4281335952848723,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7998,
+      "step": 6810
+    },
+    {
+      "epoch": 0.42907662082514736,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.8083,
+      "step": 6825
+    },
+    {
+      "epoch": 0.4300196463654224,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001,
+      "loss": 0.7701,
+      "step": 6840
+    },
+    {
+      "epoch": 0.4309626719056974,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7922,
+      "step": 6855
+    },
+    {
+      "epoch": 0.4319056974459725,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7971,
+      "step": 6870
+    },
+    {
+      "epoch": 0.43284872298624755,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.795,
+      "step": 6885
+    },
+    {
+      "epoch": 0.4337917485265226,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.8004,
+      "step": 6900
+    },
+    {
+      "epoch": 0.43473477406679767,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7965,
+      "step": 6915
+    },
+    {
+      "epoch": 0.4356777996070727,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7937,
+      "step": 6930
+    },
+    {
+      "epoch": 0.43662082514734774,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001,
+      "loss": 0.8007,
+      "step": 6945
+    },
+    {
+      "epoch": 0.43756385068762277,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7935,
+      "step": 6960
+    },
+    {
+      "epoch": 0.43850687622789786,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.001,
+      "loss": 0.8045,
+      "step": 6975
+    },
+    {
+      "epoch": 0.4394499017681729,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.8055,
+      "step": 6990
+    },
+    {
+      "epoch": 0.4403929273084479,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.001,
+      "loss": 0.8005,
+      "step": 7005
+    },
+    {
+      "epoch": 0.441335952848723,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001,
+      "loss": 0.7881,
+      "step": 7020
+    },
+    {
+      "epoch": 0.44227897838899805,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.001,
+      "loss": 0.8212,
+      "step": 7035
+    },
+    {
+      "epoch": 0.4432220039292731,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7984,
+      "step": 7050
+    },
+    {
+      "epoch": 0.4441650294695481,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.8078,
+      "step": 7065
+    },
+    {
+      "epoch": 0.4451080550098232,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.001,
+      "loss": 0.7773,
+      "step": 7080
+    },
+    {
+      "epoch": 0.44605108055009823,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.7884,
+      "step": 7095
+    },
+    {
+      "epoch": 0.44699410609037327,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7842,
+      "step": 7110
+    },
+    {
+      "epoch": 0.44793713163064836,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7854,
+      "step": 7125
+    },
+    {
+      "epoch": 0.4488801571709234,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7913,
+      "step": 7140
+    },
+    {
+      "epoch": 0.4498231827111984,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7944,
+      "step": 7155
+    },
+    {
+      "epoch": 0.45076620825147345,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7935,
+      "step": 7170
+    },
+    {
+      "epoch": 0.45170923379174854,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 0.7915,
+      "step": 7185
+    },
+    {
+      "epoch": 0.4526522593320236,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.7893,
+      "step": 7200
+    },
+    {
+      "epoch": 0.4535952848722986,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.7749,
+      "step": 7215
+    },
+    {
+      "epoch": 0.4545383104125737,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.001,
+      "loss": 0.7738,
+      "step": 7230
+    },
+    {
+      "epoch": 0.45548133595284873,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7832,
+      "step": 7245
+    },
+    {
+      "epoch": 0.45642436149312376,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001,
+      "loss": 0.7935,
+      "step": 7260
+    },
+    {
+      "epoch": 0.4573673870333988,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7969,
+      "step": 7275
+    },
+    {
+      "epoch": 0.4583104125736739,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7891,
+      "step": 7290
+    },
+    {
+      "epoch": 0.4592534381139489,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7854,
+      "step": 7305
+    },
+    {
+      "epoch": 0.46019646365422395,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.8013,
+      "step": 7320
+    },
+    {
+      "epoch": 0.46113948919449904,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7864,
+      "step": 7335
+    },
+    {
+      "epoch": 0.4620825147347741,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.7932,
+      "step": 7350
+    },
+    {
+      "epoch": 0.4630255402750491,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7866,
+      "step": 7365
+    },
+    {
+      "epoch": 0.46396856581532414,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.8011,
+      "step": 7380
+    },
+    {
+      "epoch": 0.46491159135559923,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7743,
+      "step": 7395
+    },
+    {
+      "epoch": 0.46585461689587426,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.7784,
+      "step": 7410
+    },
+    {
+      "epoch": 0.4667976424361493,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7953,
+      "step": 7425
+    },
+    {
+      "epoch": 0.4677406679764244,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7807,
+      "step": 7440
+    },
+    {
+      "epoch": 0.4686836935166994,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7713,
+      "step": 7455
+    },
+    {
+      "epoch": 0.46962671905697445,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7636,
+      "step": 7470
+    },
+    {
+      "epoch": 0.4705697445972495,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.001,
+      "loss": 0.7773,
+      "step": 7485
+    },
+    {
+      "epoch": 0.4715127701375246,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.8002,
+      "step": 7500
+    },
+    {
+      "epoch": 0.4724557956777996,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.001,
+      "loss": 0.7799,
+      "step": 7515
+    },
+    {
+      "epoch": 0.47339882121807464,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7776,
+      "step": 7530
+    },
+    {
+      "epoch": 0.47434184675834973,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7823,
+      "step": 7545
+    },
+    {
+      "epoch": 0.47528487229862476,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.8059,
+      "step": 7560
+    },
+    {
+      "epoch": 0.4762278978388998,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7908,
+      "step": 7575
+    },
+    {
+      "epoch": 0.4771709233791748,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7923,
+      "step": 7590
+    },
+    {
+      "epoch": 0.4781139489194499,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.001,
+      "loss": 0.778,
+      "step": 7605
+    },
+    {
+      "epoch": 0.47905697445972495,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.8007,
+      "step": 7620
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.7842,
+      "step": 7635
+    },
+    {
+      "epoch": 0.48094302554027507,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.001,
+      "loss": 0.7968,
+      "step": 7650
+    },
+    {
+      "epoch": 0.4818860510805501,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7812,
+      "step": 7665
+    },
+    {
+      "epoch": 0.48282907662082514,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7832,
+      "step": 7680
+    },
+    {
+      "epoch": 0.48377210216110017,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7915,
+      "step": 7695
+    },
+    {
+      "epoch": 0.48471512770137526,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.8046,
+      "step": 7710
+    },
+    {
+      "epoch": 0.4856581532416503,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.7674,
+      "step": 7725
+    },
+    {
+      "epoch": 0.4866011787819253,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.7795,
+      "step": 7740
+    },
+    {
+      "epoch": 0.4875442043222004,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7983,
+      "step": 7755
+    },
+    {
+      "epoch": 0.48848722986247545,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7897,
+      "step": 7770
+    },
+    {
+      "epoch": 0.4894302554027505,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.001,
+      "loss": 0.772,
+      "step": 7785
+    },
+    {
+      "epoch": 0.4903732809430255,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7795,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4913163064833006,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7739,
+      "step": 7815
+    },
+    {
+      "epoch": 0.49225933202357564,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001,
+      "loss": 0.7891,
+      "step": 7830
+    },
+    {
+      "epoch": 0.49320235756385067,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7802,
+      "step": 7845
+    },
+    {
+      "epoch": 0.49414538310412576,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7843,
+      "step": 7860
+    },
+    {
+      "epoch": 0.4950884086444008,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7756,
+      "step": 7875
+    },
+    {
+      "epoch": 0.4960314341846758,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.77,
+      "step": 7890
+    },
+    {
+      "epoch": 0.49697445972495086,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7633,
+      "step": 7905
+    },
+    {
+      "epoch": 0.49791748526522595,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7842,
+      "step": 7920
+    },
+    {
+      "epoch": 0.498860510805501,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7742,
+      "step": 7935
+    },
+    {
+      "epoch": 0.499803536345776,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7608,
+      "step": 7950
+    },
+    {
+      "epoch": 0.499803536345776,
+      "eval_loss": 0.9156466126441956,
+      "eval_runtime": 9.6921,
+      "eval_samples_per_second": 103.176,
+      "eval_steps_per_second": 1.444,
+      "step": 7950
+    },
+    {
+      "epoch": 0.500746561886051,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7861,
+      "step": 7965
+    },
+    {
+      "epoch": 0.5016895874263261,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7726,
+      "step": 7980
+    },
+    {
+      "epoch": 0.5026326129666012,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7749,
+      "step": 7995
+    },
+    {
+      "epoch": 0.5035756385068763,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7686,
+      "step": 8010
+    },
+    {
+      "epoch": 0.5045186640471513,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7797,
+      "step": 8025
+    },
+    {
+      "epoch": 0.5054616895874263,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7622,
+      "step": 8040
+    },
+    {
+      "epoch": 0.5064047151277014,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.7753,
+      "step": 8055
+    },
+    {
+      "epoch": 0.5073477406679764,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.7744,
+      "step": 8070
+    },
+    {
+      "epoch": 0.5082907662082514,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7659,
+      "step": 8085
+    },
+    {
+      "epoch": 0.5092337917485266,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.001,
+      "loss": 0.7883,
+      "step": 8100
+    },
+    {
+      "epoch": 0.5101768172888016,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7809,
+      "step": 8115
+    },
+    {
+      "epoch": 0.5111198428290766,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.7701,
+      "step": 8130
+    },
+    {
+      "epoch": 0.5120628683693517,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001,
+      "loss": 0.7659,
+      "step": 8145
+    },
+    {
+      "epoch": 0.5130058939096267,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7772,
+      "step": 8160
+    },
+    {
+      "epoch": 0.5139489194499017,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.7769,
+      "step": 8175
+    },
+    {
+      "epoch": 0.5148919449901768,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7706,
+      "step": 8190
+    },
+    {
+      "epoch": 0.5158349705304519,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.7645,
+      "step": 8205
+    },
+    {
+      "epoch": 0.5167779960707269,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.7724,
+      "step": 8220
+    },
+    {
+      "epoch": 0.517721021611002,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7651,
+      "step": 8235
+    },
+    {
+      "epoch": 0.518664047151277,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7703,
+      "step": 8250
+    },
+    {
+      "epoch": 0.519607072691552,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7709,
+      "step": 8265
+    },
+    {
+      "epoch": 0.5205500982318271,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.001,
+      "loss": 0.7759,
+      "step": 8280
+    },
+    {
+      "epoch": 0.5214931237721021,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7687,
+      "step": 8295
+    },
+    {
+      "epoch": 0.5224361493123773,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7735,
+      "step": 8310
+    },
+    {
+      "epoch": 0.5233791748526523,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7653,
+      "step": 8325
+    },
+    {
+      "epoch": 0.5243222003929273,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.766,
+      "step": 8340
+    },
+    {
+      "epoch": 0.5252652259332024,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.768,
+      "step": 8355
+    },
+    {
+      "epoch": 0.5262082514734774,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7651,
+      "step": 8370
+    },
+    {
+      "epoch": 0.5271512770137524,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.77,
+      "step": 8385
+    },
+    {
+      "epoch": 0.5280943025540275,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7671,
+      "step": 8400
+    },
+    {
+      "epoch": 0.5290373280943026,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001,
+      "loss": 0.7568,
+      "step": 8415
+    },
+    {
+      "epoch": 0.5299803536345776,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7719,
+      "step": 8430
+    },
+    {
+      "epoch": 0.5309233791748527,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7765,
+      "step": 8445
+    },
+    {
+      "epoch": 0.5318664047151277,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7713,
+      "step": 8460
+    },
+    {
+      "epoch": 0.5328094302554027,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001,
+      "loss": 0.7779,
+      "step": 8475
+    },
+    {
+      "epoch": 0.5337524557956778,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7675,
+      "step": 8490
+    },
+    {
+      "epoch": 0.5346954813359528,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.001,
+      "loss": 0.7652,
+      "step": 8505
+    },
+    {
+      "epoch": 0.5356385068762279,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7692,
+      "step": 8520
+    },
+    {
+      "epoch": 0.536581532416503,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7781,
+      "step": 8535
+    },
+    {
+      "epoch": 0.537524557956778,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.765,
+      "step": 8550
+    },
+    {
+      "epoch": 0.538467583497053,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.001,
+      "loss": 0.7549,
+      "step": 8565
+    },
+    {
+      "epoch": 0.5394106090373281,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.7709,
+      "step": 8580
+    },
+    {
+      "epoch": 0.5403536345776031,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7739,
+      "step": 8595
+    },
+    {
+      "epoch": 0.5412966601178782,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.001,
+      "loss": 0.769,
+      "step": 8610
+    },
+    {
+      "epoch": 0.5422396856581533,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001,
+      "loss": 0.7737,
+      "step": 8625
+    },
+    {
+      "epoch": 0.5431827111984283,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7638,
+      "step": 8640
+    },
+    {
+      "epoch": 0.5441257367387033,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001,
+      "loss": 0.7392,
+      "step": 8655
+    },
+    {
+      "epoch": 0.5450687622789784,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001,
+      "loss": 0.7566,
+      "step": 8670
+    },
+    {
+      "epoch": 0.5460117878192534,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7592,
+      "step": 8685
+    },
+    {
+      "epoch": 0.5469548133595284,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 0.7485,
+      "step": 8700
+    },
+    {
+      "epoch": 0.5478978388998036,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7678,
+      "step": 8715
+    },
+    {
+      "epoch": 0.5488408644400786,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7634,
+      "step": 8730
+    },
+    {
+      "epoch": 0.5497838899803537,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.001,
+      "loss": 0.7471,
+      "step": 8745
+    },
+    {
+      "epoch": 0.5507269155206287,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.001,
+      "loss": 0.7561,
+      "step": 8760
+    },
+    {
+      "epoch": 0.5516699410609037,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7622,
+      "step": 8775
+    },
+    {
+      "epoch": 0.5526129666011788,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 0.7701,
+      "step": 8790
+    },
+    {
+      "epoch": 0.5535559921414538,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.001,
+      "loss": 0.7728,
+      "step": 8805
+    },
+    {
+      "epoch": 0.5544990176817289,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7813,
+      "step": 8820
+    },
+    {
+      "epoch": 0.555442043222004,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7614,
+      "step": 8835
+    },
+    {
+      "epoch": 0.556385068762279,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001,
+      "loss": 0.7766,
+      "step": 8850
+    },
+    {
+      "epoch": 0.557328094302554,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7735,
+      "step": 8865
+    },
+    {
+      "epoch": 0.5582711198428291,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7641,
+      "step": 8880
+    },
+    {
+      "epoch": 0.5592141453831041,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7798,
+      "step": 8895
+    },
+    {
+      "epoch": 0.5601571709233791,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7471,
+      "step": 8910
+    },
+    {
+      "epoch": 0.5611001964636543,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001,
+      "loss": 0.7625,
+      "step": 8925
+    },
+    {
+      "epoch": 0.5620432220039293,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7631,
+      "step": 8940
+    },
+    {
+      "epoch": 0.5629862475442043,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7679,
+      "step": 8955
+    },
+    {
+      "epoch": 0.5639292730844794,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7647,
+      "step": 8970
+    },
+    {
+      "epoch": 0.5648722986247544,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.7674,
+      "step": 8985
+    },
+    {
+      "epoch": 0.5658153241650294,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7735,
+      "step": 9000
+    },
+    {
+      "epoch": 0.5667583497053045,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7826,
+      "step": 9015
+    },
+    {
+      "epoch": 0.5677013752455796,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.001,
+      "loss": 0.764,
+      "step": 9030
+    },
+    {
+      "epoch": 0.5686444007858547,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7535,
+      "step": 9045
+    },
+    {
+      "epoch": 0.5695874263261297,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7588,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5705304518664047,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7622,
+      "step": 9075
+    },
+    {
+      "epoch": 0.5714734774066798,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7514,
+      "step": 9090
+    },
+    {
+      "epoch": 0.5724165029469548,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7593,
+      "step": 9105
+    },
+    {
+      "epoch": 0.5733595284872298,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.7677,
+      "step": 9120
+    },
+    {
+      "epoch": 0.574302554027505,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7539,
+      "step": 9135
+    },
+    {
+      "epoch": 0.57524557956778,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.7475,
+      "step": 9150
+    },
+    {
+      "epoch": 0.576188605108055,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 0.741,
+      "step": 9165
+    },
+    {
+      "epoch": 0.5771316306483301,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7533,
+      "step": 9180
+    },
+    {
+      "epoch": 0.5780746561886051,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.765,
+      "step": 9195
+    },
+    {
+      "epoch": 0.5790176817288801,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7741,
+      "step": 9210
+    },
+    {
+      "epoch": 0.5799607072691552,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7598,
+      "step": 9225
+    },
+    {
+      "epoch": 0.5809037328094303,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.001,
+      "loss": 0.7539,
+      "step": 9240
+    },
+    {
+      "epoch": 0.5818467583497053,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7455,
+      "step": 9255
+    },
+    {
+      "epoch": 0.5827897838899804,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.7506,
+      "step": 9270
+    },
+    {
+      "epoch": 0.5837328094302554,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001,
+      "loss": 0.7555,
+      "step": 9285
+    },
+    {
+      "epoch": 0.5846758349705304,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.7635,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5856188605108055,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7351,
+      "step": 9315
+    },
+    {
+      "epoch": 0.5865618860510805,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7341,
+      "step": 9330
+    },
+    {
+      "epoch": 0.5875049115913556,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7525,
+      "step": 9345
+    },
+    {
+      "epoch": 0.5884479371316307,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7575,
+      "step": 9360
+    },
+    {
+      "epoch": 0.5893909626719057,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7608,
+      "step": 9375
+    },
+    {
+      "epoch": 0.5903339882121807,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7602,
+      "step": 9390
+    },
+    {
+      "epoch": 0.5912770137524558,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.7615,
+      "step": 9405
+    },
+    {
+      "epoch": 0.5922200392927308,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 0.762,
+      "step": 9420
+    },
+    {
+      "epoch": 0.5931630648330058,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7635,
+      "step": 9435
+    },
+    {
+      "epoch": 0.594106090373281,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001,
+      "loss": 0.7556,
+      "step": 9450
+    },
+    {
+      "epoch": 0.595049115913556,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.001,
+      "loss": 0.7497,
+      "step": 9465
+    },
+    {
+      "epoch": 0.5959921414538311,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7419,
+      "step": 9480
+    },
+    {
+      "epoch": 0.5969351669941061,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7562,
+      "step": 9495
+    },
+    {
+      "epoch": 0.5978781925343811,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7468,
+      "step": 9510
+    },
+    {
+      "epoch": 0.5988212180746562,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.7499,
+      "step": 9525
+    },
+    {
+      "epoch": 0.5997642436149312,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7683,
+      "step": 9540
+    },
+    {
+      "epoch": 0.5997642436149312,
+      "eval_loss": 0.8865543603897095,
+      "eval_runtime": 9.6786,
+      "eval_samples_per_second": 103.32,
+      "eval_steps_per_second": 1.446,
+      "step": 9540
+    },
+    {
+      "epoch": 0.6007072691552063,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7574,
+      "step": 9555
+    },
+    {
+      "epoch": 0.6016502946954814,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7518,
+      "step": 9570
+    },
+    {
+      "epoch": 0.6025933202357564,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7391,
+      "step": 9585
+    },
+    {
+      "epoch": 0.6035363457760314,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.001,
+      "loss": 0.7425,
+      "step": 9600
+    },
+    {
+      "epoch": 0.6044793713163065,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7606,
+      "step": 9615
+    },
+    {
+      "epoch": 0.6054223968565815,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.7292,
+      "step": 9630
+    },
+    {
+      "epoch": 0.6063654223968565,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7356,
+      "step": 9645
+    },
+    {
+      "epoch": 0.6073084479371317,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7513,
+      "step": 9660
+    },
+    {
+      "epoch": 0.6082514734774067,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7522,
+      "step": 9675
+    },
+    {
+      "epoch": 0.6091944990176817,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7563,
+      "step": 9690
+    },
+    {
+      "epoch": 0.6101375245579568,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7473,
+      "step": 9705
+    },
+    {
+      "epoch": 0.6110805500982318,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001,
+      "loss": 0.76,
+      "step": 9720
+    },
+    {
+      "epoch": 0.6120235756385068,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7473,
+      "step": 9735
+    },
+    {
+      "epoch": 0.6129666011787819,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7416,
+      "step": 9750
+    },
+    {
+      "epoch": 0.613909626719057,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7449,
+      "step": 9765
+    },
+    {
+      "epoch": 0.614852652259332,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001,
+      "loss": 0.7509,
+      "step": 9780
+    },
+    {
+      "epoch": 0.6157956777996071,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 0.7468,
+      "step": 9795
+    },
+    {
+      "epoch": 0.6167387033398821,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7632,
+      "step": 9810
+    },
+    {
+      "epoch": 0.6176817288801572,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7586,
+      "step": 9825
+    },
+    {
+      "epoch": 0.6186247544204322,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7495,
+      "step": 9840
+    },
+    {
+      "epoch": 0.6195677799607072,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.7548,
+      "step": 9855
+    },
+    {
+      "epoch": 0.6205108055009824,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7484,
+      "step": 9870
+    },
+    {
+      "epoch": 0.6214538310412574,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7683,
+      "step": 9885
+    },
+    {
+      "epoch": 0.6223968565815324,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.7332,
+      "step": 9900
+    },
+    {
+      "epoch": 0.6233398821218075,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001,
+      "loss": 0.743,
+      "step": 9915
+    },
+    {
+      "epoch": 0.6242829076620825,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7527,
+      "step": 9930
+    },
+    {
+      "epoch": 0.6252259332023575,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7407,
+      "step": 9945
+    },
+    {
+      "epoch": 0.6261689587426326,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.001,
+      "loss": 0.756,
+      "step": 9960
+    },
+    {
+      "epoch": 0.6271119842829077,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001,
+      "loss": 0.7505,
+      "step": 9975
+    },
+    {
+      "epoch": 0.6280550098231827,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7517,
+      "step": 9990
+    },
+    {
+      "epoch": 0.6289980353634578,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.766,
+      "step": 10005
+    },
+    {
+      "epoch": 0.6299410609037328,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001,
+      "loss": 0.7385,
+      "step": 10020
+    },
+    {
+      "epoch": 0.6308840864440078,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.001,
+      "loss": 0.7565,
+      "step": 10035
+    },
+    {
+      "epoch": 0.6318271119842829,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7508,
+      "step": 10050
+    },
+    {
+      "epoch": 0.6327701375245579,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7519,
+      "step": 10065
+    },
+    {
+      "epoch": 0.633713163064833,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.76,
+      "step": 10080
+    },
+    {
+      "epoch": 0.6346561886051081,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7326,
+      "step": 10095
+    },
+    {
+      "epoch": 0.6355992141453831,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7506,
+      "step": 10110
+    },
+    {
+      "epoch": 0.6365422396856582,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7419,
+      "step": 10125
+    },
+    {
+      "epoch": 0.6374852652259332,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.001,
+      "loss": 0.7309,
+      "step": 10140
+    },
+    {
+      "epoch": 0.6384282907662082,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7367,
+      "step": 10155
+    },
+    {
+      "epoch": 0.6393713163064833,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001,
+      "loss": 0.7472,
+      "step": 10170
+    },
+    {
+      "epoch": 0.6403143418467584,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7431,
+      "step": 10185
+    },
+    {
+      "epoch": 0.6412573673870334,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.7496,
+      "step": 10200
+    },
+    {
+      "epoch": 0.6422003929273085,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.741,
+      "step": 10215
+    },
+    {
+      "epoch": 0.6431434184675835,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7548,
+      "step": 10230
+    },
+    {
+      "epoch": 0.6440864440078585,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7615,
+      "step": 10245
+    },
+    {
+      "epoch": 0.6450294695481336,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.001,
+      "loss": 0.764,
+      "step": 10260
+    },
+    {
+      "epoch": 0.6459724950884086,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7467,
+      "step": 10275
+    },
+    {
+      "epoch": 0.6469155206286837,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.752,
+      "step": 10290
+    },
+    {
+      "epoch": 0.6478585461689588,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.7238,
+      "step": 10305
+    },
+    {
+      "epoch": 0.6488015717092338,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.7464,
+      "step": 10320
+    },
+    {
+      "epoch": 0.6497445972495088,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.001,
+      "loss": 0.7376,
+      "step": 10335
+    },
+    {
+      "epoch": 0.6506876227897839,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.7378,
+      "step": 10350
+    },
+    {
+      "epoch": 0.6516306483300589,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 0.7536,
+      "step": 10365
+    },
+    {
+      "epoch": 0.6525736738703339,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.001,
+      "loss": 0.732,
+      "step": 10380
+    },
+    {
+      "epoch": 0.6535166994106091,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7554,
+      "step": 10395
+    },
+    {
+      "epoch": 0.6544597249508841,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.7348,
+      "step": 10410
+    },
+    {
+      "epoch": 0.6554027504911591,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7446,
+      "step": 10425
+    },
+    {
+      "epoch": 0.6563457760314342,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7386,
+      "step": 10440
+    },
+    {
+      "epoch": 0.6572888015717092,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7456,
+      "step": 10455
+    },
+    {
+      "epoch": 0.6582318271119842,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7447,
+      "step": 10470
+    },
+    {
+      "epoch": 0.6591748526522593,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7466,
+      "step": 10485
+    },
+    {
+      "epoch": 0.6601178781925344,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001,
+      "loss": 0.7638,
+      "step": 10500
+    },
+    {
+      "epoch": 0.6610609037328095,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7454,
+      "step": 10515
+    },
+    {
+      "epoch": 0.6620039292730845,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.738,
+      "step": 10530
+    },
+    {
+      "epoch": 0.6629469548133595,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.001,
+      "loss": 0.7443,
+      "step": 10545
+    },
+    {
+      "epoch": 0.6638899803536346,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7433,
+      "step": 10560
+    },
+    {
+      "epoch": 0.6648330058939096,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.001,
+      "loss": 0.7328,
+      "step": 10575
+    },
+    {
+      "epoch": 0.6657760314341846,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.001,
+      "loss": 0.7419,
+      "step": 10590
+    },
+    {
+      "epoch": 0.6667190569744598,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7387,
+      "step": 10605
+    },
+    {
+      "epoch": 0.6676620825147348,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.7325,
+      "step": 10620
+    },
+    {
+      "epoch": 0.6686051080550098,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.737,
+      "step": 10635
+    },
+    {
+      "epoch": 0.6695481335952849,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7447,
+      "step": 10650
+    },
+    {
+      "epoch": 0.6704911591355599,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7332,
+      "step": 10665
+    },
+    {
+      "epoch": 0.6714341846758349,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.001,
+      "loss": 0.7459,
+      "step": 10680
+    },
+    {
+      "epoch": 0.67237721021611,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7389,
+      "step": 10695
+    },
+    {
+      "epoch": 0.6733202357563851,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7362,
+      "step": 10710
+    },
+    {
+      "epoch": 0.6742632612966601,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001,
+      "loss": 0.7297,
+      "step": 10725
+    },
+    {
+      "epoch": 0.6752062868369352,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.7506,
+      "step": 10740
+    },
+    {
+      "epoch": 0.6761493123772102,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7279,
+      "step": 10755
+    },
+    {
+      "epoch": 0.6770923379174852,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.7329,
+      "step": 10770
+    },
+    {
+      "epoch": 0.6780353634577603,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.001,
+      "loss": 0.736,
+      "step": 10785
+    },
+    {
+      "epoch": 0.6789783889980353,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7168,
+      "step": 10800
+    },
+    {
+      "epoch": 0.6799214145383105,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7394,
+      "step": 10815
+    },
+    {
+      "epoch": 0.6808644400785855,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7165,
+      "step": 10830
+    },
+    {
+      "epoch": 0.6818074656188605,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7249,
+      "step": 10845
+    },
+    {
+      "epoch": 0.6827504911591356,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.732,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6836935166994106,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.747,
+      "step": 10875
+    },
+    {
+      "epoch": 0.6846365422396856,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7268,
+      "step": 10890
+    },
+    {
+      "epoch": 0.6855795677799607,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7334,
+      "step": 10905
+    },
+    {
+      "epoch": 0.6865225933202358,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7243,
+      "step": 10920
+    },
+    {
+      "epoch": 0.6874656188605108,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7402,
+      "step": 10935
+    },
+    {
+      "epoch": 0.6884086444007859,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.738,
+      "step": 10950
+    },
+    {
+      "epoch": 0.6893516699410609,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7309,
+      "step": 10965
+    },
+    {
+      "epoch": 0.6902946954813359,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7551,
+      "step": 10980
+    },
+    {
+      "epoch": 0.691237721021611,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7438,
+      "step": 10995
+    },
+    {
+      "epoch": 0.692180746561886,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7353,
+      "step": 11010
+    },
+    {
+      "epoch": 0.6931237721021611,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.728,
+      "step": 11025
+    },
+    {
+      "epoch": 0.6940667976424362,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.7366,
+      "step": 11040
+    },
+    {
+      "epoch": 0.6950098231827112,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7424,
+      "step": 11055
+    },
+    {
+      "epoch": 0.6959528487229862,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7434,
+      "step": 11070
+    },
+    {
+      "epoch": 0.6968958742632613,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001,
+      "loss": 0.7371,
+      "step": 11085
+    },
+    {
+      "epoch": 0.6978388998035363,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 0.7326,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6987819253438114,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7272,
+      "step": 11115
+    },
+    {
+      "epoch": 0.6997249508840865,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.738,
+      "step": 11130
+    },
+    {
+      "epoch": 0.6997249508840865,
+      "eval_loss": 0.8602269291877747,
+      "eval_runtime": 9.6753,
+      "eval_samples_per_second": 103.356,
+      "eval_steps_per_second": 1.447,
+      "step": 11130
+    },
+    {
+      "epoch": 0.7006679764243615,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7375,
+      "step": 11145
+    },
+    {
+      "epoch": 0.7016110019646365,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7545,
+      "step": 11160
+    },
+    {
+      "epoch": 0.7025540275049116,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7482,
+      "step": 11175
+    },
+    {
+      "epoch": 0.7034970530451866,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7274,
+      "step": 11190
+    },
+    {
+      "epoch": 0.7044400785854616,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7241,
+      "step": 11205
+    },
+    {
+      "epoch": 0.7053831041257368,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7303,
+      "step": 11220
+    },
+    {
+      "epoch": 0.7063261296660118,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7267,
+      "step": 11235
+    },
+    {
+      "epoch": 0.7072691552062869,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7267,
+      "step": 11250
+    },
+    {
+      "epoch": 0.7082121807465619,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.7309,
+      "step": 11265
+    },
+    {
+      "epoch": 0.7091552062868369,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7377,
+      "step": 11280
+    },
+    {
+      "epoch": 0.710098231827112,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7306,
+      "step": 11295
+    },
+    {
+      "epoch": 0.711041257367387,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7341,
+      "step": 11310
+    },
+    {
+      "epoch": 0.7119842829076621,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.7349,
+      "step": 11325
+    },
+    {
+      "epoch": 0.7129273084479372,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7407,
+      "step": 11340
+    },
+    {
+      "epoch": 0.7138703339882122,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7358,
+      "step": 11355
+    },
+    {
+      "epoch": 0.7148133595284872,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.7254,
+      "step": 11370
+    },
+    {
+      "epoch": 0.7157563850687623,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7328,
+      "step": 11385
+    },
+    {
+      "epoch": 0.7166994106090373,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7304,
+      "step": 11400
+    },
+    {
+      "epoch": 0.7176424361493123,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7317,
+      "step": 11415
+    },
+    {
+      "epoch": 0.7185854616895875,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001,
+      "loss": 0.732,
+      "step": 11430
+    },
+    {
+      "epoch": 0.7195284872298625,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.7433,
+      "step": 11445
+    },
+    {
+      "epoch": 0.7204715127701375,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.001,
+      "loss": 0.7415,
+      "step": 11460
+    },
+    {
+      "epoch": 0.7214145383104126,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7285,
+      "step": 11475
+    },
+    {
+      "epoch": 0.7223575638506876,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7299,
+      "step": 11490
+    },
+    {
+      "epoch": 0.7233005893909626,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001,
+      "loss": 0.7314,
+      "step": 11505
+    },
+    {
+      "epoch": 0.7242436149312377,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7413,
+      "step": 11520
+    },
+    {
+      "epoch": 0.7251866404715128,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7266,
+      "step": 11535
+    },
+    {
+      "epoch": 0.7261296660117879,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.7104,
+      "step": 11550
+    },
+    {
+      "epoch": 0.7270726915520629,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7252,
+      "step": 11565
+    },
+    {
+      "epoch": 0.7280157170923379,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.726,
+      "step": 11580
+    },
+    {
+      "epoch": 0.728958742632613,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.7242,
+      "step": 11595
+    },
+    {
+      "epoch": 0.729901768172888,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7313,
+      "step": 11610
+    },
+    {
+      "epoch": 0.730844793713163,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001,
+      "loss": 0.7379,
+      "step": 11625
+    },
+    {
+      "epoch": 0.7317878192534382,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7394,
+      "step": 11640
+    },
+    {
+      "epoch": 0.7327308447937132,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7332,
+      "step": 11655
+    },
+    {
+      "epoch": 0.7336738703339882,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.7154,
+      "step": 11670
+    },
+    {
+      "epoch": 0.7346168958742633,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.7351,
+      "step": 11685
+    },
+    {
+      "epoch": 0.7355599214145383,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7375,
+      "step": 11700
+    },
+    {
+      "epoch": 0.7365029469548133,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.001,
+      "loss": 0.7363,
+      "step": 11715
+    },
+    {
+      "epoch": 0.7374459724950884,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.7301,
+      "step": 11730
+    },
+    {
+      "epoch": 0.7383889980353635,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.001,
+      "loss": 0.7287,
+      "step": 11745
+    },
+    {
+      "epoch": 0.7393320235756385,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7237,
+      "step": 11760
+    },
+    {
+      "epoch": 0.7402750491159136,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7242,
+      "step": 11775
+    },
+    {
+      "epoch": 0.7412180746561886,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.7242,
+      "step": 11790
+    },
+    {
+      "epoch": 0.7421611001964636,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 0.7171,
+      "step": 11805
+    },
+    {
+      "epoch": 0.7431041257367387,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7191,
+      "step": 11820
+    },
+    {
+      "epoch": 0.7440471512770137,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.001,
+      "loss": 0.7323,
+      "step": 11835
+    },
+    {
+      "epoch": 0.7449901768172889,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7139,
+      "step": 11850
+    },
+    {
+      "epoch": 0.7459332023575639,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7237,
+      "step": 11865
+    },
+    {
+      "epoch": 0.7468762278978389,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7136,
+      "step": 11880
+    },
+    {
+      "epoch": 0.747819253438114,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7375,
+      "step": 11895
+    },
+    {
+      "epoch": 0.748762278978389,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7236,
+      "step": 11910
+    },
+    {
+      "epoch": 0.749705304518664,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7416,
+      "step": 11925
+    },
+    {
+      "epoch": 0.750648330058939,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001,
+      "loss": 0.7376,
+      "step": 11940
+    },
+    {
+      "epoch": 0.7515913555992142,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.7293,
+      "step": 11955
+    },
+    {
+      "epoch": 0.7525343811394892,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7274,
+      "step": 11970
+    },
+    {
+      "epoch": 0.7534774066797643,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7251,
+      "step": 11985
+    },
+    {
+      "epoch": 0.7544204322200393,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7221,
+      "step": 12000
+    },
+    {
+      "epoch": 0.7553634577603143,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7269,
+      "step": 12015
+    },
+    {
+      "epoch": 0.7563064833005894,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7229,
+      "step": 12030
+    },
+    {
+      "epoch": 0.7572495088408644,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7332,
+      "step": 12045
+    },
+    {
+      "epoch": 0.7581925343811395,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7425,
+      "step": 12060
+    },
+    {
+      "epoch": 0.7591355599214146,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7084,
+      "step": 12075
+    },
+    {
+      "epoch": 0.7600785854616896,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.001,
+      "loss": 0.7212,
+      "step": 12090
+    },
+    {
+      "epoch": 0.7610216110019646,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7221,
+      "step": 12105
+    },
+    {
+      "epoch": 0.7619646365422397,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7125,
+      "step": 12120
+    },
+    {
+      "epoch": 0.7629076620825147,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.001,
+      "loss": 0.7214,
+      "step": 12135
+    },
+    {
+      "epoch": 0.7638506876227897,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7211,
+      "step": 12150
+    },
+    {
+      "epoch": 0.7647937131630649,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7126,
+      "step": 12165
+    },
+    {
+      "epoch": 0.7657367387033399,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.726,
+      "step": 12180
+    },
+    {
+      "epoch": 0.766679764243615,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7079,
+      "step": 12195
+    },
+    {
+      "epoch": 0.76762278978389,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7282,
+      "step": 12210
+    },
+    {
+      "epoch": 0.768565815324165,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001,
+      "loss": 0.7293,
+      "step": 12225
+    },
+    {
+      "epoch": 0.76950884086444,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7245,
+      "step": 12240
+    },
+    {
+      "epoch": 0.7704518664047151,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7263,
+      "step": 12255
+    },
+    {
+      "epoch": 0.7713948919449902,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.001,
+      "loss": 0.7483,
+      "step": 12270
+    },
+    {
+      "epoch": 0.7723379174852653,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7243,
+      "step": 12285
+    },
+    {
+      "epoch": 0.7732809430255403,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.001,
+      "loss": 0.72,
+      "step": 12300
+    },
+    {
+      "epoch": 0.7742239685658153,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.001,
+      "loss": 0.7145,
+      "step": 12315
+    },
+    {
+      "epoch": 0.7751669941060904,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7264,
+      "step": 12330
+    },
+    {
+      "epoch": 0.7761100196463654,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7233,
+      "step": 12345
+    },
+    {
+      "epoch": 0.7770530451866404,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7132,
+      "step": 12360
+    },
+    {
+      "epoch": 0.7779960707269156,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7218,
+      "step": 12375
+    },
+    {
+      "epoch": 0.7789390962671906,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.7229,
+      "step": 12390
+    },
+    {
+      "epoch": 0.7798821218074656,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.7244,
+      "step": 12405
+    },
+    {
+      "epoch": 0.7808251473477407,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7133,
+      "step": 12420
+    },
+    {
+      "epoch": 0.7817681728880157,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7165,
+      "step": 12435
+    },
+    {
+      "epoch": 0.7827111984282907,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.7125,
+      "step": 12450
+    },
+    {
+      "epoch": 0.7836542239685658,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.001,
+      "loss": 0.7025,
+      "step": 12465
+    },
+    {
+      "epoch": 0.7845972495088409,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7143,
+      "step": 12480
+    },
+    {
+      "epoch": 0.7855402750491159,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7217,
+      "step": 12495
+    },
+    {
+      "epoch": 0.786483300589391,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.001,
+      "loss": 0.7194,
+      "step": 12510
+    },
+    {
+      "epoch": 0.787426326129666,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7117,
+      "step": 12525
+    },
+    {
+      "epoch": 0.788369351669941,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7125,
+      "step": 12540
+    },
+    {
+      "epoch": 0.7893123772102161,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.001,
+      "loss": 0.7107,
+      "step": 12555
+    },
+    {
+      "epoch": 0.7902554027504911,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7392,
+      "step": 12570
+    },
+    {
+      "epoch": 0.7911984282907663,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001,
+      "loss": 0.7211,
+      "step": 12585
+    },
+    {
+      "epoch": 0.7921414538310413,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.001,
+      "loss": 0.7139,
+      "step": 12600
+    },
+    {
+      "epoch": 0.7930844793713163,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.721,
+      "step": 12615
+    },
+    {
+      "epoch": 0.7940275049115914,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.001,
+      "loss": 0.7258,
+      "step": 12630
+    },
+    {
+      "epoch": 0.7949705304518664,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7079,
+      "step": 12645
+    },
+    {
+      "epoch": 0.7959135559921414,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.712,
+      "step": 12660
+    },
+    {
+      "epoch": 0.7968565815324165,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7296,
+      "step": 12675
+    },
+    {
+      "epoch": 0.7977996070726916,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.7146,
+      "step": 12690
+    },
+    {
+      "epoch": 0.7987426326129666,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.001,
+      "loss": 0.7202,
+      "step": 12705
+    },
+    {
+      "epoch": 0.7996856581532417,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7257,
+      "step": 12720
+    },
+    {
+      "epoch": 0.7996856581532417,
+      "eval_loss": 0.8420960307121277,
+      "eval_runtime": 9.6794,
+      "eval_samples_per_second": 103.312,
+      "eval_steps_per_second": 1.446,
+      "step": 12720
+    },
+    {
+      "epoch": 0.8006286836935167,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7087,
+      "step": 12735
+    },
+    {
+      "epoch": 0.8015717092337917,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7219,
+      "step": 12750
+    },
+    {
+      "epoch": 0.8025147347740668,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.001,
+      "loss": 0.7241,
+      "step": 12765
+    },
+    {
+      "epoch": 0.8034577603143418,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001,
+      "loss": 0.7211,
+      "step": 12780
+    },
+    {
+      "epoch": 0.8044007858546169,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7234,
+      "step": 12795
+    },
+    {
+      "epoch": 0.805343811394892,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7214,
+      "step": 12810
+    },
+    {
+      "epoch": 0.806286836935167,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7201,
+      "step": 12825
+    },
+    {
+      "epoch": 0.807229862475442,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7261,
+      "step": 12840
+    },
+    {
+      "epoch": 0.8081728880157171,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7301,
+      "step": 12855
+    },
+    {
+      "epoch": 0.8091159135559921,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7104,
+      "step": 12870
+    },
+    {
+      "epoch": 0.8100589390962671,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7094,
+      "step": 12885
+    },
+    {
+      "epoch": 0.8110019646365423,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.7188,
+      "step": 12900
+    },
+    {
+      "epoch": 0.8119449901768173,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7296,
+      "step": 12915
+    },
+    {
+      "epoch": 0.8128880157170923,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.725,
+      "step": 12930
+    },
+    {
+      "epoch": 0.8138310412573674,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.716,
+      "step": 12945
+    },
+    {
+      "epoch": 0.8147740667976424,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7199,
+      "step": 12960
+    },
+    {
+      "epoch": 0.8157170923379174,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7251,
+      "step": 12975
+    },
+    {
+      "epoch": 0.8166601178781925,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.7153,
+      "step": 12990
+    },
+    {
+      "epoch": 0.8176031434184676,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001,
+      "loss": 0.7172,
+      "step": 13005
+    },
+    {
+      "epoch": 0.8185461689587427,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7241,
+      "step": 13020
+    },
+    {
+      "epoch": 0.8194891944990177,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.001,
+      "loss": 0.7087,
+      "step": 13035
+    },
+    {
+      "epoch": 0.8204322200392927,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.001,
+      "loss": 0.7146,
+      "step": 13050
+    },
+    {
+      "epoch": 0.8213752455795678,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.001,
+      "loss": 0.7137,
+      "step": 13065
+    },
+    {
+      "epoch": 0.8223182711198428,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.001,
+      "loss": 0.7309,
+      "step": 13080
+    },
+    {
+      "epoch": 0.8232612966601178,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.001,
+      "loss": 0.7075,
+      "step": 13095
+    },
+    {
+      "epoch": 0.824204322200393,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7187,
+      "step": 13110
+    },
+    {
+      "epoch": 0.825147347740668,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7133,
+      "step": 13125
+    },
+    {
+      "epoch": 0.826090373280943,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.001,
+      "loss": 0.7062,
+      "step": 13140
+    },
+    {
+      "epoch": 0.8270333988212181,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7139,
+      "step": 13155
+    },
+    {
+      "epoch": 0.8279764243614931,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.001,
+      "loss": 0.7122,
+      "step": 13170
+    },
+    {
+      "epoch": 0.8289194499017681,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7089,
+      "step": 13185
+    },
+    {
+      "epoch": 0.8298624754420432,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.7148,
+      "step": 13200
+    },
+    {
+      "epoch": 0.8308055009823183,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.001,
+      "loss": 0.7165,
+      "step": 13215
+    },
+    {
+      "epoch": 0.8317485265225933,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.716,
+      "step": 13230
+    },
+    {
+      "epoch": 0.8326915520628684,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7324,
+      "step": 13245
+    },
+    {
+      "epoch": 0.8336345776031434,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7028,
+      "step": 13260
+    },
+    {
+      "epoch": 0.8345776031434184,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7162,
+      "step": 13275
+    },
+    {
+      "epoch": 0.8355206286836935,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7115,
+      "step": 13290
+    },
+    {
+      "epoch": 0.8364636542239685,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7323,
+      "step": 13305
+    },
+    {
+      "epoch": 0.8374066797642437,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7154,
+      "step": 13320
+    },
+    {
+      "epoch": 0.8383497053045187,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.7231,
+      "step": 13335
+    },
+    {
+      "epoch": 0.8392927308447937,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7308,
+      "step": 13350
+    },
+    {
+      "epoch": 0.8402357563850688,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7169,
+      "step": 13365
+    },
+    {
+      "epoch": 0.8411787819253438,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7209,
+      "step": 13380
+    },
+    {
+      "epoch": 0.8421218074656188,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.706,
+      "step": 13395
+    },
+    {
+      "epoch": 0.8430648330058939,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7079,
+      "step": 13410
+    },
+    {
+      "epoch": 0.844007858546169,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7062,
+      "step": 13425
+    },
+    {
+      "epoch": 0.844950884086444,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7047,
+      "step": 13440
+    },
+    {
+      "epoch": 0.8458939096267191,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7179,
+      "step": 13455
+    },
+    {
+      "epoch": 0.8468369351669941,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001,
+      "loss": 0.7159,
+      "step": 13470
+    },
+    {
+      "epoch": 0.8477799607072691,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7152,
+      "step": 13485
+    },
+    {
+      "epoch": 0.8487229862475442,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.709,
+      "step": 13500
+    },
+    {
+      "epoch": 0.8496660117878193,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.001,
+      "loss": 0.7158,
+      "step": 13515
+    },
+    {
+      "epoch": 0.8506090373280943,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7026,
+      "step": 13530
+    },
+    {
+      "epoch": 0.8515520628683694,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.7197,
+      "step": 13545
+    },
+    {
+      "epoch": 0.8524950884086444,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7271,
+      "step": 13560
+    },
+    {
+      "epoch": 0.8534381139489194,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7241,
+      "step": 13575
+    },
+    {
+      "epoch": 0.8543811394891945,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7244,
+      "step": 13590
+    },
+    {
+      "epoch": 0.8553241650294695,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.7154,
+      "step": 13605
+    },
+    {
+      "epoch": 0.8562671905697447,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.7135,
+      "step": 13620
+    },
+    {
+      "epoch": 0.8572102161100197,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7095,
+      "step": 13635
+    },
+    {
+      "epoch": 0.8581532416502947,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.001,
+      "loss": 0.7245,
+      "step": 13650
+    },
+    {
+      "epoch": 0.8590962671905698,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.7174,
+      "step": 13665
+    },
+    {
+      "epoch": 0.8600392927308448,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.001,
+      "loss": 0.7131,
+      "step": 13680
+    },
+    {
+      "epoch": 0.8609823182711198,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7227,
+      "step": 13695
+    },
+    {
+      "epoch": 0.8619253438113949,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7067,
+      "step": 13710
+    },
+    {
+      "epoch": 0.86286836935167,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 0.7013,
+      "step": 13725
+    },
+    {
+      "epoch": 0.863811394891945,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7046,
+      "step": 13740
+    },
+    {
+      "epoch": 0.8647544204322201,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7077,
+      "step": 13755
+    },
+    {
+      "epoch": 0.8656974459724951,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7064,
+      "step": 13770
+    },
+    {
+      "epoch": 0.8666404715127701,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.7177,
+      "step": 13785
+    },
+    {
+      "epoch": 0.8675834970530452,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7128,
+      "step": 13800
+    },
+    {
+      "epoch": 0.8685265225933202,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.7131,
+      "step": 13815
+    },
+    {
+      "epoch": 0.8694695481335953,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.7048,
+      "step": 13830
+    },
+    {
+      "epoch": 0.8704125736738704,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7183,
+      "step": 13845
+    },
+    {
+      "epoch": 0.8713555992141454,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7087,
+      "step": 13860
+    },
+    {
+      "epoch": 0.8722986247544204,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 0.7117,
+      "step": 13875
+    },
+    {
+      "epoch": 0.8732416502946955,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.001,
+      "loss": 0.7216,
+      "step": 13890
+    },
+    {
+      "epoch": 0.8741846758349705,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.7159,
+      "step": 13905
+    },
+    {
+      "epoch": 0.8751277013752455,
+      "grad_norm": 0.75,
+      "learning_rate": 0.001,
+      "loss": 0.7096,
+      "step": 13920
+    },
+    {
+      "epoch": 0.8760707269155207,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.702,
+      "step": 13935
+    },
+    {
+      "epoch": 0.8770137524557957,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7101,
+      "step": 13950
+    },
+    {
+      "epoch": 0.8779567779960707,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7212,
+      "step": 13965
+    },
+    {
+      "epoch": 0.8788998035363458,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.7126,
+      "step": 13980
+    },
+    {
+      "epoch": 0.8798428290766208,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.001,
+      "loss": 0.7036,
+      "step": 13995
+    },
+    {
+      "epoch": 0.8807858546168958,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7071,
+      "step": 14010
+    },
+    {
+      "epoch": 0.8817288801571709,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7051,
+      "step": 14025
+    },
+    {
+      "epoch": 0.882671905697446,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.7156,
+      "step": 14040
+    },
+    {
+      "epoch": 0.8836149312377211,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.709,
+      "step": 14055
+    },
+    {
+      "epoch": 0.8845579567779961,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.001,
+      "loss": 0.7062,
+      "step": 14070
+    },
+    {
+      "epoch": 0.8855009823182711,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.001,
+      "loss": 0.7142,
+      "step": 14085
+    },
+    {
+      "epoch": 0.8864440078585462,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7143,
+      "step": 14100
+    },
+    {
+      "epoch": 0.8873870333988212,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7093,
+      "step": 14115
+    },
+    {
+      "epoch": 0.8883300589390962,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.712,
+      "step": 14130
+    },
+    {
+      "epoch": 0.8892730844793714,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.001,
+      "loss": 0.7085,
+      "step": 14145
+    },
+    {
+      "epoch": 0.8902161100196464,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7197,
+      "step": 14160
+    },
+    {
+      "epoch": 0.8911591355599214,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.7022,
+      "step": 14175
+    },
+    {
+      "epoch": 0.8921021611001965,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7019,
+      "step": 14190
+    },
+    {
+      "epoch": 0.8930451866404715,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7171,
+      "step": 14205
+    },
+    {
+      "epoch": 0.8939882121807465,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001,
+      "loss": 0.7052,
+      "step": 14220
+    },
+    {
+      "epoch": 0.8949312377210216,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.7029,
+      "step": 14235
+    },
+    {
+      "epoch": 0.8958742632612967,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7067,
+      "step": 14250
+    },
+    {
+      "epoch": 0.8968172888015717,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.6962,
+      "step": 14265
+    },
+    {
+      "epoch": 0.8977603143418468,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.702,
+      "step": 14280
+    },
+    {
+      "epoch": 0.8987033398821218,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7066,
+      "step": 14295
+    },
+    {
+      "epoch": 0.8996463654223968,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7141,
+      "step": 14310
+    },
+    {
+      "epoch": 0.8996463654223968,
+      "eval_loss": 0.8242524266242981,
+      "eval_runtime": 9.6736,
+      "eval_samples_per_second": 103.374,
+      "eval_steps_per_second": 1.447,
+      "step": 14310
+    },
+    {
+      "epoch": 0.9005893909626719,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7051,
+      "step": 14325
+    },
+    {
+      "epoch": 0.9015324165029469,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.001,
+      "loss": 0.7161,
+      "step": 14340
+    },
+    {
+      "epoch": 0.902475442043222,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 0.6994,
+      "step": 14355
+    },
+    {
+      "epoch": 0.9034184675834971,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.001,
+      "loss": 0.7121,
+      "step": 14370
+    },
+    {
+      "epoch": 0.9043614931237721,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.7232,
+      "step": 14385
+    },
+    {
+      "epoch": 0.9053045186640472,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.7122,
+      "step": 14400
+    },
+    {
+      "epoch": 0.9062475442043222,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.001,
+      "loss": 0.7168,
+      "step": 14415
+    },
+    {
+      "epoch": 0.9071905697445972,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.001,
+      "loss": 0.6997,
+      "step": 14430
+    },
+    {
+      "epoch": 0.9081335952848723,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.001,
+      "loss": 0.7124,
+      "step": 14445
+    },
+    {
+      "epoch": 0.9090766208251474,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.001,
+      "loss": 0.6995,
+      "step": 14460
+    },
+    {
+      "epoch": 0.9100196463654224,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7087,
+      "step": 14475
+    },
+    {
+      "epoch": 0.9109626719056975,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.6991,
+      "step": 14490
+    },
+    {
+      "epoch": 0.9119056974459725,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.7069,
+      "step": 14505
+    },
+    {
+      "epoch": 0.9128487229862475,
+      "grad_norm": 0.625,
+      "learning_rate": 0.001,
+      "loss": 0.701,
+      "step": 14520
+    },
+    {
+      "epoch": 0.9137917485265226,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.001,
+      "loss": 0.7111,
+      "step": 14535
+    },
+    {
+      "epoch": 0.9147347740667976,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.6989,
+      "step": 14550
+    },
+    {
+      "epoch": 0.9156777996070727,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.7243,
+      "step": 14565
+    },
+    {
+      "epoch": 0.9166208251473478,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7043,
+      "step": 14580
+    },
+    {
+      "epoch": 0.9175638506876228,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.6925,
+      "step": 14595
+    },
+    {
+      "epoch": 0.9185068762278978,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.001,
+      "loss": 0.7129,
+      "step": 14610
+    },
+    {
+      "epoch": 0.9194499017681729,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.001,
+      "loss": 0.7064,
+      "step": 14625
+    },
+    {
+      "epoch": 0.9203929273084479,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.001,
+      "loss": 0.6876,
+      "step": 14640
+    },
+    {
+      "epoch": 0.9213359528487229,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.001,
+      "loss": 0.6978,
+      "step": 14655
+    },
+    {
+      "epoch": 0.9222789783889981,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7003,
+      "step": 14670
+    },
+    {
+      "epoch": 0.9232220039292731,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.001,
+      "loss": 0.7009,
+      "step": 14685
+    },
+    {
+      "epoch": 0.9241650294695481,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.7093,
+      "step": 14700
+    },
+    {
+      "epoch": 0.9251080550098232,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.6927,
+      "step": 14715
+    },
+    {
+      "epoch": 0.9260510805500982,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.6995,
+      "step": 14730
+    },
+    {
+      "epoch": 0.9269941060903732,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.711,
+      "step": 14745
+    },
+    {
+      "epoch": 0.9279371316306483,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.7156,
+      "step": 14760
+    },
+    {
+      "epoch": 0.9288801571709234,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.001,
+      "loss": 0.7173,
+      "step": 14775
+    },
+    {
+      "epoch": 0.9298231827111985,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.001,
+      "loss": 0.7132,
+      "step": 14790
+    },
+    {
+      "epoch": 0.9307662082514735,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.6983,
+      "step": 14805
+    },
+    {
+      "epoch": 0.9317092337917485,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7047,
+      "step": 14820
+    },
+    {
+      "epoch": 0.9326522593320236,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.7115,
+      "step": 14835
+    },
+    {
+      "epoch": 0.9335952848722986,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.001,
+      "loss": 0.7038,
+      "step": 14850
+    },
+    {
+      "epoch": 0.9345383104125736,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7066,
+      "step": 14865
+    },
+    {
+      "epoch": 0.9354813359528488,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 0.7062,
+      "step": 14880
+    },
+    {
+      "epoch": 0.9364243614931238,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.001,
+      "loss": 0.6915,
+      "step": 14895
+    },
+    {
+      "epoch": 0.9373673870333988,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.001,
+      "loss": 0.7031,
+      "step": 14910
+    },
+    {
+      "epoch": 0.9383104125736739,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.001,
+      "loss": 0.7072,
+      "step": 14925
+    },
+    {
+      "epoch": 0.9392534381139489,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.7012,
+      "step": 14940
+    },
+    {
+      "epoch": 0.9401964636542239,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.001,
+      "loss": 0.7211,
+      "step": 14955
+    },
+    {
+      "epoch": 0.941139489194499,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.001,
+      "loss": 0.7048,
+      "step": 14970
+    },
+    {
+      "epoch": 0.9420825147347741,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7016,
+      "step": 14985
+    },
+    {
+      "epoch": 0.9430255402750491,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.001,
+      "loss": 0.7095,
+      "step": 15000
+    },
+    {
+      "epoch": 0.9439685658153242,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.001,
+      "loss": 0.705,
+      "step": 15015
+    },
+    {
+      "epoch": 0.9449115913555992,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.6986,
+      "step": 15030
+    },
+    {
+      "epoch": 0.9458546168958742,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.001,
+      "loss": 0.7026,
+      "step": 15045
+    },
+    {
+      "epoch": 0.9467976424361493,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.001,
+      "loss": 0.709,
+      "step": 15060
+    },
+    {
+      "epoch": 0.9477406679764243,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.001,
+      "loss": 0.712,
+      "step": 15075
+    },
+    {
+      "epoch": 0.9486836935166995,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.001,
+      "loss": 0.7126,
+      "step": 15090
+    },
+    {
+      "epoch": 0.9496267190569745,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.001,
+      "loss": 0.6879,
+      "step": 15105
+    },
+    {
+      "epoch": 0.9505697445972495,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7031,
+      "step": 15120
+    },
+    {
+      "epoch": 0.9515127701375246,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.7146,
+      "step": 15135
+    },
+    {
+      "epoch": 0.9524557956777996,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.001,
+      "loss": 0.6882,
+      "step": 15150
+    },
+    {
+      "epoch": 0.9533988212180746,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.001,
+      "loss": 0.6981,
+      "step": 15165
+    },
+    {
+      "epoch": 0.9543418467583497,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7011,
+      "step": 15180
+    },
+    {
+      "epoch": 0.9552848722986248,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.698,
+      "step": 15195
+    },
+    {
+      "epoch": 0.9562278978388998,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.6932,
+      "step": 15210
+    },
+    {
+      "epoch": 0.9571709233791749,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.69,
+      "step": 15225
+    },
+    {
+      "epoch": 0.9581139489194499,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.001,
+      "loss": 0.695,
+      "step": 15240
+    },
+    {
+      "epoch": 0.9590569744597249,
+      "grad_norm": 0.5,
+      "learning_rate": 0.001,
+      "loss": 0.7002,
+      "step": 15255
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.001,
+      "loss": 0.6943,
+      "step": 15270
+    },
+    {
+      "epoch": 0.960943025540275,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.001,
+      "loss": 0.7044,
+      "step": 15285
+    },
+    {
+      "epoch": 0.9618860510805501,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.7069,
+      "step": 15300
+    },
+    {
+      "epoch": 0.9628290766208252,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.6985,
+      "step": 15315
+    },
+    {
+      "epoch": 0.9637721021611002,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.001,
+      "loss": 0.7049,
+      "step": 15330
+    },
+    {
+      "epoch": 0.9647151277013752,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.001,
+      "loss": 0.7035,
+      "step": 15345
+    },
+    {
+      "epoch": 0.9656581532416503,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.001,
+      "loss": 0.7016,
+      "step": 15360
+    },
+    {
+      "epoch": 0.9666011787819253,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.6954,
+      "step": 15375
+    },
+    {
+      "epoch": 0.9675442043222003,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.001,
+      "loss": 0.7014,
+      "step": 15390
+    },
+    {
+      "epoch": 0.9684872298624755,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.7129,
+      "step": 15405
+    },
+    {
+      "epoch": 0.9694302554027505,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.001,
+      "loss": 0.6999,
+      "step": 15420
+    },
+    {
+      "epoch": 0.9703732809430256,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7017,
+      "step": 15435
+    },
+    {
+      "epoch": 0.9713163064833006,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.001,
+      "loss": 0.6893,
+      "step": 15450
+    },
+    {
+      "epoch": 0.9722593320235756,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.001,
+      "loss": 0.6993,
+      "step": 15465
+    },
+    {
+      "epoch": 0.9732023575638507,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.001,
+      "loss": 0.6999,
+      "step": 15480
+    },
+    {
+      "epoch": 0.9741453831041257,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.6864,
+      "step": 15495
+    },
+    {
+      "epoch": 0.9750884086444008,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.001,
+      "loss": 0.7057,
+      "step": 15510
+    },
+    {
+      "epoch": 0.9760314341846759,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.6957,
+      "step": 15525
+    },
+    {
+      "epoch": 0.9769744597249509,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.001,
+      "loss": 0.709,
+      "step": 15540
+    },
+    {
+      "epoch": 0.9779174852652259,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.001,
+      "loss": 0.6965,
+      "step": 15555
+    },
+    {
+      "epoch": 0.978860510805501,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.6989,
+      "step": 15570
+    },
+    {
+      "epoch": 0.979803536345776,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.6995,
+      "step": 15585
+    },
+    {
+      "epoch": 0.980746561886051,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.6894,
+      "step": 15600
+    },
+    {
+      "epoch": 0.9816895874263262,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.001,
+      "loss": 0.7084,
+      "step": 15615
+    },
+    {
+      "epoch": 0.9826326129666012,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.001,
+      "loss": 0.7021,
+      "step": 15630
+    },
+    {
+      "epoch": 0.9835756385068762,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.001,
+      "loss": 0.6892,
+      "step": 15645
+    },
+    {
+      "epoch": 0.9845186640471513,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.001,
+      "loss": 0.7147,
+      "step": 15660
+    },
+    {
+      "epoch": 0.9854616895874263,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.001,
+      "loss": 0.7007,
+      "step": 15675
+    },
+    {
+      "epoch": 0.9864047151277013,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.699,
+      "step": 15690
+    },
+    {
+      "epoch": 0.9873477406679764,
+      "grad_norm": 0.875,
+      "learning_rate": 0.001,
+      "loss": 0.6943,
+      "step": 15705
+    },
+    {
+      "epoch": 0.9882907662082515,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.001,
+      "loss": 0.6943,
+      "step": 15720
+    },
+    {
+      "epoch": 0.9892337917485265,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.001,
+      "loss": 0.703,
+      "step": 15735
+    },
+    {
+      "epoch": 0.9901768172888016,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.6953,
+      "step": 15750
+    },
+    {
+      "epoch": 0.9911198428290766,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.001,
+      "loss": 0.6884,
+      "step": 15765
+    },
+    {
+      "epoch": 0.9920628683693516,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.001,
+      "loss": 0.6972,
+      "step": 15780
+    },
+    {
+      "epoch": 0.9930058939096267,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.001,
+      "loss": 0.6929,
+      "step": 15795
+    },
+    {
+      "epoch": 0.9939489194499017,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.001,
+      "loss": 0.6849,
+      "step": 15810
+    },
+    {
+      "epoch": 0.9948919449901769,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.001,
+      "loss": 0.6932,
+      "step": 15825
+    },
+    {
+      "epoch": 0.9958349705304519,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.001,
+      "loss": 0.7042,
+      "step": 15840
+    },
+    {
+      "epoch": 0.9967779960707269,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.001,
+      "loss": 0.6924,
+      "step": 15855
+    },
+    {
+      "epoch": 0.997721021611002,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.001,
+      "loss": 0.7009,
+      "step": 15870
+    },
+    {
+      "epoch": 0.998664047151277,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.001,
+      "loss": 0.7059,
+      "step": 15885
+    },
+    {
+      "epoch": 0.999607072691552,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.001,
+      "loss": 0.691,
+      "step": 15900
+    },
+    {
+      "epoch": 0.999607072691552,
+      "eval_loss": 0.8118711709976196,
+      "eval_runtime": 9.6839,
+      "eval_samples_per_second": 103.264,
+      "eval_steps_per_second": 1.446,
+      "step": 15900
+    }
+  ],
+  "logging_steps": 15,
+  "max_steps": 15906,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1590,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.185992916964999e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}