diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,23422 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.972075108329321,
+  "eval_steps": 500,
+  "global_step": 16500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001203659123736158,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.3437477745175867e-05,
+      "loss": 1.9394,
+      "step": 5
+    },
+    {
+      "epoch": 0.002407318247472316,
+      "grad_norm": 4.09375,
+      "learning_rate": 5.2734324926645696e-05,
+      "loss": 1.8185,
+      "step": 10
+    },
+    {
+      "epoch": 0.0036109773712084737,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.203117210811555e-05,
+      "loss": 1.6969,
+      "step": 15
+    },
+    {
+      "epoch": 0.004814636494944632,
+      "grad_norm": 3.09375,
+      "learning_rate": 0.00011132801928958537,
+      "loss": 1.5691,
+      "step": 20
+    },
+    {
+      "epoch": 0.00601829561868079,
+      "grad_norm": 3.1875,
+      "learning_rate": 0.00014062486647105523,
+      "loss": 1.511,
+      "step": 25
+    },
+    {
+      "epoch": 0.007221954742416947,
+      "grad_norm": 2.84375,
+      "learning_rate": 0.00016992171365252506,
+      "loss": 1.4285,
+      "step": 30
+    },
+    {
+      "epoch": 0.008425613866153106,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.0001992185608339949,
+      "loss": 1.3826,
+      "step": 35
+    },
+    {
+      "epoch": 0.009629272989889264,
+      "grad_norm": 3.109375,
+      "learning_rate": 0.00020507790818421773,
+      "loss": 1.3233,
+      "step": 40
+    },
+    {
+      "epoch": 0.010832932113625422,
+      "grad_norm": 3.125,
+      "learning_rate": 0.00020507781845957552,
+      "loss": 1.3161,
+      "step": 45
+    },
+    {
+      "epoch": 0.01203659123736158,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.00020507765971606326,
+      "loss": 1.2634,
+      "step": 50
+    },
+    {
+      "epoch": 0.013240250361097737,
+      "grad_norm": 3.15625,
+      "learning_rate": 0.00020507743195382337,
+      "loss": 1.2576,
+      "step": 55
+    },
+    {
+      "epoch": 0.014443909484833895,
+      "grad_norm": 2.859375,
+      "learning_rate": 0.00020507713517306031,
+      "loss": 1.2109,
+      "step": 60
+    },
+    {
+      "epoch": 0.015647568608570053,
+      "grad_norm": 2.703125,
+      "learning_rate": 0.00020507676937404044,
+      "loss": 1.2114,
+      "step": 65
+    },
+    {
+      "epoch": 0.016851227732306212,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00020507633455709202,
+      "loss": 1.1827,
+      "step": 70
+    },
+    {
+      "epoch": 0.018054886856042368,
+      "grad_norm": 2.578125,
+      "learning_rate": 0.00020507583072260527,
+      "loss": 1.1745,
+      "step": 75
+    },
+    {
+      "epoch": 0.019258545979778528,
+      "grad_norm": 2.765625,
+      "learning_rate": 0.00020507525787103237,
+      "loss": 1.1547,
+      "step": 80
+    },
+    {
+      "epoch": 0.020462205103514684,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00020507461600288745,
+      "loss": 1.1566,
+      "step": 85
+    },
+    {
+      "epoch": 0.021665864227250843,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.0002050739051187466,
+      "loss": 1.1486,
+      "step": 90
+    },
+    {
+      "epoch": 0.022869523350987,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00020507312521924776,
+      "loss": 1.1269,
+      "step": 95
+    },
+    {
+      "epoch": 0.02407318247472316,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00020507227630509089,
+      "loss": 1.1181,
+      "step": 100
+    },
+    {
+      "epoch": 0.025276841598459315,
+      "grad_norm": 2.703125,
+      "learning_rate": 0.00020507135837703782,
+      "loss": 1.0867,
+      "step": 105
+    },
+    {
+      "epoch": 0.026480500722195474,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00020507037143591242,
+      "loss": 1.1229,
+      "step": 110
+    },
+    {
+      "epoch": 0.027684159845931634,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.00020506931548260044,
+      "loss": 1.1203,
+      "step": 115
+    },
+    {
+      "epoch": 0.02888781896966779,
+      "grad_norm": 3.125,
+      "learning_rate": 0.00020506819051804948,
+      "loss": 1.0955,
+      "step": 120
+    },
+    {
+      "epoch": 0.03009147809340395,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00020506699654326924,
+      "loss": 1.0462,
+      "step": 125
+    },
+    {
+      "epoch": 0.031295137217140105,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00020506573355933124,
+      "loss": 1.1156,
+      "step": 130
+    },
+    {
+      "epoch": 0.03249879634087626,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.00020506440156736898,
+      "loss": 1.0528,
+      "step": 135
+    },
+    {
+      "epoch": 0.033702455464612424,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00020506300056857786,
+      "loss": 1.0555,
+      "step": 140
+    },
+    {
+      "epoch": 0.03490611458834858,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.00020506153056421524,
+      "loss": 1.0468,
+      "step": 145
+    },
+    {
+      "epoch": 0.036109773712084736,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00020505999155560043,
+      "loss": 1.0505,
+      "step": 150
+    },
+    {
+      "epoch": 0.03731343283582089,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00020505838354411456,
+      "loss": 1.0497,
+      "step": 155
+    },
+    {
+      "epoch": 0.038517091959557055,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.0002050567065312008,
+      "loss": 1.0736,
+      "step": 160
+    },
+    {
+      "epoch": 0.03972075108329321,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00020505496051836426,
+      "loss": 1.0362,
+      "step": 165
+    },
+    {
+      "epoch": 0.04092441020702937,
+      "grad_norm": 2.5,
+      "learning_rate": 0.0002050531455071719,
+      "loss": 1.0406,
+      "step": 170
+    },
+    {
+      "epoch": 0.04212806933076553,
+      "grad_norm": 2.0,
+      "learning_rate": 0.0002050512614992526,
+      "loss": 1.0319,
+      "step": 175
+    },
+    {
+      "epoch": 0.043331728454501686,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00020504930849629726,
+      "loss": 1.0327,
+      "step": 180
+    },
+    {
+      "epoch": 0.04453538757823784,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00020504728650005854,
+      "loss": 0.9765,
+      "step": 185
+    },
+    {
+      "epoch": 0.045739046701974,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00020504519551235124,
+      "loss": 1.0428,
+      "step": 190
+    },
+    {
+      "epoch": 0.04694270582571016,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00020504303553505184,
+      "loss": 0.9931,
+      "step": 195
+    },
+    {
+      "epoch": 0.04814636494944632,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00020504080657009885,
+      "loss": 0.9917,
+      "step": 200
+    },
+    {
+      "epoch": 0.04935002407318247,
+      "grad_norm": 1.875,
+      "learning_rate": 0.0002050385086194928,
+      "loss": 0.9869,
+      "step": 205
+    },
+    {
+      "epoch": 0.05055368319691863,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00020503614168529592,
+      "loss": 0.9892,
+      "step": 210
+    },
+    {
+      "epoch": 0.05175734232065479,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00020503370576963252,
+      "loss": 0.9916,
+      "step": 215
+    },
+    {
+      "epoch": 0.05296100144439095,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00020503120087468868,
+      "loss": 0.9903,
+      "step": 220
+    },
+    {
+      "epoch": 0.054164660568127104,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00020502862700271253,
+      "loss": 0.9884,
+      "step": 225
+    },
+    {
+      "epoch": 0.05536831969186327,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00020502598415601396,
+      "loss": 0.9797,
+      "step": 230
+    },
+    {
+      "epoch": 0.05657197881559942,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00020502327233696492,
+      "loss": 0.9547,
+      "step": 235
+    },
+    {
+      "epoch": 0.05777563793933558,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00020502049154799912,
+      "loss": 0.9535,
+      "step": 240
+    },
+    {
+      "epoch": 0.058979297063071735,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00020501764179161228,
+      "loss": 0.972,
+      "step": 245
+    },
+    {
+      "epoch": 0.0601829561868079,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00020501472307036186,
+      "loss": 0.9618,
+      "step": 250
+    },
+    {
+      "epoch": 0.061386615310544054,
+      "grad_norm": 1.9921875,
+      "learning_rate": 0.00020501173538686743,
+      "loss": 0.9849,
+      "step": 255
+    },
+    {
+      "epoch": 0.06259027443428021,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00020500867874381027,
+      "loss": 0.923,
+      "step": 260
+    },
+    {
+      "epoch": 0.06379393355801637,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00020500555314393362,
+      "loss": 0.9549,
+      "step": 265
+    },
+    {
+      "epoch": 0.06499759268175252,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00020500235859004264,
+      "loss": 0.9438,
+      "step": 270
+    },
+    {
+      "epoch": 0.06620125180548869,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.0002049990950850043,
+      "loss": 0.9447,
+      "step": 275
+    },
+    {
+      "epoch": 0.06740491092922485,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.0002049957626317475,
+      "loss": 0.94,
+      "step": 280
+    },
+    {
+      "epoch": 0.068608570052961,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00020499236123326293,
+      "loss": 0.9199,
+      "step": 285
+    },
+    {
+      "epoch": 0.06981222917669716,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.0002049888908926034,
+      "loss": 0.9234,
+      "step": 290
+    },
+    {
+      "epoch": 0.07101588830043332,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00020498535161288326,
+      "loss": 0.9207,
+      "step": 295
+    },
+    {
+      "epoch": 0.07221954742416947,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00020498174339727898,
+      "loss": 0.899,
+      "step": 300
+    },
+    {
+      "epoch": 0.07342320654790563,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00020497806624902878,
+      "loss": 0.9195,
+      "step": 305
+    },
+    {
+      "epoch": 0.07462686567164178,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00020497432017143284,
+      "loss": 0.9605,
+      "step": 310
+    },
+    {
+      "epoch": 0.07583052479537795,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00020497050516785306,
+      "loss": 0.8905,
+      "step": 315
+    },
+    {
+      "epoch": 0.07703418391911411,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00020496662124171327,
+      "loss": 0.901,
+      "step": 320
+    },
+    {
+      "epoch": 0.07823784304285027,
+      "grad_norm": 1.9921875,
+      "learning_rate": 0.00020496266839649927,
+      "loss": 0.9063,
+      "step": 325
+    },
+    {
+      "epoch": 0.07944150216658642,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.0002049586466357585,
+      "loss": 0.9248,
+      "step": 330
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00020495455596310038,
+      "loss": 0.9111,
+      "step": 335
+    },
+    {
+      "epoch": 0.08184882041405873,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00020495039638219618,
+      "loss": 0.907,
+      "step": 340
+    },
+    {
+      "epoch": 0.08305247953779489,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00020494616789677898,
+      "loss": 0.8876,
+      "step": 345
+    },
+    {
+      "epoch": 0.08425613866153106,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.0002049418705106437,
+      "loss": 0.9286,
+      "step": 350
+    },
+    {
+      "epoch": 0.08545979778526722,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0002049375042276471,
+      "loss": 0.9021,
+      "step": 355
+    },
+    {
+      "epoch": 0.08666345690900337,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00020493306905170776,
+      "loss": 0.904,
+      "step": 360
+    },
+    {
+      "epoch": 0.08786711603273953,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00020492856498680612,
+      "loss": 0.9037,
+      "step": 365
+    },
+    {
+      "epoch": 0.08907077515647568,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.0002049239920369844,
+      "loss": 0.9056,
+      "step": 370
+    },
+    {
+      "epoch": 0.09027443428021184,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.00020491935020634672,
+      "loss": 0.9118,
+      "step": 375
+    },
+    {
+      "epoch": 0.091478093403948,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.0002049146394990589,
+      "loss": 0.8825,
+      "step": 380
+    },
+    {
+      "epoch": 0.09268175252768417,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.0002049098599193487,
+      "loss": 0.8976,
+      "step": 385
+    },
+    {
+      "epoch": 0.09388541165142032,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.0002049050114715056,
+      "loss": 0.8852,
+      "step": 390
+    },
+    {
+      "epoch": 0.09508907077515648,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00020490009415988098,
+      "loss": 0.8853,
+      "step": 395
+    },
+    {
+      "epoch": 0.09629272989889263,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00020489510798888785,
+      "loss": 0.882,
+      "step": 400
+    },
+    {
+      "epoch": 0.09749638902262879,
+      "grad_norm": 1.8828125,
+      "learning_rate": 0.00020489005296300122,
+      "loss": 0.8837,
+      "step": 405
+    },
+    {
+      "epoch": 0.09870004814636495,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.0002048849290867578,
+      "loss": 0.8927,
+      "step": 410
+    },
+    {
+      "epoch": 0.0999037072701011,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00020487973636475608,
+      "loss": 0.8974,
+      "step": 415
+    },
+    {
+      "epoch": 0.10110736639383726,
+      "grad_norm": 2.046875,
+      "learning_rate": 0.00020487447480165634,
+      "loss": 0.8939,
+      "step": 420
+    },
+    {
+      "epoch": 0.10231102551757343,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.00020486914440218066,
+      "loss": 0.8905,
+      "step": 425
+    },
+    {
+      "epoch": 0.10351468464130958,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00020486374517111295,
+      "loss": 0.8357,
+      "step": 430
+    },
+    {
+      "epoch": 0.10471834376504574,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00020485827711329876,
+      "loss": 0.8793,
+      "step": 435
+    },
+    {
+      "epoch": 0.1059220028887819,
+      "grad_norm": 1.9765625,
+      "learning_rate": 0.00020485274023364553,
+      "loss": 0.8499,
+      "step": 440
+    },
+    {
+      "epoch": 0.10712566201251805,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00020484713453712244,
+      "loss": 0.8867,
+      "step": 445
+    },
+    {
+      "epoch": 0.10832932113625421,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00020484146002876038,
+      "loss": 0.8901,
+      "step": 450
+    },
+    {
+      "epoch": 0.10953298025999036,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00020483571671365205,
+      "loss": 0.8975,
+      "step": 455
+    },
+    {
+      "epoch": 0.11073663938372653,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.0002048299045969519,
+      "loss": 0.8592,
+      "step": 460
+    },
+    {
+      "epoch": 0.11194029850746269,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.00020482402368387606,
+      "loss": 0.85,
+      "step": 465
+    },
+    {
+      "epoch": 0.11314395763119885,
+      "grad_norm": 1.75,
+      "learning_rate": 0.0002048180739797025,
+      "loss": 0.8685,
+      "step": 470
+    },
+    {
+      "epoch": 0.114347616754935,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.00020481205548977086,
+      "loss": 0.9002,
+      "step": 475
+    },
+    {
+      "epoch": 0.11555127587867116,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00020480596821948255,
+      "loss": 0.8456,
+      "step": 480
+    },
+    {
+      "epoch": 0.11675493500240731,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.0002047998121743007,
+      "loss": 0.8625,
+      "step": 485
+    },
+    {
+      "epoch": 0.11795859412614347,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.0002047935873597501,
+      "loss": 0.803,
+      "step": 490
+    },
+    {
+      "epoch": 0.11916225324987964,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00020478729378141735,
+      "loss": 0.839,
+      "step": 495
+    },
+    {
+      "epoch": 0.1203659123736158,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00020478093144495074,
+      "loss": 0.8499,
+      "step": 500
+    },
+    {
+      "epoch": 0.1203659123736158,
+      "eval_loss": 0.758273184299469,
+      "eval_runtime": 2.3484,
+      "eval_samples_per_second": 85.165,
+      "eval_steps_per_second": 85.165,
+      "step": 500
+    },
+    {
+      "epoch": 0.12156957149735195,
+      "grad_norm": 1.953125,
+      "learning_rate": 0.00020477450035606023,
+      "loss": 0.8103,
+      "step": 505
+    },
+    {
+      "epoch": 0.12277323062108811,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00020476800052051752,
+      "loss": 0.847,
+      "step": 510
+    },
+    {
+      "epoch": 0.12397688974482426,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00020476143194415598,
+      "loss": 0.8835,
+      "step": 515
+    },
+    {
+      "epoch": 0.12518054886856042,
+      "grad_norm": 1.9453125,
+      "learning_rate": 0.0002047547946328707,
+      "loss": 0.9029,
+      "step": 520
+    },
+    {
+      "epoch": 0.12638420799229658,
+      "grad_norm": 1.828125,
+      "learning_rate": 0.00020474808859261847,
+      "loss": 0.8808,
+      "step": 525
+    },
+    {
+      "epoch": 0.12758786711603273,
+      "grad_norm": 1.8984375,
+      "learning_rate": 0.0002047413138294177,
+      "loss": 0.8423,
+      "step": 530
+    },
+    {
+      "epoch": 0.1287915262397689,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.0002047344703493485,
+      "loss": 0.8329,
+      "step": 535
+    },
+    {
+      "epoch": 0.12999518536350504,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00020472755815855272,
+      "loss": 0.8337,
+      "step": 540
+    },
+    {
+      "epoch": 0.1311988444872412,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.00020472057726323377,
+      "loss": 0.8542,
+      "step": 545
+    },
+    {
+      "epoch": 0.13240250361097738,
+      "grad_norm": 1.890625,
+      "learning_rate": 0.0002047135276696568,
+      "loss": 0.8087,
+      "step": 550
+    },
+    {
+      "epoch": 0.13360616273471354,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00020470640938414854,
+      "loss": 0.8493,
+      "step": 555
+    },
+    {
+      "epoch": 0.1348098218584497,
+      "grad_norm": 1.75,
+      "learning_rate": 0.0002046992224130975,
+      "loss": 0.8346,
+      "step": 560
+    },
+    {
+      "epoch": 0.13601348098218585,
+      "grad_norm": 1.8671875,
+      "learning_rate": 0.00020469196676295366,
+      "loss": 0.8155,
+      "step": 565
+    },
+    {
+      "epoch": 0.137217140105922,
+      "grad_norm": 1.8046875,
+      "learning_rate": 0.00020468464244022876,
+      "loss": 0.8205,
+      "step": 570
+    },
+    {
+      "epoch": 0.13842079922965816,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020467724945149617,
+      "loss": 0.7741,
+      "step": 575
+    },
+    {
+      "epoch": 0.13962445835339432,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00020466978780339076,
+      "loss": 0.8312,
+      "step": 580
+    },
+    {
+      "epoch": 0.14082811747713048,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00020466225750260917,
+      "loss": 0.8201,
+      "step": 585
+    },
+    {
+      "epoch": 0.14203177660086663,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.00020465465855590962,
+      "loss": 0.8271,
+      "step": 590
+    },
+    {
+      "epoch": 0.1432354357246028,
+      "grad_norm": 1.90625,
+      "learning_rate": 0.00020464699097011186,
+      "loss": 0.827,
+      "step": 595
+    },
+    {
+      "epoch": 0.14443909484833894,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.0002046392547520973,
+      "loss": 0.8012,
+      "step": 600
+    },
+    {
+      "epoch": 0.1456427539720751,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00020463144990880895,
+      "loss": 0.8208,
+      "step": 605
+    },
+    {
+      "epoch": 0.14684641309581126,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00020462357644725136,
+      "loss": 0.8098,
+      "step": 610
+    },
+    {
+      "epoch": 0.1480500722195474,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.00020461563437449077,
+      "loss": 0.8341,
+      "step": 615
+    },
+    {
+      "epoch": 0.14925373134328357,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00020460762369765487,
+      "loss": 0.8568,
+      "step": 620
+    },
+    {
+      "epoch": 0.15045739046701975,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00020459954442393302,
+      "loss": 0.8193,
+      "step": 625
+    },
+    {
+      "epoch": 0.1516610495907559,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.000204591396560576,
+      "loss": 0.8497,
+      "step": 630
+    },
+    {
+      "epoch": 0.15286470871449206,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00020458318011489635,
+      "loss": 0.8303,
+      "step": 635
+    },
+    {
+      "epoch": 0.15406836783822822,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.000204574895094268,
+      "loss": 0.8073,
+      "step": 640
+    },
+    {
+      "epoch": 0.15527202696196438,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00020456654150612655,
+      "loss": 0.8241,
+      "step": 645
+    },
+    {
+      "epoch": 0.15647568608570053,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00020455811935796896,
+      "loss": 0.8294,
+      "step": 650
+    },
+    {
+      "epoch": 0.1576793452094367,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00020454962865735393,
+      "loss": 0.8556,
+      "step": 655
+    },
+    {
+      "epoch": 0.15888300433317284,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.00020454106941190147,
+      "loss": 0.8176,
+      "step": 660
+    },
+    {
+      "epoch": 0.160086663456909,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00020453244162929333,
+      "loss": 0.8422,
+      "step": 665
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00020452374531727255,
+      "loss": 0.8189,
+      "step": 670
+    },
+    {
+      "epoch": 0.1624939817043813,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00020451498048364386,
+      "loss": 0.7864,
+      "step": 675
+    },
+    {
+      "epoch": 0.16369764082811747,
+      "grad_norm": 1.8359375,
+      "learning_rate": 0.00020450614713627338,
+      "loss": 0.8094,
+      "step": 680
+    },
+    {
+      "epoch": 0.16490129995185362,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00020449724528308875,
+      "loss": 0.801,
+      "step": 685
+    },
+    {
+      "epoch": 0.16610495907558978,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00020448827493207904,
+      "loss": 0.8038,
+      "step": 690
+    },
+    {
+      "epoch": 0.16730861819932596,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.0002044792360912949,
+      "loss": 0.7928,
+      "step": 695
+    },
+    {
+      "epoch": 0.16851227732306212,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00020447012876884837,
+      "loss": 0.7884,
+      "step": 700
+    },
+    {
+      "epoch": 0.16971593644679828,
+      "grad_norm": 1.8203125,
+      "learning_rate": 0.00020446095297291296,
+      "loss": 0.8327,
+      "step": 705
+    },
+    {
+      "epoch": 0.17091959557053443,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.00020445170871172364,
+      "loss": 0.8146,
+      "step": 710
+    },
+    {
+      "epoch": 0.1721232546942706,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.00020444239599357682,
+      "loss": 0.8139,
+      "step": 715
+    },
+    {
+      "epoch": 0.17332691381800674,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00020443301482683037,
+      "loss": 0.7951,
+      "step": 720
+    },
+    {
+      "epoch": 0.1745305729417429,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00020442356521990356,
+      "loss": 0.853,
+      "step": 725
+    },
+    {
+      "epoch": 0.17573423206547906,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.0002044140471812771,
+      "loss": 0.8058,
+      "step": 730
+    },
+    {
+      "epoch": 0.1769378911892152,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.0002044044607194931,
+      "loss": 0.8169,
+      "step": 735
+    },
+    {
+      "epoch": 0.17814155031295137,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0002043948058431551,
+      "loss": 0.7831,
+      "step": 740
+    },
+    {
+      "epoch": 0.17934520943668752,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020438508256092805,
+      "loss": 0.7756,
+      "step": 745
+    },
+    {
+      "epoch": 0.18054886856042368,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00020437529088153825,
+      "loss": 0.7584,
+      "step": 750
+    },
+    {
+      "epoch": 0.18175252768415984,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.0002043654308137734,
+      "loss": 0.7844,
+      "step": 755
+    },
+    {
+      "epoch": 0.182956186807896,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00020435550236648256,
+      "loss": 0.7515,
+      "step": 760
+    },
+    {
+      "epoch": 0.18415984593163215,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00020434550554857626,
+      "loss": 0.7984,
+      "step": 765
+    },
+    {
+      "epoch": 0.18536350505536833,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00020433544036902625,
+      "loss": 0.7523,
+      "step": 770
+    },
+    {
+      "epoch": 0.1865671641791045,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00020432530683686567,
+      "loss": 0.8343,
+      "step": 775
+    },
+    {
+      "epoch": 0.18777082330284064,
+      "grad_norm": 1.796875,
+      "learning_rate": 0.00020431510496118907,
+      "loss": 0.791,
+      "step": 780
+    },
+    {
+      "epoch": 0.1889744824265768,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00020430483475115228,
+      "loss": 0.7973,
+      "step": 785
+    },
+    {
+      "epoch": 0.19017814155031296,
+      "grad_norm": 1.8125,
+      "learning_rate": 0.00020429449621597247,
+      "loss": 0.7755,
+      "step": 790
+    },
+    {
+      "epoch": 0.1913818006740491,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00020428408936492808,
+      "loss": 0.7479,
+      "step": 795
+    },
+    {
+      "epoch": 0.19258545979778527,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00020427361420735903,
+      "loss": 0.7886,
+      "step": 800
+    },
+    {
+      "epoch": 0.19378911892152142,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00020426307075266633,
+      "loss": 0.7919,
+      "step": 805
+    },
+    {
+      "epoch": 0.19499277804525758,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00020425245901031238,
+      "loss": 0.7854,
+      "step": 810
+    },
+    {
+      "epoch": 0.19619643716899374,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0002042417789898209,
+      "loss": 0.7661,
+      "step": 815
+    },
+    {
+      "epoch": 0.1974000962927299,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00020423103070077683,
+      "loss": 0.7862,
+      "step": 820
+    },
+    {
+      "epoch": 0.19860375541646605,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020422021415282643,
+      "loss": 0.7762,
+      "step": 825
+    },
+    {
+      "epoch": 0.1998074145402022,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00020420932935567715,
+      "loss": 0.7851,
+      "step": 830
+    },
+    {
+      "epoch": 0.20101107366393836,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.00020419837631909774,
+      "loss": 0.7898,
+      "step": 835
+    },
+    {
+      "epoch": 0.20221473278767452,
+      "grad_norm": 1.6640625,
+      "learning_rate": 0.00020418735505291815,
+      "loss": 0.7651,
+      "step": 840
+    },
+    {
+      "epoch": 0.2034183919114107,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.0002041762655670297,
+      "loss": 0.787,
+      "step": 845
+    },
+    {
+      "epoch": 0.20462205103514686,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00020416510787138475,
+      "loss": 0.7895,
+      "step": 850
+    },
+    {
+      "epoch": 0.205825710158883,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.000204153881975997,
+      "loss": 0.8062,
+      "step": 855
+    },
+    {
+      "epoch": 0.20702936928261917,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.0002041425878909413,
+      "loss": 0.781,
+      "step": 860
+    },
+    {
+      "epoch": 0.20823302840635532,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00020413122562635367,
+      "loss": 0.8269,
+      "step": 865
+    },
+    {
+      "epoch": 0.20943668753009148,
+      "grad_norm": 1.671875,
+      "learning_rate": 0.00020411979519243146,
+      "loss": 0.7963,
+      "step": 870
+    },
+    {
+      "epoch": 0.21064034665382764,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00020410829659943305,
+      "loss": 0.7691,
+      "step": 875
+    },
+    {
+      "epoch": 0.2118440057775638,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.000204096729857678,
+      "loss": 0.7997,
+      "step": 880
+    },
+    {
+      "epoch": 0.21304766490129995,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00020408509497754713,
+      "loss": 0.7765,
+      "step": 885
+    },
+    {
+      "epoch": 0.2142513240250361,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00020407339196948233,
+      "loss": 0.7658,
+      "step": 890
+    },
+    {
+      "epoch": 0.21545498314877226,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020406162084398666,
+      "loss": 0.7685,
+      "step": 895
+    },
+    {
+      "epoch": 0.21665864227250842,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00020404978161162437,
+      "loss": 0.7525,
+      "step": 900
+    },
+    {
+      "epoch": 0.21786230139624457,
+      "grad_norm": 1.75,
+      "learning_rate": 0.0002040378742830207,
+      "loss": 0.7878,
+      "step": 905
+    },
+    {
+      "epoch": 0.21906596051998073,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00020402589886886209,
+      "loss": 0.7852,
+      "step": 910
+    },
+    {
+      "epoch": 0.2202696196437169,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.00020401385537989608,
+      "loss": 0.7589,
+      "step": 915
+    },
+    {
+      "epoch": 0.22147327876745307,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.00020400174382693132,
+      "loss": 0.7792,
+      "step": 920
+    },
+    {
+      "epoch": 0.22267693789118922,
+      "grad_norm": 6.34375,
+      "learning_rate": 0.0002039895642208375,
+      "loss": 0.7764,
+      "step": 925
+    },
+    {
+      "epoch": 0.22388059701492538,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00020397731657254546,
+      "loss": 0.7667,
+      "step": 930
+    },
+    {
+      "epoch": 0.22508425613866154,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00020396500089304696,
+      "loss": 0.778,
+      "step": 935
+    },
+    {
+      "epoch": 0.2262879152623977,
+      "grad_norm": 1.765625,
+      "learning_rate": 0.00020395261719339503,
+      "loss": 0.7851,
+      "step": 940
+    },
+    {
+      "epoch": 0.22749157438613385,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00020394016548470353,
+      "loss": 0.7762,
+      "step": 945
+    },
+    {
+      "epoch": 0.22869523350987,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.0002039276457781475,
+      "loss": 0.7955,
+      "step": 950
+    },
+    {
+      "epoch": 0.22989889263360616,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00020391505808496293,
+      "loss": 0.7441,
+      "step": 955
+    },
+    {
+      "epoch": 0.23110255175734232,
+      "grad_norm": 1.7578125,
+      "learning_rate": 0.00020390240241644687,
+      "loss": 0.7631,
+      "step": 960
+    },
+    {
+      "epoch": 0.23230621088107847,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00020388967878395736,
+      "loss": 0.7693,
+      "step": 965
+    },
+    {
+      "epoch": 0.23350987000481463,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0002038768871989134,
+      "loss": 0.7661,
+      "step": 970
+    },
+    {
+      "epoch": 0.23471352912855079,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0002038640276727951,
+      "loss": 0.7397,
+      "step": 975
+    },
+    {
+      "epoch": 0.23591718825228694,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00020385110021714338,
+      "loss": 0.7375,
+      "step": 980
+    },
+    {
+      "epoch": 0.2371208473760231,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0002038381048435602,
+      "loss": 0.7431,
+      "step": 985
+    },
+    {
+      "epoch": 0.23832450649975928,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020382504156370848,
+      "loss": 0.7465,
+      "step": 990
+    },
+    {
+      "epoch": 0.23952816562349544,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.0002038119103893121,
+      "loss": 0.7611,
+      "step": 995
+    },
+    {
+      "epoch": 0.2407318247472316,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00020379871133215585,
+      "loss": 0.7453,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2407318247472316,
+      "eval_loss": 0.6567254662513733,
+      "eval_runtime": 2.3374,
+      "eval_samples_per_second": 85.567,
+      "eval_steps_per_second": 85.567,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00020378544440408542,
+      "loss": 0.7755,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2431391429947039,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00020377210961700743,
+      "loss": 0.7583,
+      "step": 1010
+    },
+    {
+      "epoch": 0.24434280211844006,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00020375870698288942,
+      "loss": 0.755,
+      "step": 1015
+    },
+    {
+      "epoch": 0.24554646124217622,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020374523651375983,
+      "loss": 0.7317,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24675012036591237,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0002037316982217079,
+      "loss": 0.7217,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24795377948964853,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020371809211888384,
+      "loss": 0.718,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24915743861338469,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.0002037044182174987,
+      "loss": 0.7659,
+      "step": 1035
+    },
+    {
+      "epoch": 0.25036109773712084,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00020369067652982426,
+      "loss": 0.7615,
+      "step": 1040
+    },
+    {
+      "epoch": 0.251564756860857,
+      "grad_norm": 1.6796875,
+      "learning_rate": 0.00020367686706819325,
+      "loss": 0.7083,
+      "step": 1045
+    },
+    {
+      "epoch": 0.25276841598459315,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0002036629898449993,
+      "loss": 0.7435,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2539720751083293,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00020364904487269662,
+      "loss": 0.7013,
+      "step": 1055
+    },
+    {
+      "epoch": 0.25517573423206547,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.00020363503216380046,
+      "loss": 0.7681,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2563793933558016,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00020362095173088668,
+      "loss": 0.7653,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2575830524795378,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020360680358659202,
+      "loss": 0.741,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25878671160327393,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.000203592587743614,
+      "loss": 0.773,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2599903707270101,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020357830421471082,
+      "loss": 0.8058,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26119402985074625,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.0002035639530127015,
+      "loss": 0.7148,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2623976889744824,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00020354953415046577,
+      "loss": 0.7497,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26360134809821856,
+      "grad_norm": 1.6171875,
+      "learning_rate": 0.00020353504764094403,
+      "loss": 0.7325,
+      "step": 1095
+    },
+    {
+      "epoch": 0.26480500722195477,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0002035204934971375,
+      "loss": 0.7436,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2660086663456909,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00020350587173210798,
+      "loss": 0.7258,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2672123254694271,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00020349118235897808,
+      "loss": 0.6951,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26841598459316324,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00020347642539093096,
+      "loss": 0.7554,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2696196437168994,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00020346160084121052,
+      "loss": 0.7085,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27082330284063555,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00020344670872312132,
+      "loss": 0.7583,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2720269619643717,
+      "grad_norm": 1.5859375,
+      "learning_rate": 0.00020343174905002853,
+      "loss": 0.7423,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27323062108810786,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00020341672183535794,
+      "loss": 0.771,
+      "step": 1135
+    },
+    {
+      "epoch": 0.274434280211844,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.000203401627092596,
+      "loss": 0.7473,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2756379393355802,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00020338646483528968,
+      "loss": 0.7536,
+      "step": 1145
+    },
+    {
+      "epoch": 0.27684159845931633,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00020337123507704668,
+      "loss": 0.7703,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2780452575830525,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00020335593783153515,
+      "loss": 0.7462,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27924891670678864,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.0002033405731124839,
+      "loss": 0.7671,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2804525758305248,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020332514093368217,
+      "loss": 0.7071,
+      "step": 1165
+    },
+    {
+      "epoch": 0.28165623495426095,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0002033096413089799,
+      "loss": 0.7375,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2828598940779971,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00020329407425228745,
+      "loss": 0.7355,
+      "step": 1175
+    },
+    {
+      "epoch": 0.28406355320173327,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020327843977757573,
+      "loss": 0.7515,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2852672123254694,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0002032627378988762,
+      "loss": 0.7331,
+      "step": 1185
+    },
+    {
+      "epoch": 0.2864708714492056,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020324696863028072,
+      "loss": 0.7656,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28767453057294173,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00020323113198594169,
+      "loss": 0.7379,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2888781896966779,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.000203215227980072,
+      "loss": 0.7435,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29008184882041405,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00020319925662694497,
+      "loss": 0.7092,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2912855079441502,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0002031832179408943,
+      "loss": 0.7432,
+      "step": 1210
+    },
+    {
+      "epoch": 0.29248916706788636,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.0002031671119363142,
+      "loss": 0.7442,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2936928261916225,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020315093862765927,
+      "loss": 0.7407,
+      "step": 1220
+    },
+    {
+      "epoch": 0.29489648531535867,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020313469802944455,
+      "loss": 0.7336,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2961001444390948,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.0002031183901562454,
+      "loss": 0.7458,
+      "step": 1230
+    },
+    {
+      "epoch": 0.297303803562831,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020310201502269759,
+      "loss": 0.7132,
+      "step": 1235
+    },
+    {
+      "epoch": 0.29850746268656714,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00020308557264349722,
+      "loss": 0.7207,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29971112181030335,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00020306906303340088,
+      "loss": 0.7316,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3009147809340395,
+      "grad_norm": 1.5078125,
+      "learning_rate": 0.00020305248620722529,
+      "loss": 0.734,
+      "step": 1250
+    },
+    {
+      "epoch": 0.30211844005777566,
+      "grad_norm": 1.578125,
+      "learning_rate": 0.00020303584217984768,
+      "loss": 0.7246,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3033220991815118,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.0002030191309662054,
+      "loss": 0.7322,
+      "step": 1260
+    },
+    {
+      "epoch": 0.304525758305248,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0002030023525812963,
+      "loss": 0.7073,
+      "step": 1265
+    },
+    {
+      "epoch": 0.30572941742898413,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0002029855070401784,
+      "loss": 0.7307,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3069330765527203,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00020296859435797,
+      "loss": 0.72,
+      "step": 1275
+    },
+    {
+      "epoch": 0.30813673567645644,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00020295161454984966,
+      "loss": 0.7064,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3093403948001926,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020293456763105622,
+      "loss": 0.7452,
+      "step": 1285
+    },
+    {
+      "epoch": 0.31054405392392875,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020291745361688868,
+      "loss": 0.73,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3117477130476649,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0002029002725227063,
+      "loss": 0.7469,
+      "step": 1295
+    },
+    {
+      "epoch": 0.31295137217140107,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00020288302436392855,
+      "loss": 0.6994,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3141550312951372,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.0002028657091560351,
+      "loss": 0.7116,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3153586904188734,
+      "grad_norm": 1.7265625,
+      "learning_rate": 0.00020284832691456575,
+      "loss": 0.726,
+      "step": 1310
+    },
+    {
+      "epoch": 0.31656234954260953,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00020283087765512044,
+      "loss": 0.7522,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3177660086663457,
+      "grad_norm": 1.6015625,
+      "learning_rate": 0.00020281336139335934,
+      "loss": 0.7305,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31896966779008185,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020279577814500275,
+      "loss": 0.7365,
+      "step": 1325
+    },
+    {
+      "epoch": 0.320173326913818,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00020277812792583098,
+      "loss": 0.7176,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32137698603755416,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020276041075168453,
+      "loss": 0.7182,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00020274262663846397,
+      "loss": 0.6926,
+      "step": 1340
+    },
+    {
+      "epoch": 0.32378430428502647,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00020272477560212996,
+      "loss": 0.6864,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3249879634087626,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0002027068576587032,
+      "loss": 0.6841,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3261916225324988,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00020268887282426444,
+      "loss": 0.7243,
+      "step": 1355
+    },
+    {
+      "epoch": 0.32739528165623494,
+      "grad_norm": 1.625,
+      "learning_rate": 0.00020267082111495452,
+      "loss": 0.71,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3285989407799711,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020265270254697413,
+      "loss": 0.7578,
+      "step": 1365
+    },
+    {
+      "epoch": 0.32980259990370725,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0002026345171365842,
+      "loss": 0.6661,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3310062590274434,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00020261626490010543,
+      "loss": 0.7308,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33220991815117956,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00020259794585391868,
+      "loss": 0.6862,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3334135772749157,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00020257956001446464,
+      "loss": 0.6837,
+      "step": 1385
+    },
+    {
+      "epoch": 0.33461723639865193,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00020256110739824394,
+      "loss": 0.7164,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3358208955223881,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.0002025425880218173,
+      "loss": 0.7228,
+      "step": 1395
+    },
+    {
+      "epoch": 0.33702455464612424,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00020252400190180512,
+      "loss": 0.7357,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3382282137698604,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00020250534905488793,
+      "loss": 0.6781,
+      "step": 1405
+    },
+    {
+      "epoch": 0.33943187289359655,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00020248662949780595,
+      "loss": 0.713,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3406355320173327,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020246784324735937,
+      "loss": 0.716,
+      "step": 1415
+    },
+    {
+      "epoch": 0.34183919114106887,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0002024489903204083,
+      "loss": 0.7322,
+      "step": 1420
+    },
+    {
+      "epoch": 0.343042850264805,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0002024300707338725,
+      "loss": 0.731,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3442465093885412,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00020241108450473174,
+      "loss": 0.7077,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34545016851227733,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0002023920316500255,
+      "loss": 0.7216,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3466538276360135,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00020237291218685308,
+      "loss": 0.7019,
+      "step": 1440
+    },
+    {
+      "epoch": 0.34785748675974965,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0002023537261323736,
+      "loss": 0.6703,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3490611458834858,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00020233447350380586,
+      "loss": 0.6877,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35026480500722196,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00020231515431842845,
+      "loss": 0.6752,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3514684641309581,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00020229576859357966,
+      "loss": 0.6921,
+      "step": 1460
+    },
+    {
+      "epoch": 0.35267212325469427,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00020227631634665762,
+      "loss": 0.7102,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3538757823784304,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00020225679759512002,
+      "loss": 0.739,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3550794415021666,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00020223721235648428,
+      "loss": 0.7211,
+      "step": 1475
+    },
+    {
+      "epoch": 0.35628310062590274,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00020221756064832752,
+      "loss": 0.7093,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3574867597496389,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00020219784248828645,
+      "loss": 0.6881,
+      "step": 1485
+    },
+    {
+      "epoch": 0.35869041887337505,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0002021780578940575,
+      "loss": 0.6753,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3598940779971112,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0002021582068833967,
+      "loss": 0.6865,
+      "step": 1495
+    },
+    {
+      "epoch": 0.36109773712084736,
+      "grad_norm": 1.6328125,
+      "learning_rate": 0.00020213828947411962,
+      "loss": 0.6809,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36109773712084736,
+      "eval_loss": 0.6072394847869873,
+      "eval_runtime": 2.3749,
+      "eval_samples_per_second": 84.214,
+      "eval_steps_per_second": 84.214,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3623013962445835,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020211830568410148,
+      "loss": 0.7175,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3635050553683197,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00020209825553127708,
+      "loss": 0.701,
+      "step": 1510
+    },
+    {
+      "epoch": 0.36470871449205583,
+      "grad_norm": 1.515625,
+      "learning_rate": 0.00020207813903364076,
+      "loss": 0.7149,
+      "step": 1515
+    },
+    {
+      "epoch": 0.365912373615792,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0002020579562092464,
+      "loss": 0.6791,
+      "step": 1520
+    },
+    {
+      "epoch": 0.36711603273952814,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0002020377070762074,
+      "loss": 0.6658,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3683196918632643,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00020201739165269666,
+      "loss": 0.7309,
+      "step": 1530
+    },
+    {
+      "epoch": 0.36952335098700045,
+      "grad_norm": 1.5703125,
+      "learning_rate": 0.00020199700995694666,
+      "loss": 0.7247,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37072701011073667,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.00020197656200724923,
+      "loss": 0.7183,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3719306692344728,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00020195604782195578,
+      "loss": 0.6884,
+      "step": 1545
+    },
+    {
+      "epoch": 0.373134328358209,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00020193546741947702,
+      "loss": 0.691,
+      "step": 1550
+    },
+    {
+      "epoch": 0.37433798748194513,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00020191482081828325,
+      "loss": 0.7117,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3755416466056813,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00020189410803690406,
+      "loss": 0.7099,
+      "step": 1560
+    },
+    {
+      "epoch": 0.37674530572941745,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00020187332909392852,
+      "loss": 0.6907,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3779489648531536,
+      "grad_norm": 1.375,
+      "learning_rate": 0.000201852484008005,
+      "loss": 0.7284,
+      "step": 1570
+    },
+    {
+      "epoch": 0.37915262397688976,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00020183157279784128,
+      "loss": 0.7115,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3803562831006259,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0002018105954822045,
+      "loss": 0.6897,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38155994222436207,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00020178955207992102,
+      "loss": 0.6898,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3827636013480982,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00020176844260987668,
+      "loss": 0.7069,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3839672604718344,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0002017472670910165,
+      "loss": 0.6938,
+      "step": 1595
+    },
+    {
+      "epoch": 0.38517091959557054,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00020172602554234474,
+      "loss": 0.7031,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3863745787193067,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00020170471798292507,
+      "loss": 0.6911,
+      "step": 1605
+    },
+    {
+      "epoch": 0.38757823784304285,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020168334443188023,
+      "loss": 0.6918,
+      "step": 1610
+    },
+    {
+      "epoch": 0.388781896966779,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00020166190490839234,
+      "loss": 0.7089,
+      "step": 1615
+    },
+    {
+      "epoch": 0.38998555609051516,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00020164039943170264,
+      "loss": 0.6818,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3911892152142513,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00020161882802111154,
+      "loss": 0.752,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3923928743379875,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00020159719069597865,
+      "loss": 0.6934,
+      "step": 1630
+    },
+    {
+      "epoch": 0.39359653346172363,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0002015754874757228,
+      "loss": 0.6998,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3948001925854598,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020155371837982187,
+      "loss": 0.7089,
+      "step": 1640
+    },
+    {
+      "epoch": 0.39600385170919594,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0002015318834278129,
+      "loss": 0.6729,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3972075108329321,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00020150998263929206,
+      "loss": 0.6835,
+      "step": 1650
+    },
+    {
+      "epoch": 0.39841116995666825,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00020148801603391456,
+      "loss": 0.7027,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3996148290804044,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00020146598363139466,
+      "loss": 0.6962,
+      "step": 1660
+    },
+    {
+      "epoch": 0.40081848820414057,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00020144388545150573,
+      "loss": 0.6746,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4020221473278767,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00020142172151408014,
+      "loss": 0.6756,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 1.6484375,
+      "learning_rate": 0.00020139949183900933,
+      "loss": 0.685,
+      "step": 1675
+    },
+    {
+      "epoch": 0.40442946557534903,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0002013771964462436,
+      "loss": 0.7223,
+      "step": 1680
+    },
+    {
+      "epoch": 0.40563312469908525,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.00020135483535579243,
+      "loss": 0.6715,
+      "step": 1685
+    },
+    {
+      "epoch": 0.4068367838228214,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00020133240858772404,
+      "loss": 0.7027,
+      "step": 1690
+    },
+    {
+      "epoch": 0.40804044294655756,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0002013099161621658,
+      "loss": 0.6796,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4092441020702937,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00020128735809930382,
+      "loss": 0.6772,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41044776119402987,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00020126473441938332,
+      "loss": 0.7071,
+      "step": 1705
+    },
+    {
+      "epoch": 0.411651420317766,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00020124204514270817,
+      "loss": 0.6642,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4128550794415022,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00020121929028964132,
+      "loss": 0.6734,
+      "step": 1715
+    },
+    {
+      "epoch": 0.41405873856523834,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00020119646988060448,
+      "loss": 0.696,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4152623976889745,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0002011735839360782,
+      "loss": 0.6878,
+      "step": 1725
+    },
+    {
+      "epoch": 0.41646605681271065,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0002011506324766018,
+      "loss": 0.654,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4176697159364468,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.0002011276155227735,
+      "loss": 0.6922,
+      "step": 1735
+    },
+    {
+      "epoch": 0.41887337506018296,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020110453309525028,
+      "loss": 0.6812,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4200770341839191,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00020108138521474778,
+      "loss": 0.6771,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4212806933076553,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.0002010581719020405,
+      "loss": 0.6896,
+      "step": 1750
+    },
+    {
+      "epoch": 0.42248435243139143,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00020103489317796155,
+      "loss": 0.6948,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4236880115551276,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00020101154906340286,
+      "loss": 0.6696,
+      "step": 1760
+    },
+    {
+      "epoch": 0.42489167067886374,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00020098813957931496,
+      "loss": 0.6867,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4260953298025999,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00020096466474670712,
+      "loss": 0.6711,
+      "step": 1770
+    },
+    {
+      "epoch": 0.42729898892633605,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00020094112458664715,
+      "loss": 0.6475,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4285026480500722,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00020091751912026164,
+      "loss": 0.6793,
+      "step": 1780
+    },
+    {
+      "epoch": 0.42970630717380837,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00020089384836873567,
+      "loss": 0.701,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4309099662975445,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00020087011235331293,
+      "loss": 0.7153,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4321136254212807,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0002008463110952957,
+      "loss": 0.6996,
+      "step": 1795
+    },
+    {
+      "epoch": 0.43331728454501683,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00020082244461604485,
+      "loss": 0.6874,
+      "step": 1800
+    },
+    {
+      "epoch": 0.434520943668753,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00020079851293697973,
+      "loss": 0.7237,
+      "step": 1805
+    },
+    {
+      "epoch": 0.43572460279248915,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00020077451607957821,
+      "loss": 0.6801,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4369282619162253,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0002007504540653767,
+      "loss": 0.6645,
+      "step": 1815
+    },
+    {
+      "epoch": 0.43813192103996146,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.00020072632691597,
+      "loss": 0.6698,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4393355801636976,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00020070213465301146,
+      "loss": 0.6834,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4405392392874338,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0002006778772982128,
+      "loss": 0.6623,
+      "step": 1830
+    },
+    {
+      "epoch": 0.44174289841117,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00020065355487334418,
+      "loss": 0.689,
+      "step": 1835
+    },
+    {
+      "epoch": 0.44294655753490614,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00020062916740023427,
+      "loss": 0.7091,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4441502166586423,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00020060471490076988,
+      "loss": 0.6626,
+      "step": 1845
+    },
+    {
+      "epoch": 0.44535387578237845,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00020058019739689634,
+      "loss": 0.6956,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4465575349061146,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00020055561491061732,
+      "loss": 0.6821,
+      "step": 1855
+    },
+    {
+      "epoch": 0.44776119402985076,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00020053096746399477,
+      "loss": 0.6926,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4489648531535869,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00020050625507914897,
+      "loss": 0.6317,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4501685122773231,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0002004814777782584,
+      "loss": 0.694,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45137217140105923,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00020045663558355994,
+      "loss": 0.686,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4525758305247954,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00020043172851734854,
+      "loss": 0.6752,
+      "step": 1880
+    },
+    {
+      "epoch": 0.45377948964853154,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00020040675660197755,
+      "loss": 0.6888,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4549831487722677,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00020038171985985835,
+      "loss": 0.6399,
+      "step": 1890
+    },
+    {
+      "epoch": 0.45618680789600385,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00020035661831346063,
+      "loss": 0.6438,
+      "step": 1895
+    },
+    {
+      "epoch": 0.45739046701974,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00020033145198531218,
+      "loss": 0.7266,
+      "step": 1900
+    },
+    {
+      "epoch": 0.45859412614347617,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00020030622089799892,
+      "loss": 0.6877,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4597977852672123,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00020028092507416492,
+      "loss": 0.6844,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4610014443909485,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00020025556453651236,
+      "loss": 0.6749,
+      "step": 1915
+    },
+    {
+      "epoch": 0.46220510351468463,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.0002002301393078014,
+      "loss": 0.6717,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4634087626384208,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00020020464941085042,
+      "loss": 0.7072,
+      "step": 1925
+    },
+    {
+      "epoch": 0.46461242176215695,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0002001790948685357,
+      "loss": 0.6732,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4658160808858931,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0002001534757037916,
+      "loss": 0.6695,
+      "step": 1935
+    },
+    {
+      "epoch": 0.46701974000962926,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00020012779193961046,
+      "loss": 0.6712,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4682233991333654,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00020010204359904263,
+      "loss": 0.667,
+      "step": 1945
+    },
+    {
+      "epoch": 0.46942705825710157,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00020007623070519633,
+      "loss": 0.6518,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4706307173808377,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00020005035328123782,
+      "loss": 0.6548,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4718343765045739,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0002000244113503912,
+      "loss": 0.6339,
+      "step": 1960
+    },
+    {
+      "epoch": 0.47303803562831004,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0001999984049359385,
+      "loss": 0.6876,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4742416947520462,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001999723340612196,
+      "loss": 0.6948,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4754453538757824,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001999461987496322,
+      "loss": 0.6809,
+      "step": 1975
+    },
+    {
+      "epoch": 0.47664901299951856,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019991999902463194,
+      "loss": 0.6741,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4778526721232547,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019989373490973211,
+      "loss": 0.6463,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4790563312469909,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019986740642850395,
+      "loss": 0.6633,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48025999037072703,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.0001998410136045764,
+      "loss": 0.6469,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4814636494944632,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00019981455646163606,
+      "loss": 0.6366,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4814636494944632,
+      "eval_loss": 0.5777117609977722,
+      "eval_runtime": 2.3295,
+      "eval_samples_per_second": 85.854,
+      "eval_steps_per_second": 85.854,
+      "step": 2000
+    },
+    {
+      "epoch": 0.48266730861819934,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00019978803502342741,
+      "loss": 0.6697,
+      "step": 2005
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001997614493137525,
+      "loss": 0.6508,
+      "step": 2010
+    },
+    {
+      "epoch": 0.48507462686567165,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019973479935647117,
+      "loss": 0.6693,
+      "step": 2015
+    },
+    {
+      "epoch": 0.4862782859894078,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001997080851755008,
+      "loss": 0.6508,
+      "step": 2020
+    },
+    {
+      "epoch": 0.48748194511314397,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019968130679481652,
+      "loss": 0.6663,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4886856042368801,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019965446423845102,
+      "loss": 0.6342,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4898892633606163,
+      "grad_norm": 1.4921875,
+      "learning_rate": 0.00019962755753049458,
+      "loss": 0.6939,
+      "step": 2035
+    },
+    {
+      "epoch": 0.49109292248435243,
+      "grad_norm": 1.640625,
+      "learning_rate": 0.0001996005866950951,
+      "loss": 0.6656,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4922965816080886,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.000199573551756458,
+      "loss": 0.643,
+      "step": 2045
+    },
+    {
+      "epoch": 0.49350024073182475,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001995464527388462,
+      "loss": 0.6738,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4947038998555609,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001995192896665802,
+      "loss": 0.6455,
+      "step": 2055
+    },
+    {
+      "epoch": 0.49590755897929706,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00019949206256403794,
+      "loss": 0.6361,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4971112181030332,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.0001994647714556548,
+      "loss": 0.6584,
+      "step": 2065
+    },
+    {
+      "epoch": 0.49831487722676937,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019943741636592374,
+      "loss": 0.6715,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4995185363505055,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0001994099973193949,
+      "loss": 0.6349,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5007221954742417,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00019938251434067607,
+      "loss": 0.6497,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5019258545979779,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00019935496745443222,
+      "loss": 0.6906,
+      "step": 2085
+    },
+    {
+      "epoch": 0.503129513721714,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001993273566853858,
+      "loss": 0.6885,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5043331728454502,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019929968205831657,
+      "loss": 0.6248,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5055368319691863,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019927194359806158,
+      "loss": 0.6865,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5067404910929225,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019924414132951513,
+      "loss": 0.6369,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5079441502166586,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00019921627527762884,
+      "loss": 0.6639,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5091478093403948,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001991883454674116,
+      "loss": 0.6742,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5103514684641309,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00019916035192392943,
+      "loss": 0.6588,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5115551275878671,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019913229467230566,
+      "loss": 0.6904,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5127587867116032,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019910417373772066,
+      "loss": 0.6035,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5139624458353395,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001990759891454121,
+      "loss": 0.6288,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5151661049590756,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019904774092067463,
+      "loss": 0.6856,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5163697640828118,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00019901942908886021,
+      "loss": 0.6526,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5175734232065479,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019899105367537765,
+      "loss": 0.6428,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5187770823302841,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.00019896261470569293,
+      "loss": 0.6646,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5199807414540202,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019893411220532915,
+      "loss": 0.6503,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5211844005777564,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00019890554619986632,
+      "loss": 0.671,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5223880597014925,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001988769167149415,
+      "loss": 0.6784,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5235917188252287,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019884822377624862,
+      "loss": 0.6494,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5247953779489648,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001988194674095387,
+      "loss": 0.6571,
+      "step": 2180
+    },
+    {
+      "epoch": 0.525999037072701,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001987906476406196,
+      "loss": 0.6686,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5272026961964371,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019876176449535614,
+      "loss": 0.6523,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5284063553201733,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001987328179996699,
+      "loss": 0.6585,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5296100144439095,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001987038081795394,
+      "loss": 0.6549,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5308136735676456,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00019867473506100003,
+      "loss": 0.6718,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5320173326913819,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019864559867014398,
+      "loss": 0.6528,
+      "step": 2210
+    },
+    {
+      "epoch": 0.533220991815118,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001986163990331201,
+      "loss": 0.6411,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5344246509388542,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.0001985871361761341,
+      "loss": 0.6726,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5356283100625903,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00019855781012544842,
+      "loss": 0.6683,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5368319691863265,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00019852842090738225,
+      "loss": 0.6499,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5380356283100626,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019849896854831143,
+      "loss": 0.679,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5392392874337988,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019846945307466847,
+      "loss": 0.6388,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5404429465575349,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019843987451294244,
+      "loss": 0.6282,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5416466056812711,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001984102328896792,
+      "loss": 0.6185,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5428502648050072,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00019838052823148105,
+      "loss": 0.6518,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5440539239287434,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.000198350760565007,
+      "loss": 0.6655,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5452575830524795,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019832092991697248,
+      "loss": 0.6806,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5464612421762157,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001982910363141495,
+      "loss": 0.6591,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5476649012999518,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.0001982610797833666,
+      "loss": 0.648,
+      "step": 2275
+    },
+    {
+      "epoch": 0.548868560423688,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00019823106035150875,
+      "loss": 0.6801,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5500722195474241,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019820097804551742,
+      "loss": 0.6577,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5512758786711603,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001981708328923904,
+      "loss": 0.6477,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5524795377948964,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.000198140624919182,
+      "loss": 0.637,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5536831969186327,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0001981103541530029,
+      "loss": 0.6511,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5548868560423688,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019808002062102003,
+      "loss": 0.665,
+      "step": 2305
+    },
+    {
+      "epoch": 0.556090515166105,
+      "grad_norm": 1.5234375,
+      "learning_rate": 0.00019804962435045678,
+      "loss": 0.6343,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5572941742898411,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.00019801916536859275,
+      "loss": 0.6293,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5584978334135773,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019798864370276388,
+      "loss": 0.6783,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5597014925373134,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019795805938036234,
+      "loss": 0.6325,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5609051516610496,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019792741242883656,
+      "loss": 0.6406,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5621088107847857,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019789670287569108,
+      "loss": 0.6101,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5633124699085219,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019786593074848675,
+      "loss": 0.6657,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019783509607484054,
+      "loss": 0.6452,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5657197881559942,
+      "grad_norm": 1.4296875,
+      "learning_rate": 0.00019780419888242555,
+      "loss": 0.6516,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5669234472797304,
+      "grad_norm": 1.4140625,
+      "learning_rate": 0.0001977732391989709,
+      "loss": 0.6181,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5681271064034665,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019774221705226195,
+      "loss": 0.6215,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5693307655272027,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019771113247014,
+      "loss": 0.6163,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5705344246509388,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.0001976799854805024,
+      "loss": 0.6159,
+      "step": 2370
+    },
+    {
+      "epoch": 0.571738083774675,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019764877611130261,
+      "loss": 0.6381,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5729417428984112,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019761750439054997,
+      "loss": 0.6282,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5741454020221474,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019758617034630973,
+      "loss": 0.6404,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5753490611458835,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019755477400670323,
+      "loss": 0.6867,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5765527202696197,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001975233153999076,
+      "loss": 0.6404,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5777563793933558,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0001974917945541559,
+      "loss": 0.6496,
+      "step": 2400
+    },
+    {
+      "epoch": 0.578960038517092,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019746021149773702,
+      "loss": 0.6515,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5801636976408281,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019742856625899573,
+      "loss": 0.6092,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5813673567645643,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019739685886633254,
+      "loss": 0.6167,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5825710158883004,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001973650893482038,
+      "loss": 0.671,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5837746750120366,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00019733325773312158,
+      "loss": 0.6301,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5849783341357727,
+      "grad_norm": 1.6953125,
+      "learning_rate": 0.00019730136404965366,
+      "loss": 0.6606,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5861819932595089,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00019726940832642362,
+      "loss": 0.6323,
+      "step": 2435
+    },
+    {
+      "epoch": 0.587385652383245,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019723739059211056,
+      "loss": 0.6246,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5885893115069812,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019720531087544945,
+      "loss": 0.6423,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5897929706307173,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019717316920523067,
+      "loss": 0.6505,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5909966297544536,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001971409656103003,
+      "loss": 0.6303,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5922002888781897,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019710870011956005,
+      "loss": 0.6268,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5934039480019259,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019707637276196704,
+      "loss": 0.64,
+      "step": 2465
+    },
+    {
+      "epoch": 0.594607607125662,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.0001970439835665341,
+      "loss": 0.6486,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5958112662493982,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019701153256232933,
+      "loss": 0.6377,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5970149253731343,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019697901977847645,
+      "loss": 0.6665,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5982185844968705,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.0001969464452441547,
+      "loss": 0.7109,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5994222436206067,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019691380898859853,
+      "loss": 0.6202,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6006259027443428,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001968811110410979,
+      "loss": 0.6243,
+      "step": 2495
+    },
+    {
+      "epoch": 0.601829561868079,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019684835143099816,
+      "loss": 0.6391,
+      "step": 2500
+    },
+    {
+      "epoch": 0.601829561868079,
+      "eval_loss": 0.5394683480262756,
+      "eval_runtime": 2.3314,
+      "eval_samples_per_second": 85.786,
+      "eval_steps_per_second": 85.786,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6030332209918151,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019681553018769993,
+      "loss": 0.6186,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6042368801155513,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001967826473406592,
+      "loss": 0.6417,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6054405392392874,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00019674970291938722,
+      "loss": 0.6211,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6066441983630236,
+      "grad_norm": 1.3828125,
+      "learning_rate": 0.00019671669695345054,
+      "loss": 0.6473,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6078478574867597,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019668362947247085,
+      "loss": 0.6459,
+      "step": 2525
+    },
+    {
+      "epoch": 0.609051516610496,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019665050050612516,
+      "loss": 0.6277,
+      "step": 2530
+    },
+    {
+      "epoch": 0.610255175734232,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019661731008414557,
+      "loss": 0.6127,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6114588348579683,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001965840582363194,
+      "loss": 0.6143,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6126624939817044,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001965507449924891,
+      "loss": 0.6354,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6138661531054406,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001965173703825521,
+      "loss": 0.6347,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6150698122291767,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019648393443646108,
+      "loss": 0.6147,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6162734713529129,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019645043718422364,
+      "loss": 0.6159,
+      "step": 2560
+    },
+    {
+      "epoch": 0.617477130476649,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019641687865590252,
+      "loss": 0.6448,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6186807896003852,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001963832588816153,
+      "loss": 0.6239,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6198844487241213,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019634957789153458,
+      "loss": 0.6028,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6210881078478575,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.000196315835715888,
+      "loss": 0.5975,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6222917669715936,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019628203238495799,
+      "loss": 0.66,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6234954260953298,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00019624816792908193,
+      "loss": 0.6242,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6246990852190659,
+      "grad_norm": 1.3671875,
+      "learning_rate": 0.00019621424237865206,
+      "loss": 0.6346,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6259027443428021,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019618025576411535,
+      "loss": 0.626,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6271064034665382,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019614620811597374,
+      "loss": 0.6328,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6283100625902744,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001961120994647838,
+      "loss": 0.6538,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6295137217140105,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.0001960779298411569,
+      "loss": 0.6415,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6307173808377468,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019604369927575917,
+      "loss": 0.6209,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6319210399614829,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019600940779931135,
+      "loss": 0.6679,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6331246990852191,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019597505544258895,
+      "loss": 0.6373,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6343283582089553,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019594064223642205,
+      "loss": 0.6236,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6355320173326914,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019590616821169534,
+      "loss": 0.669,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6367356764564276,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00019587163339934808,
+      "loss": 0.6373,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6379393355801637,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019583703783037414,
+      "loss": 0.6248,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6391429947038999,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.00019580238153582186,
+      "loss": 0.6446,
+      "step": 2655
+    },
+    {
+      "epoch": 0.640346653827636,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001957676645467942,
+      "loss": 0.6258,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6415503129513722,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019573288689444837,
+      "loss": 0.6269,
+      "step": 2665
+    },
+    {
+      "epoch": 0.6427539720751083,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019569804860999622,
+      "loss": 0.6046,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6439576311988445,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00019566314972470394,
+      "loss": 0.6218,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00019562819026989206,
+      "loss": 0.6336,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6463649494463168,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.0001955931702769356,
+      "loss": 0.6342,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6475686085700529,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.0001955580897772638,
+      "loss": 0.6243,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6487722676937892,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019552294880236022,
+      "loss": 0.618,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6499759268175253,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019548774738376275,
+      "loss": 0.6258,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6511795859412615,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019545248555306345,
+      "loss": 0.6808,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6523832450649976,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019541716334190866,
+      "loss": 0.6091,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6535869041887338,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019538178078199885,
+      "loss": 0.6413,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6547905633124699,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001953463379050887,
+      "loss": 0.6222,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6559942224362061,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019531083474298702,
+      "loss": 0.6199,
+      "step": 2725
+    },
+    {
+      "epoch": 0.6571978815599422,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.00019527527132755668,
+      "loss": 0.6621,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6584015406836784,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019523964769071475,
+      "loss": 0.6661,
+      "step": 2735
+    },
+    {
+      "epoch": 0.6596051998074145,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019520396386443214,
+      "loss": 0.6169,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6608088589311507,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001951682198807339,
+      "loss": 0.5924,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6620125180548868,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019513241577169915,
+      "loss": 0.6339,
+      "step": 2750
+    },
+    {
+      "epoch": 0.663216177178623,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019509655156946083,
+      "loss": 0.6463,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6644198363023591,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019506062730620583,
+      "loss": 0.6212,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6656234954260953,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019502464301417502,
+      "loss": 0.6358,
+      "step": 2765
+    },
+    {
+      "epoch": 0.6668271545498314,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019498859872566307,
+      "loss": 0.6378,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6680308136735676,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019495249447301855,
+      "loss": 0.6233,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6692344727973039,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019491633028864382,
+      "loss": 0.6398,
+      "step": 2780
+    },
+    {
+      "epoch": 0.67043813192104,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.000194880106204995,
+      "loss": 0.6337,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6716417910447762,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.000194843822254582,
+      "loss": 0.6026,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6728454501685123,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001948074784699685,
+      "loss": 0.6318,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6740491092922485,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019477107488377176,
+      "loss": 0.6322,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6752527684159846,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019473461152866281,
+      "loss": 0.614,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6764564275397208,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00019469808843736633,
+      "loss": 0.6345,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6776600866634569,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00019466150564266053,
+      "loss": 0.6519,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6788637457871931,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019462486317737727,
+      "loss": 0.6325,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6800674049109292,
+      "grad_norm": 1.5625,
+      "learning_rate": 0.0001945881610744019,
+      "loss": 0.62,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6812710640346654,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019455139936667339,
+      "loss": 0.6023,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6824747231584015,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00019451457808718407,
+      "loss": 0.6236,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6836783822821377,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019447769726897986,
+      "loss": 0.6382,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6848820414058738,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019444075694516008,
+      "loss": 0.6218,
+      "step": 2845
+    },
+    {
+      "epoch": 0.68608570052961,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019440375714887734,
+      "loss": 0.6166,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6872893596533461,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00019436669791333787,
+      "loss": 0.6012,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6884930187770824,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019432957927180094,
+      "loss": 0.6195,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6896966779008185,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019429240125757943,
+      "loss": 0.6233,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6909003370245547,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019425516390403922,
+      "loss": 0.6175,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6921039961482908,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019421786724459973,
+      "loss": 0.6079,
+      "step": 2875
+    },
+    {
+      "epoch": 0.693307655272027,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00019418051131273335,
+      "loss": 0.6204,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6945113143957631,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019414309614196586,
+      "loss": 0.6401,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6957149735194993,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001941056217658761,
+      "loss": 0.6285,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6969186326432354,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.0001940680882180961,
+      "loss": 0.6413,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6981222917669716,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00019403049553231086,
+      "loss": 0.5924,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6993259508907077,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001939928437422587,
+      "loss": 0.6299,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7005296100144439,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00019395513288173073,
+      "loss": 0.5845,
+      "step": 2910
+    },
+    {
+      "epoch": 0.70173326913818,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001939173629845712,
+      "loss": 0.6312,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7029369282619162,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001938795340846774,
+      "loss": 0.5586,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7041405873856523,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019384164621599948,
+      "loss": 0.633,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7053442465093885,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001938036994125405,
+      "loss": 0.5886,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7065479056331248,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019376569370835648,
+      "loss": 0.6014,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7077515647568609,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00019372762913755624,
+      "loss": 0.5822,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7089552238805971,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019368950573430148,
+      "loss": 0.634,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7101588830043332,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.0001936513235328067,
+      "loss": 0.6208,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7113625421280694,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001936130825673391,
+      "loss": 0.6068,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7125662012518055,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001935747828722187,
+      "loss": 0.5787,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7137698603755417,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001935364244818182,
+      "loss": 0.6003,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7149735194992778,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019349800743056296,
+      "loss": 0.6187,
+      "step": 2970
+    },
+    {
+      "epoch": 0.716177178623014,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.000193459531752931,
+      "loss": 0.5991,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7173808377467501,
+      "grad_norm": 1.25,
+      "learning_rate": 0.000193420997483453,
+      "loss": 0.6105,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7185844968704863,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019338240465671208,
+      "loss": 0.6234,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7197881559942224,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019334375330734408,
+      "loss": 0.6185,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7209918151179586,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019330504347003726,
+      "loss": 0.6016,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7221954742416947,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019326627517953247,
+      "loss": 0.5947,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7221954742416947,
+      "eval_loss": 0.5148986577987671,
+      "eval_runtime": 2.3459,
+      "eval_samples_per_second": 85.255,
+      "eval_steps_per_second": 85.255,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7233991333654309,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00019322744847062291,
+      "loss": 0.639,
+      "step": 3005
+    },
+    {
+      "epoch": 0.724602792489167,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00019318856337815425,
+      "loss": 0.6763,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019314961993702456,
+      "loss": 0.6059,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7270101107366393,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019311061818218432,
+      "loss": 0.5965,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7282137698603756,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001930715581486362,
+      "loss": 0.5941,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7294174289841117,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019303243987143538,
+      "loss": 0.6248,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7306210881078479,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019299326338568915,
+      "loss": 0.5996,
+      "step": 3035
+    },
+    {
+      "epoch": 0.731824747231584,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019295402872655717,
+      "loss": 0.5812,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7330284063553202,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001929147359292511,
+      "loss": 0.5873,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7342320654790563,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019287538502903506,
+      "loss": 0.6372,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7354357246027925,
+      "grad_norm": 1.4609375,
+      "learning_rate": 0.00019283597606122504,
+      "loss": 0.579,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7366393837265286,
+      "grad_norm": 1.375,
+      "learning_rate": 0.00019279650906118938,
+      "loss": 0.6259,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7378430428502648,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001927569840643483,
+      "loss": 0.6138,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7390467019740009,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019271740110617425,
+      "loss": 0.6437,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7402503610977371,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00019267776022219155,
+      "loss": 0.6368,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7414540202214733,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00019263806144797656,
+      "loss": 0.6152,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7426576793452094,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019259830481915767,
+      "loss": 0.5404,
+      "step": 3085
+    },
+    {
+      "epoch": 0.7438613384689456,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019255849037141506,
+      "loss": 0.6364,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7450649975926817,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019251861814048092,
+      "loss": 0.5972,
+      "step": 3095
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00019247868816213917,
+      "loss": 0.6147,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7474723158401541,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019243870047222568,
+      "loss": 0.6057,
+      "step": 3105
+    },
+    {
+      "epoch": 0.7486759749638903,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019239865510662808,
+      "loss": 0.5704,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7498796340876264,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00019235855210128573,
+      "loss": 0.6119,
+      "step": 3115
+    },
+    {
+      "epoch": 0.7510832932113626,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019231839149218977,
+      "loss": 0.5928,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7522869523350987,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00019227817331538292,
+      "loss": 0.6013,
+      "step": 3125
+    },
+    {
+      "epoch": 0.7534906114588349,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001922378976069597,
+      "loss": 0.6183,
+      "step": 3130
+    },
+    {
+      "epoch": 0.754694270582571,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019219756440306624,
+      "loss": 0.5788,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7558979297063072,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019215717373990022,
+      "loss": 0.6394,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7571015888300433,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00019211672565371088,
+      "loss": 0.5905,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7583052479537795,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019207622018079906,
+      "loss": 0.615,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7595089070775156,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019203565735751704,
+      "loss": 0.5802,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7607125662012518,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019199503722026862,
+      "loss": 0.6187,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7619162253249879,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019195435980550902,
+      "loss": 0.6073,
+      "step": 3165
+    },
+    {
+      "epoch": 0.7631198844487241,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019191362514974487,
+      "loss": 0.5869,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7643235435724602,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019187283328953413,
+      "loss": 0.6094,
+      "step": 3175
+    },
+    {
+      "epoch": 0.7655272026961965,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019183198426148614,
+      "loss": 0.5951,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7667308618199326,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019179107810226155,
+      "loss": 0.6064,
+      "step": 3185
+    },
+    {
+      "epoch": 0.7679345209436688,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019175011484857222,
+      "loss": 0.6124,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7691381800674049,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00019170909453718142,
+      "loss": 0.6183,
+      "step": 3195
+    },
+    {
+      "epoch": 0.7703418391911411,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019166801720490336,
+      "loss": 0.6389,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7715454983148772,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019162688288860367,
+      "loss": 0.5935,
+      "step": 3205
+    },
+    {
+      "epoch": 0.7727491574386134,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00019158569162519904,
+      "loss": 0.6266,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7739528165623495,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019154444345165716,
+      "loss": 0.6139,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7751564756860857,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019150313840499695,
+      "loss": 0.5838,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7763601348098219,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001914617765222883,
+      "loss": 0.6252,
+      "step": 3225
+    },
+    {
+      "epoch": 0.777563793933558,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0001914203578406521,
+      "loss": 0.6568,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7787674530572942,
+      "grad_norm": 1.3125,
+      "learning_rate": 0.00019137888239726024,
+      "loss": 0.6226,
+      "step": 3235
+    },
+    {
+      "epoch": 0.7799711121810303,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019133735022933558,
+      "loss": 0.5844,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7811747713047665,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001912957613741518,
+      "loss": 0.5578,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7823784304285026,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019125411586903355,
+      "loss": 0.6047,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7835820895522388,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00019121241375135624,
+      "loss": 0.5762,
+      "step": 3255
+    },
+    {
+      "epoch": 0.784785748675975,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019117065505854614,
+      "loss": 0.6227,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7859894077997112,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00019112883982808036,
+      "loss": 0.5868,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7871930669234473,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00019108696809748651,
+      "loss": 0.5681,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7883967260471835,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001910450399043432,
+      "loss": 0.6221,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7896003851709196,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019100305528627952,
+      "loss": 0.5701,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7908040442946558,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001909610142809753,
+      "loss": 0.6213,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7920077034183919,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001909189169261609,
+      "loss": 0.5885,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7932113625421281,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001908767632596173,
+      "loss": 0.6039,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7944150216658642,
+      "grad_norm": 1.25,
+      "learning_rate": 0.00019083455331917604,
+      "loss": 0.6027,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7956186807896004,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00019079228714271907,
+      "loss": 0.601,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7968223399133365,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00019074996476817893,
+      "loss": 0.5832,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7980259990370727,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001907075862335385,
+      "loss": 0.604,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7992296581608088,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019066515157683113,
+      "loss": 0.6115,
+      "step": 3320
+    },
+    {
+      "epoch": 0.800433317284545,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019062266083614044,
+      "loss": 0.6115,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8016369764082811,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019058011404960053,
+      "loss": 0.5968,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8028406355320173,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001905375112553957,
+      "loss": 0.5939,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8040442946557534,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00019049485249176046,
+      "loss": 0.6154,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8052479537794897,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00019045213779697972,
+      "loss": 0.6037,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00019040936720938844,
+      "loss": 0.5619,
+      "step": 3350
+    },
+    {
+      "epoch": 0.807655272026962,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00019036654076737183,
+      "loss": 0.5984,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8088589311506981,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019032365850936512,
+      "loss": 0.5342,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8100625902744343,
+      "grad_norm": 1.34375,
+      "learning_rate": 0.0001902807204738538,
+      "loss": 0.5779,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8112662493981705,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019023772669937326,
+      "loss": 0.594,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8124699085219066,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00019019467722450895,
+      "loss": 0.599,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8136735676456428,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.0001901515720878964,
+      "loss": 0.6092,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8148772267693789,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.000190108411328221,
+      "loss": 0.5535,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8160808858931151,
+      "grad_norm": 1.390625,
+      "learning_rate": 0.00019006519498421812,
+      "loss": 0.6414,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8172845450168512,
+      "grad_norm": 1.25,
+      "learning_rate": 0.0001900219230946729,
+      "loss": 0.599,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8184882041405874,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018997859569842046,
+      "loss": 0.5924,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8196918632643235,
+      "grad_norm": 1.421875,
+      "learning_rate": 0.00018993521283434573,
+      "loss": 0.6377,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8208955223880597,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00018989177454138337,
+      "loss": 0.5973,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8220991815117958,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00018984828085851776,
+      "loss": 0.603,
+      "step": 3415
+    },
+    {
+      "epoch": 0.823302840635532,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00018980473182478303,
+      "loss": 0.5744,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8245064997592682,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.000189761127479263,
+      "loss": 0.6272,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8257101588830044,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00018971746786109107,
+      "loss": 0.6359,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8269138180067405,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018967375300945038,
+      "loss": 0.5578,
+      "step": 3435
+    },
+    {
+      "epoch": 0.8281174771304767,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001896299829635735,
+      "loss": 0.563,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8293211362542128,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001895861577627425,
+      "loss": 0.5758,
+      "step": 3445
+    },
+    {
+      "epoch": 0.830524795377949,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018954227744628918,
+      "loss": 0.5843,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8317284545016851,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018949834205359454,
+      "loss": 0.6058,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8329321136254213,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001894543516240892,
+      "loss": 0.6144,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8341357727491574,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018941030619725306,
+      "loss": 0.5773,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8353394318728936,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018936620581261543,
+      "loss": 0.5878,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8365430909966297,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018932205050975495,
+      "loss": 0.5736,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8377467501203659,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001892778403282995,
+      "loss": 0.5935,
+      "step": 3480
+    },
+    {
+      "epoch": 0.838950409244102,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001892335753079262,
+      "loss": 0.6066,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8401540683678382,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018918925548836154,
+      "loss": 0.5443,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8413577274915743,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.000189144880909381,
+      "loss": 0.6198,
+      "step": 3495
+    },
+    {
+      "epoch": 0.8425613866153105,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001891004516108093,
+      "loss": 0.6282,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8425613866153105,
+      "eval_loss": 0.4995540380477905,
+      "eval_runtime": 2.3523,
+      "eval_samples_per_second": 85.024,
+      "eval_steps_per_second": 85.024,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8437650457390466,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018905596763252024,
+      "loss": 0.6045,
+      "step": 3505
+    },
+    {
+      "epoch": 0.8449687048627829,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001890114290144367,
+      "loss": 0.5966,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8461723639865191,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001889668357965306,
+      "loss": 0.5985,
+      "step": 3515
+    },
+    {
+      "epoch": 0.8473760231102552,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001889221880188229,
+      "loss": 0.5791,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8485796822339914,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00018887748572138347,
+      "loss": 0.606,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8497833413577275,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00018883272894433108,
+      "loss": 0.5975,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8509870004814637,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.0001887879177278335,
+      "loss": 0.6532,
+      "step": 3535
+    },
+    {
+      "epoch": 0.8521906596051998,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00018874305211210728,
+      "loss": 0.6029,
+      "step": 3540
+    },
+    {
+      "epoch": 0.853394318728936,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001886981321374178,
+      "loss": 0.593,
+      "step": 3545
+    },
+    {
+      "epoch": 0.8545979778526721,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018865315784407925,
+      "loss": 0.5687,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8558016369764083,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018860812927245458,
+      "loss": 0.6154,
+      "step": 3555
+    },
+    {
+      "epoch": 0.8570052961001444,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018856304646295533,
+      "loss": 0.5943,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8582089552238806,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001885179094560419,
+      "loss": 0.6202,
+      "step": 3565
+    },
+    {
+      "epoch": 0.8594126143476167,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0001884727182922233,
+      "loss": 0.5927,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8606162734713529,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018842747301205695,
+      "loss": 0.6032,
+      "step": 3575
+    },
+    {
+      "epoch": 0.861819932595089,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00018838217365614907,
+      "loss": 0.5836,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8630235917188253,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00018833682026515427,
+      "loss": 0.5684,
+      "step": 3585
+    },
+    {
+      "epoch": 0.8642272508425614,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00018829141287977574,
+      "loss": 0.5857,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8654309099662976,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001882459515407651,
+      "loss": 0.6066,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8666345690900337,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.0001882004362889223,
+      "loss": 0.5835,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8678382282137699,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018815486716509586,
+      "loss": 0.5878,
+      "step": 3605
+    },
+    {
+      "epoch": 0.869041887337506,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00018810924421018246,
+      "loss": 0.573,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8702455464612422,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018806356746512725,
+      "loss": 0.6186,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8714492055849783,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018801783697092354,
+      "loss": 0.5795,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8726528647087145,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00018797205276861292,
+      "loss": 0.602,
+      "step": 3625
+    },
+    {
+      "epoch": 0.8738565238324506,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00018792621489928518,
+      "loss": 0.6181,
+      "step": 3630
+    },
+    {
+      "epoch": 0.8750601829561868,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.0001878803234040783,
+      "loss": 0.5741,
+      "step": 3635
+    },
+    {
+      "epoch": 0.8762638420799229,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00018783437832417833,
+      "loss": 0.6213,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8774675012036591,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001877883797008195,
+      "loss": 0.5992,
+      "step": 3645
+    },
+    {
+      "epoch": 0.8786711603273952,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00018774232757528398,
+      "loss": 0.5908,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8798748194511314,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00018769622198890203,
+      "loss": 0.5768,
+      "step": 3655
+    },
+    {
+      "epoch": 0.8810784785748677,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001876500629830519,
+      "loss": 0.5549,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8822821376986038,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00018760385059915968,
+      "loss": 0.5473,
+      "step": 3665
+    },
+    {
+      "epoch": 0.88348579682234,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018755758487869952,
+      "loss": 0.6139,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8846894559460761,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.0001875112658631933,
+      "loss": 0.5976,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8858931150698123,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00018746489359421084,
+      "loss": 0.6029,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001874184681133697,
+      "loss": 0.5581,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8883004333172846,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018737198946233512,
+      "loss": 0.5848,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8895040924410207,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00018732545768282026,
+      "loss": 0.5881,
+      "step": 3695
+    },
+    {
+      "epoch": 0.8907077515647569,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018727887281658572,
+      "loss": 0.608,
+      "step": 3700
+    },
+    {
+      "epoch": 0.891911410688493,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00018723223490543992,
+      "loss": 0.5593,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8931150698122292,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001871855439912388,
+      "loss": 0.5866,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8943187289359653,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018713880011588592,
+      "loss": 0.5867,
+      "step": 3715
+    },
+    {
+      "epoch": 0.8955223880597015,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00018709200332133234,
+      "loss": 0.5776,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8967260471834376,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001870451536495766,
+      "loss": 0.5885,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8979297063071738,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00018699825114266473,
+      "loss": 0.5994,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8991333654309099,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00018695129584269017,
+      "loss": 0.5474,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9003370245546461,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00018690428779179373,
+      "loss": 0.5751,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9015406836783822,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018685722703216354,
+      "loss": 0.5515,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9027443428021185,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018681011360603507,
+      "loss": 0.5795,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9039480019258546,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00018676294755569107,
+      "loss": 0.593,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9051516610495908,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00018671572892346144,
+      "loss": 0.5936,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9063553201733269,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001866684577517234,
+      "loss": 0.6217,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9075589792970631,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00018662113408290118,
+      "loss": 0.5803,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9087626384207992,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018657375795946623,
+      "loss": 0.5893,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9099662975445354,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.000186526329423937,
+      "loss": 0.5926,
+      "step": 3780
+    },
+    {
+      "epoch": 0.9111699566682715,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001864788485188791,
+      "loss": 0.5846,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9123736157920077,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018643131528690493,
+      "loss": 0.5814,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9135772749157438,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00018638372977067407,
+      "loss": 0.5633,
+      "step": 3795
+    },
+    {
+      "epoch": 0.91478093403948,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018633609201289286,
+      "loss": 0.6015,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9159845931632162,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001862884020563147,
+      "loss": 0.587,
+      "step": 3805
+    },
+    {
+      "epoch": 0.9171882522869523,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00018624065994373964,
+      "loss": 0.614,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9183919114106885,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00018619286571801465,
+      "loss": 0.5697,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9195955705344246,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00018614501942203348,
+      "loss": 0.6092,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9207992296581609,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018609712109873654,
+      "loss": 0.5751,
+      "step": 3825
+    },
+    {
+      "epoch": 0.922002888781897,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.000186049170791111,
+      "loss": 0.593,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9232065479056332,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018600116854219066,
+      "loss": 0.6005,
+      "step": 3835
+    },
+    {
+      "epoch": 0.9244102070293693,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018595311439505594,
+      "loss": 0.5475,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9256138661531055,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00018590500839283378,
+      "loss": 0.5714,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9268175252768416,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00018585685057869776,
+      "loss": 0.5859,
+      "step": 3850
+    },
+    {
+      "epoch": 0.9280211844005778,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00018580864099586794,
+      "loss": 0.6297,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9292248435243139,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001857603796876107,
+      "loss": 0.5792,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9304285026480501,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018571206669723903,
+      "loss": 0.5965,
+      "step": 3865
+    },
+    {
+      "epoch": 0.9316321617717862,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001856637020681122,
+      "loss": 0.6071,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9328358208955224,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001856152858436359,
+      "loss": 0.5739,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9340394800192585,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.000185566818067262,
+      "loss": 0.5897,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9352431391429947,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00018551829878248876,
+      "loss": 0.5476,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9364467982667308,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001854697280328606,
+      "loss": 0.5725,
+      "step": 3890
+    },
+    {
+      "epoch": 0.937650457390467,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00018542110586196816,
+      "loss": 0.5668,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9388541165142031,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018537243231344814,
+      "loss": 0.5941,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9400577756379394,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001853237074309835,
+      "loss": 0.5715,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9412614347616755,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00018527493125830317,
+      "loss": 0.6027,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9424650938854117,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00018522610383918213,
+      "loss": 0.5993,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9436687530091478,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00018517722521744134,
+      "loss": 0.6044,
+      "step": 3920
+    },
+    {
+      "epoch": 0.944872412132884,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001851282954369478,
+      "loss": 0.5557,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9460760712566201,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00018507931454161427,
+      "loss": 0.5728,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9472797303803563,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001850302825753995,
+      "loss": 0.5711,
+      "step": 3935
+    },
+    {
+      "epoch": 0.9484833895040924,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00018498119958230806,
+      "loss": 0.5567,
+      "step": 3940
+    },
+    {
+      "epoch": 0.9496870486278286,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00018493206560639026,
+      "loss": 0.5543,
+      "step": 3945
+    },
+    {
+      "epoch": 0.9508907077515648,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018488288069174223,
+      "loss": 0.5571,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9520943668753009,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00018483364488250576,
+      "loss": 0.5442,
+      "step": 3955
+    },
+    {
+      "epoch": 0.9532980259990371,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018478435822286837,
+      "loss": 0.5445,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9545016851227732,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001847350207570632,
+      "loss": 0.5922,
+      "step": 3965
+    },
+    {
+      "epoch": 0.9557053442465094,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018468563252936891,
+      "loss": 0.5655,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9569090033702455,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018463619358410985,
+      "loss": 0.5717,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9581126624939817,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00018458670396565574,
+      "loss": 0.587,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9593163216177178,
+      "grad_norm": 1.3203125,
+      "learning_rate": 0.00018453716371842204,
+      "loss": 0.6037,
+      "step": 3985
+    },
+    {
+      "epoch": 0.9605199807414541,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00018448757288686923,
+      "loss": 0.5477,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9617236398651902,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001844379315155036,
+      "loss": 0.5953,
+      "step": 3995
+    },
+    {
+      "epoch": 0.9629272989889264,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0001843882396488765,
+      "loss": 0.5538,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9629272989889264,
+      "eval_loss": 0.4796777665615082,
+      "eval_runtime": 2.3459,
+      "eval_samples_per_second": 85.254,
+      "eval_steps_per_second": 85.254,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9641309581126625,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001843384973315848,
+      "loss": 0.6233,
+      "step": 4005
+    },
+    {
+      "epoch": 0.9653346172363987,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018428870460827058,
+      "loss": 0.5766,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9665382763601348,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001842388615236211,
+      "loss": 0.576,
+      "step": 4015
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00018418896812236884,
+      "loss": 0.5802,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9689455946076071,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00018413902444929152,
+      "loss": 0.559,
+      "step": 4025
+    },
+    {
+      "epoch": 0.9701492537313433,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00018408903054921186,
+      "loss": 0.5764,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9713529128550794,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00018403898646699777,
+      "loss": 0.5801,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9725565719788156,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001839888922475621,
+      "loss": 0.5761,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9737602311025517,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00018393874793586277,
+      "loss": 0.5683,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9749638902262879,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00018388855357690258,
+      "loss": 0.5907,
+      "step": 4050
+    },
+    {
+      "epoch": 0.976167549350024,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00018383830921572933,
+      "loss": 0.5633,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9773712084737602,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001837880148974356,
+      "loss": 0.5708,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9785748675974963,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018373767066715897,
+      "loss": 0.5555,
+      "step": 4065
+    },
+    {
+      "epoch": 0.9797785267212326,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001836872765700816,
+      "loss": 0.5653,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9809821858449687,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001836368326514305,
+      "loss": 0.6252,
+      "step": 4075
+    },
+    {
+      "epoch": 0.9821858449687049,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001835863389564775,
+      "loss": 0.5915,
+      "step": 4080
+    },
+    {
+      "epoch": 0.983389504092441,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00018353579553053893,
+      "loss": 0.5691,
+      "step": 4085
+    },
+    {
+      "epoch": 0.9845931632161772,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00018348520241897584,
+      "loss": 0.5827,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9857968223399133,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00018343455966719383,
+      "loss": 0.571,
+      "step": 4095
+    },
+    {
+      "epoch": 0.9870004814636495,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00018338386732064316,
+      "loss": 0.5897,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9882041405873857,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018333312542481841,
+      "loss": 0.5796,
+      "step": 4105
+    },
+    {
+      "epoch": 0.9894077997111218,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001832823340252588,
+      "loss": 0.572,
+      "step": 4110
+    },
+    {
+      "epoch": 0.990611458834858,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001832314931675479,
+      "loss": 0.5665,
+      "step": 4115
+    },
+    {
+      "epoch": 0.9918151179585941,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00018318060289731372,
+      "loss": 0.5833,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9930187770823303,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001831296632602285,
+      "loss": 0.5675,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9942224362060664,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00018307867430200888,
+      "loss": 0.5603,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9954260953298026,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00018302763606841576,
+      "loss": 0.57,
+      "step": 4135
+    },
+    {
+      "epoch": 0.9966297544535387,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00018297654860525433,
+      "loss": 0.5507,
+      "step": 4140
+    },
+    {
+      "epoch": 0.997833413577275,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00018292541195837373,
+      "loss": 0.5902,
+      "step": 4145
+    },
+    {
+      "epoch": 0.999037072701011,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.0001828742261736675,
+      "loss": 0.5725,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9997592681752527,
+      "eval_loss": 0.4810214340686798,
+      "eval_runtime": 2.3633,
+      "eval_samples_per_second": 84.626,
+      "eval_steps_per_second": 84.626,
+      "step": 4153
+    },
+    {
+      "epoch": 1.0002407318247473,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001828229912970731,
+      "loss": 0.5843,
+      "step": 4155
+    },
+    {
+      "epoch": 1.0014443909484834,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00018277170737457217,
+      "loss": 0.5257,
+      "step": 4160
+    },
+    {
+      "epoch": 1.0026480500722195,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00018272037445219032,
+      "loss": 0.5127,
+      "step": 4165
+    },
+    {
+      "epoch": 1.0038517091959558,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001826689925759971,
+      "loss": 0.5202,
+      "step": 4170
+    },
+    {
+      "epoch": 1.0050553683196919,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00018261756179210604,
+      "loss": 0.5105,
+      "step": 4175
+    },
+    {
+      "epoch": 1.006259027443428,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00018256608214667455,
+      "loss": 0.5246,
+      "step": 4180
+    },
+    {
+      "epoch": 1.007462686567164,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018251455368590387,
+      "loss": 0.5564,
+      "step": 4185
+    },
+    {
+      "epoch": 1.0086663456909004,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018246297645603912,
+      "loss": 0.515,
+      "step": 4190
+    },
+    {
+      "epoch": 1.0098700048146365,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001824113505033691,
+      "loss": 0.5367,
+      "step": 4195
+    },
+    {
+      "epoch": 1.0110736639383726,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001823596758742264,
+      "loss": 0.5418,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0122773230621087,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00018230795261498723,
+      "loss": 0.5572,
+      "step": 4205
+    },
+    {
+      "epoch": 1.013480982185845,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00018225618077207157,
+      "loss": 0.5422,
+      "step": 4210
+    },
+    {
+      "epoch": 1.0146846413095811,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00018220436039194284,
+      "loss": 0.501,
+      "step": 4215
+    },
+    {
+      "epoch": 1.0158883004333172,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00018215249152110812,
+      "loss": 0.53,
+      "step": 4220
+    },
+    {
+      "epoch": 1.0170919595570533,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018210057420611796,
+      "loss": 0.5213,
+      "step": 4225
+    },
+    {
+      "epoch": 1.0182956186807897,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001820486084935665,
+      "loss": 0.5377,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0194992778045258,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018199659443009117,
+      "loss": 0.5225,
+      "step": 4235
+    },
+    {
+      "epoch": 1.0207029369282619,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018194453206237283,
+      "loss": 0.5309,
+      "step": 4240
+    },
+    {
+      "epoch": 1.021906596051998,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00018189242143713577,
+      "loss": 0.5499,
+      "step": 4245
+    },
+    {
+      "epoch": 1.0231102551757343,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001818402626011475,
+      "loss": 0.5268,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0243139142994704,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00018178805560121883,
+      "loss": 0.5468,
+      "step": 4255
+    },
+    {
+      "epoch": 1.0255175734232065,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00018173580048420386,
+      "loss": 0.5463,
+      "step": 4260
+    },
+    {
+      "epoch": 1.0267212325469428,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018168349729699972,
+      "loss": 0.5491,
+      "step": 4265
+    },
+    {
+      "epoch": 1.027924891670679,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00018163114608654683,
+      "loss": 0.5286,
+      "step": 4270
+    },
+    {
+      "epoch": 1.029128550794415,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018157874689982864,
+      "loss": 0.5454,
+      "step": 4275
+    },
+    {
+      "epoch": 1.030332209918151,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001815262997838716,
+      "loss": 0.5213,
+      "step": 4280
+    },
+    {
+      "epoch": 1.0315358690418874,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00018147380478574534,
+      "loss": 0.5287,
+      "step": 4285
+    },
+    {
+      "epoch": 1.0327395281656235,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001814212619525624,
+      "loss": 0.5038,
+      "step": 4290
+    },
+    {
+      "epoch": 1.0339431872893596,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00018136867133147807,
+      "loss": 0.4808,
+      "step": 4295
+    },
+    {
+      "epoch": 1.0351468464130957,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001813160329696908,
+      "loss": 0.4892,
+      "step": 4300
+    },
+    {
+      "epoch": 1.036350505536832,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00018126334691444167,
+      "loss": 0.5366,
+      "step": 4305
+    },
+    {
+      "epoch": 1.0375541646605682,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018121061321301472,
+      "loss": 0.5313,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0387578237843043,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00018115783191273666,
+      "loss": 0.5291,
+      "step": 4315
+    },
+    {
+      "epoch": 1.0399614829080404,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.000181105003060977,
+      "loss": 0.5206,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0411651420317767,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00018105212670514772,
+      "loss": 0.551,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0423688011555128,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00018099920289270374,
+      "loss": 0.5165,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0435724602792489,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001809462316711423,
+      "loss": 0.5068,
+      "step": 4335
+    },
+    {
+      "epoch": 1.044776119402985,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00018089321308800338,
+      "loss": 0.5138,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0459797785267213,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00018084014719086934,
+      "loss": 0.5104,
+      "step": 4345
+    },
+    {
+      "epoch": 1.0471834376504574,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.0001807870340273651,
+      "loss": 0.5383,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0483870967741935,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00018073387364515788,
+      "loss": 0.5017,
+      "step": 4355
+    },
+    {
+      "epoch": 1.0495907558979296,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001806806660919574,
+      "loss": 0.5055,
+      "step": 4360
+    },
+    {
+      "epoch": 1.050794415021666,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018062741141551563,
+      "loss": 0.5104,
+      "step": 4365
+    },
+    {
+      "epoch": 1.051998074145402,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00018057410966362684,
+      "loss": 0.5105,
+      "step": 4370
+    },
+    {
+      "epoch": 1.0532017332691381,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00018052076088412766,
+      "loss": 0.5216,
+      "step": 4375
+    },
+    {
+      "epoch": 1.0544053923928742,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00018046736512489675,
+      "loss": 0.4967,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0556090515166106,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00018041392243385506,
+      "loss": 0.5398,
+      "step": 4385
+    },
+    {
+      "epoch": 1.0568127106403467,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00018036043285896556,
+      "loss": 0.5494,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0580163697640828,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00018030689644823347,
+      "loss": 0.5387,
+      "step": 4395
+    },
+    {
+      "epoch": 1.059220028887819,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00018025331324970586,
+      "loss": 0.5224,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0604236880115552,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00018019968331147183,
+      "loss": 0.5334,
+      "step": 4405
+    },
+    {
+      "epoch": 1.0616273471352913,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00018014600668166247,
+      "loss": 0.5184,
+      "step": 4410
+    },
+    {
+      "epoch": 1.0628310062590274,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001800922834084508,
+      "loss": 0.5243,
+      "step": 4415
+    },
+    {
+      "epoch": 1.0640346653827637,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00018003851354005164,
+      "loss": 0.5315,
+      "step": 4420
+    },
+    {
+      "epoch": 1.0652383245064998,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001799846971247217,
+      "loss": 0.5417,
+      "step": 4425
+    },
+    {
+      "epoch": 1.066441983630236,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001799308342107593,
+      "loss": 0.5308,
+      "step": 4430
+    },
+    {
+      "epoch": 1.067645642753972,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001798769248465047,
+      "loss": 0.5634,
+      "step": 4435
+    },
+    {
+      "epoch": 1.0688493018777083,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001798229690803397,
+      "loss": 0.5227,
+      "step": 4440
+    },
+    {
+      "epoch": 1.0700529610014444,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00017976896696068783,
+      "loss": 0.525,
+      "step": 4445
+    },
+    {
+      "epoch": 1.0712566201251805,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001797149185360142,
+      "loss": 0.5198,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0724602792489166,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00017966082385482545,
+      "loss": 0.5306,
+      "step": 4455
+    },
+    {
+      "epoch": 1.073663938372653,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001796066829656697,
+      "loss": 0.5029,
+      "step": 4460
+    },
+    {
+      "epoch": 1.074867597496389,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017955249591713668,
+      "loss": 0.5171,
+      "step": 4465
+    },
+    {
+      "epoch": 1.0760712566201251,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017949826275785735,
+      "loss": 0.5279,
+      "step": 4470
+    },
+    {
+      "epoch": 1.0772749157438612,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00017944398353650422,
+      "loss": 0.5141,
+      "step": 4475
+    },
+    {
+      "epoch": 1.0784785748675976,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017938965830179108,
+      "loss": 0.4901,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0796822339913337,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00017933528710247298,
+      "loss": 0.5182,
+      "step": 4485
+    },
+    {
+      "epoch": 1.0808858931150698,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017928086998734633,
+      "loss": 0.5293,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0820895522388059,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017922640700524856,
+      "loss": 0.5104,
+      "step": 4495
+    },
+    {
+      "epoch": 1.0832932113625422,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00017917189820505848,
+      "loss": 0.5316,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0832932113625422,
+      "eval_loss": 0.4706423878669739,
+      "eval_runtime": 2.3716,
+      "eval_samples_per_second": 84.331,
+      "eval_steps_per_second": 84.331,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0844968704862783,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001791173436356958,
+      "loss": 0.515,
+      "step": 4505
+    },
+    {
+      "epoch": 1.0857005296100144,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00017906274334612157,
+      "loss": 0.5566,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0869041887337505,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00017900809738533756,
+      "loss": 0.5197,
+      "step": 4515
+    },
+    {
+      "epoch": 1.0881078478574868,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017895340580238684,
+      "loss": 0.5884,
+      "step": 4520
+    },
+    {
+      "epoch": 1.089311506981223,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001788986686463532,
+      "loss": 0.5118,
+      "step": 4525
+    },
+    {
+      "epoch": 1.090515166104959,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00017884388596636145,
+      "loss": 0.5278,
+      "step": 4530
+    },
+    {
+      "epoch": 1.0917188252286953,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00017878905781157717,
+      "loss": 0.5652,
+      "step": 4535
+    },
+    {
+      "epoch": 1.0929224843524314,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001787341842312068,
+      "loss": 0.5057,
+      "step": 4540
+    },
+    {
+      "epoch": 1.0941261434761675,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017867926527449757,
+      "loss": 0.5338,
+      "step": 4545
+    },
+    {
+      "epoch": 1.0953298025999036,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00017862430099073738,
+      "loss": 0.5067,
+      "step": 4550
+    },
+    {
+      "epoch": 1.09653346172364,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001785692914292549,
+      "loss": 0.539,
+      "step": 4555
+    },
+    {
+      "epoch": 1.097737120847376,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.0001785142366394193,
+      "loss": 0.5525,
+      "step": 4560
+    },
+    {
+      "epoch": 1.0989407799711122,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00017845913667064044,
+      "loss": 0.5261,
+      "step": 4565
+    },
+    {
+      "epoch": 1.1001444390948483,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001784039915723687,
+      "loss": 0.535,
+      "step": 4570
+    },
+    {
+      "epoch": 1.1013480982185846,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.000178348801394095,
+      "loss": 0.5062,
+      "step": 4575
+    },
+    {
+      "epoch": 1.1025517573423207,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001782935661853506,
+      "loss": 0.5015,
+      "step": 4580
+    },
+    {
+      "epoch": 1.1037554164660568,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001782382859957073,
+      "loss": 0.5428,
+      "step": 4585
+    },
+    {
+      "epoch": 1.104959075589793,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00017818296087477724,
+      "loss": 0.5294,
+      "step": 4590
+    },
+    {
+      "epoch": 1.1061627347135292,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017812759087221282,
+      "loss": 0.4981,
+      "step": 4595
+    },
+    {
+      "epoch": 1.1073663938372653,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00017807217603770686,
+      "loss": 0.5253,
+      "step": 4600
+    },
+    {
+      "epoch": 1.1085700529610014,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001780167164209922,
+      "loss": 0.4899,
+      "step": 4605
+    },
+    {
+      "epoch": 1.1097737120847375,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017796121207184205,
+      "loss": 0.5408,
+      "step": 4610
+    },
+    {
+      "epoch": 1.1109773712084738,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001779056630400697,
+      "loss": 0.5328,
+      "step": 4615
+    },
+    {
+      "epoch": 1.11218103033221,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00017785006937552858,
+      "loss": 0.5003,
+      "step": 4620
+    },
+    {
+      "epoch": 1.113384689455946,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001777944311281121,
+      "loss": 0.5436,
+      "step": 4625
+    },
+    {
+      "epoch": 1.1145883485796821,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00017773874834775379,
+      "loss": 0.5552,
+      "step": 4630
+    },
+    {
+      "epoch": 1.1157920077034185,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.000177683021084427,
+      "loss": 0.5394,
+      "step": 4635
+    },
+    {
+      "epoch": 1.1169956668271546,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00017762724938814514,
+      "loss": 0.5216,
+      "step": 4640
+    },
+    {
+      "epoch": 1.1181993259508907,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00017757143330896145,
+      "loss": 0.5347,
+      "step": 4645
+    },
+    {
+      "epoch": 1.1194029850746268,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017751557289696896,
+      "loss": 0.5043,
+      "step": 4650
+    },
+    {
+      "epoch": 1.120606644198363,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017745966820230055,
+      "loss": 0.5087,
+      "step": 4655
+    },
+    {
+      "epoch": 1.1218103033220992,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001774037192751288,
+      "loss": 0.5139,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1230139624458353,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.000177347726165666,
+      "loss": 0.5439,
+      "step": 4665
+    },
+    {
+      "epoch": 1.1242176215695716,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001772916889241641,
+      "loss": 0.4872,
+      "step": 4670
+    },
+    {
+      "epoch": 1.1254212806933077,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00017723560760091466,
+      "loss": 0.5446,
+      "step": 4675
+    },
+    {
+      "epoch": 1.1266249398170438,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001771794822462488,
+      "loss": 0.5149,
+      "step": 4680
+    },
+    {
+      "epoch": 1.12782859894078,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001771233129105371,
+      "loss": 0.5115,
+      "step": 4685
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00017706709964418975,
+      "loss": 0.5322,
+      "step": 4690
+    },
+    {
+      "epoch": 1.1302359171882523,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00017701084249765618,
+      "loss": 0.5153,
+      "step": 4695
+    },
+    {
+      "epoch": 1.1314395763119884,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00017695454152142536,
+      "loss": 0.5189,
+      "step": 4700
+    },
+    {
+      "epoch": 1.1326432354357245,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00017689819676602552,
+      "loss": 0.5008,
+      "step": 4705
+    },
+    {
+      "epoch": 1.1338468945594609,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001768418082820242,
+      "loss": 0.5202,
+      "step": 4710
+    },
+    {
+      "epoch": 1.135050553683197,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00017678537612002816,
+      "loss": 0.5312,
+      "step": 4715
+    },
+    {
+      "epoch": 1.136254212806933,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017672890033068337,
+      "loss": 0.5455,
+      "step": 4720
+    },
+    {
+      "epoch": 1.1374578719306692,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017667238096467504,
+      "loss": 0.526,
+      "step": 4725
+    },
+    {
+      "epoch": 1.1386615310544055,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00017661581807272732,
+      "loss": 0.5238,
+      "step": 4730
+    },
+    {
+      "epoch": 1.1398651901781416,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00017655921170560356,
+      "loss": 0.5419,
+      "step": 4735
+    },
+    {
+      "epoch": 1.1410688493018777,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00017650256191410605,
+      "loss": 0.504,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1422725084256138,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017644586874907613,
+      "loss": 0.5404,
+      "step": 4745
+    },
+    {
+      "epoch": 1.14347616754935,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00017638913226139397,
+      "loss": 0.5315,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1446798266730862,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00017633235250197874,
+      "loss": 0.5091,
+      "step": 4755
+    },
+    {
+      "epoch": 1.1458834857968223,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00017627552952178833,
+      "loss": 0.5325,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1470871449205584,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00017621866337181946,
+      "loss": 0.5081,
+      "step": 4765
+    },
+    {
+      "epoch": 1.1482908040442947,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001761617541031076,
+      "loss": 0.521,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1494944631680308,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00017610480176672694,
+      "loss": 0.499,
+      "step": 4775
+    },
+    {
+      "epoch": 1.150698122291767,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001760478064137903,
+      "loss": 0.5359,
+      "step": 4780
+    },
+    {
+      "epoch": 1.151901781415503,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00017599076809544906,
+      "loss": 0.5401,
+      "step": 4785
+    },
+    {
+      "epoch": 1.1531054405392394,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00017593368686289327,
+      "loss": 0.5244,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1543090996629755,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017587656276735138,
+      "loss": 0.5201,
+      "step": 4795
+    },
+    {
+      "epoch": 1.1555127587867116,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017581939586009033,
+      "loss": 0.5031,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1567164179104479,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00017576218619241559,
+      "loss": 0.5197,
+      "step": 4805
+    },
+    {
+      "epoch": 1.157920077034184,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017570493381567086,
+      "loss": 0.5258,
+      "step": 4810
+    },
+    {
+      "epoch": 1.15912373615792,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017564763878123827,
+      "loss": 0.5382,
+      "step": 4815
+    },
+    {
+      "epoch": 1.1603273952816562,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017559030114053817,
+      "loss": 0.5308,
+      "step": 4820
+    },
+    {
+      "epoch": 1.1615310544053923,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001755329209450292,
+      "loss": 0.5421,
+      "step": 4825
+    },
+    {
+      "epoch": 1.1627347135291286,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00017547549824620815,
+      "loss": 0.4822,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1639383726528647,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00017541803309561,
+      "loss": 0.5155,
+      "step": 4835
+    },
+    {
+      "epoch": 1.1651420317766008,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001753605255448078,
+      "loss": 0.5297,
+      "step": 4840
+    },
+    {
+      "epoch": 1.1663456909003371,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017530297564541256,
+      "loss": 0.5199,
+      "step": 4845
+    },
+    {
+      "epoch": 1.1675493500240732,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00017524538344907353,
+      "loss": 0.5427,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1687530091478093,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017518774900747772,
+      "loss": 0.4721,
+      "step": 4855
+    },
+    {
+      "epoch": 1.1699566682715454,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017513007237235007,
+      "loss": 0.5255,
+      "step": 4860
+    },
+    {
+      "epoch": 1.1711603273952818,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017507235359545353,
+      "loss": 0.5042,
+      "step": 4865
+    },
+    {
+      "epoch": 1.1723639865190179,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017501459272858868,
+      "loss": 0.5018,
+      "step": 4870
+    },
+    {
+      "epoch": 1.173567645642754,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00017495678982359404,
+      "loss": 0.5463,
+      "step": 4875
+    },
+    {
+      "epoch": 1.17477130476649,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001748989449323457,
+      "loss": 0.5344,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1759749638902264,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00017484105810675762,
+      "loss": 0.5347,
+      "step": 4885
+    },
+    {
+      "epoch": 1.1771786230139625,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00017478312939878124,
+      "loss": 0.5007,
+      "step": 4890
+    },
+    {
+      "epoch": 1.1783822821376986,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001747251588604056,
+      "loss": 0.5041,
+      "step": 4895
+    },
+    {
+      "epoch": 1.1795859412614347,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017466714654365738,
+      "loss": 0.49,
+      "step": 4900
+    },
+    {
+      "epoch": 1.180789600385171,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001746090925006007,
+      "loss": 0.5114,
+      "step": 4905
+    },
+    {
+      "epoch": 1.181993259508907,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00017455099678333706,
+      "loss": 0.5697,
+      "step": 4910
+    },
+    {
+      "epoch": 1.1831969186326432,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001744928594440055,
+      "loss": 0.4936,
+      "step": 4915
+    },
+    {
+      "epoch": 1.1844005777563793,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00017443468053478229,
+      "loss": 0.5241,
+      "step": 4920
+    },
+    {
+      "epoch": 1.1856042368801156,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00017437646010788106,
+      "loss": 0.51,
+      "step": 4925
+    },
+    {
+      "epoch": 1.1868078960038517,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001743181982155527,
+      "loss": 0.5403,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1880115551275878,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001742598949100853,
+      "loss": 0.5321,
+      "step": 4935
+    },
+    {
+      "epoch": 1.1892152142513241,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00017420155024380416,
+      "loss": 0.4952,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1904188733750602,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00017414316426907166,
+      "loss": 0.5318,
+      "step": 4945
+    },
+    {
+      "epoch": 1.1916225324987963,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017408473703828725,
+      "loss": 0.5119,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1928261916225325,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017402626860388743,
+      "loss": 0.489,
+      "step": 4955
+    },
+    {
+      "epoch": 1.1940298507462686,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00017396775901834562,
+      "loss": 0.5417,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1952335098700049,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00017390920833417227,
+      "loss": 0.5087,
+      "step": 4965
+    },
+    {
+      "epoch": 1.196437168993741,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00017385061660391465,
+      "loss": 0.4913,
+      "step": 4970
+    },
+    {
+      "epoch": 1.197640828117477,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00017379198388015686,
+      "loss": 0.522,
+      "step": 4975
+    },
+    {
+      "epoch": 1.1988444872412134,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00017373331021551986,
+      "loss": 0.5076,
+      "step": 4980
+    },
+    {
+      "epoch": 1.2000481463649495,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00017367459566266122,
+      "loss": 0.5261,
+      "step": 4985
+    },
+    {
+      "epoch": 1.2012518054886856,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00017361584027427536,
+      "loss": 0.5098,
+      "step": 4990
+    },
+    {
+      "epoch": 1.2024554646124217,
+      "grad_norm": 1.2734375,
+      "learning_rate": 0.00017355704410309324,
+      "loss": 0.5317,
+      "step": 4995
+    },
+    {
+      "epoch": 1.2036591237361578,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00017349820720188248,
+      "loss": 0.5442,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2036591237361578,
+      "eval_loss": 0.4596037268638611,
+      "eval_runtime": 2.3802,
+      "eval_samples_per_second": 84.026,
+      "eval_steps_per_second": 84.026,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2048627828598941,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017343932962344722,
+      "loss": 0.493,
+      "step": 5005
+    },
+    {
+      "epoch": 1.2060664419836302,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017338041142062814,
+      "loss": 0.4901,
+      "step": 5010
+    },
+    {
+      "epoch": 1.2072701011073663,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001733214526463023,
+      "loss": 0.5146,
+      "step": 5015
+    },
+    {
+      "epoch": 1.2084737602311026,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001732624533533833,
+      "loss": 0.5151,
+      "step": 5020
+    },
+    {
+      "epoch": 1.2096774193548387,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.000173203413594821,
+      "loss": 0.5402,
+      "step": 5025
+    },
+    {
+      "epoch": 1.2108810784785748,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00017314433342360158,
+      "loss": 0.5233,
+      "step": 5030
+    },
+    {
+      "epoch": 1.212084737602311,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017308521289274758,
+      "loss": 0.521,
+      "step": 5035
+    },
+    {
+      "epoch": 1.2132883967260473,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00017302605205531766,
+      "loss": 0.5392,
+      "step": 5040
+    },
+    {
+      "epoch": 1.2144920558497834,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00017296685096440673,
+      "loss": 0.5299,
+      "step": 5045
+    },
+    {
+      "epoch": 1.2156957149735195,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017290760967314577,
+      "loss": 0.5142,
+      "step": 5050
+    },
+    {
+      "epoch": 1.2168993740972556,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00017284832823470178,
+      "loss": 0.5286,
+      "step": 5055
+    },
+    {
+      "epoch": 1.218103033220992,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.000172789006702278,
+      "loss": 0.5047,
+      "step": 5060
+    },
+    {
+      "epoch": 1.219306692344728,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00017272964512911345,
+      "loss": 0.489,
+      "step": 5065
+    },
+    {
+      "epoch": 1.220510351468464,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00017267024356848318,
+      "loss": 0.5114,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2217140105922002,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00017261080207369814,
+      "loss": 0.5132,
+      "step": 5075
+    },
+    {
+      "epoch": 1.2229176697159365,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.000172551320698105,
+      "loss": 0.5474,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2241213288396726,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00017249179949508633,
+      "loss": 0.477,
+      "step": 5085
+    },
+    {
+      "epoch": 1.2253249879634087,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00017243223851806048,
+      "loss": 0.527,
+      "step": 5090
+    },
+    {
+      "epoch": 1.2265286470871448,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001723726378204814,
+      "loss": 0.5112,
+      "step": 5095
+    },
+    {
+      "epoch": 1.2277323062108811,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001723129974558387,
+      "loss": 0.532,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2289359653346172,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00017225331747765764,
+      "loss": 0.5028,
+      "step": 5105
+    },
+    {
+      "epoch": 1.2301396244583533,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.000172193597939499,
+      "loss": 0.5074,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2313432835820897,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017213383889495907,
+      "loss": 0.5334,
+      "step": 5115
+    },
+    {
+      "epoch": 1.2325469427058258,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001720740403976696,
+      "loss": 0.5109,
+      "step": 5120
+    },
+    {
+      "epoch": 1.2337506018295619,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017201420250129772,
+      "loss": 0.5004,
+      "step": 5125
+    },
+    {
+      "epoch": 1.234954260953298,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00017195432525954598,
+      "loss": 0.4972,
+      "step": 5130
+    },
+    {
+      "epoch": 1.236157920077034,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00017189440872615217,
+      "loss": 0.4611,
+      "step": 5135
+    },
+    {
+      "epoch": 1.2373615792007704,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001718344529548894,
+      "loss": 0.543,
+      "step": 5140
+    },
+    {
+      "epoch": 1.2385652383245065,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00017177445799956595,
+      "loss": 0.5259,
+      "step": 5145
+    },
+    {
+      "epoch": 1.2397688974482426,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001717144239140253,
+      "loss": 0.5092,
+      "step": 5150
+    },
+    {
+      "epoch": 1.240972556571979,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017165435075214603,
+      "loss": 0.4841,
+      "step": 5155
+    },
+    {
+      "epoch": 1.242176215695715,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017159423856784176,
+      "loss": 0.5283,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2433798748194511,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00017153408741506126,
+      "loss": 0.5033,
+      "step": 5165
+    },
+    {
+      "epoch": 1.2445835339431872,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00017147389734778812,
+      "loss": 0.5252,
+      "step": 5170
+    },
+    {
+      "epoch": 1.2457871930669235,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.0001714136684200409,
+      "loss": 0.5088,
+      "step": 5175
+    },
+    {
+      "epoch": 1.2469908521906596,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017135340068587303,
+      "loss": 0.5282,
+      "step": 5180
+    },
+    {
+      "epoch": 1.2481945113143957,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00017129309419937287,
+      "loss": 0.5047,
+      "step": 5185
+    },
+    {
+      "epoch": 1.2493981704381318,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001712327490146634,
+      "loss": 0.5001,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2506018295618682,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00017117236518590244,
+      "loss": 0.5134,
+      "step": 5195
+    },
+    {
+      "epoch": 1.2518054886856043,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00017111194276728245,
+      "loss": 0.5247,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2530091478093404,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017105148181303053,
+      "loss": 0.5102,
+      "step": 5205
+    },
+    {
+      "epoch": 1.2542128069330767,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00017099098237740838,
+      "loss": 0.5079,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2554164660568128,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001709304445147122,
+      "loss": 0.4895,
+      "step": 5215
+    },
+    {
+      "epoch": 1.2566201251805489,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001708698682792727,
+      "loss": 0.521,
+      "step": 5220
+    },
+    {
+      "epoch": 1.257823784304285,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00017080925372545504,
+      "loss": 0.5099,
+      "step": 5225
+    },
+    {
+      "epoch": 1.259027443428021,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00017074860090765878,
+      "loss": 0.5001,
+      "step": 5230
+    },
+    {
+      "epoch": 1.2602311025517574,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00017068790988031772,
+      "loss": 0.5255,
+      "step": 5235
+    },
+    {
+      "epoch": 1.2614347616754935,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00017062718069790013,
+      "loss": 0.5225,
+      "step": 5240
+    },
+    {
+      "epoch": 1.2626384207992296,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00017056641341490832,
+      "loss": 0.5056,
+      "step": 5245
+    },
+    {
+      "epoch": 1.263842079922966,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00017050560808587898,
+      "loss": 0.5309,
+      "step": 5250
+    },
+    {
+      "epoch": 1.265045739046702,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00017044476476538279,
+      "loss": 0.5052,
+      "step": 5255
+    },
+    {
+      "epoch": 1.2662493981704381,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00017038388350802463,
+      "loss": 0.519,
+      "step": 5260
+    },
+    {
+      "epoch": 1.2674530572941742,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017032296436844343,
+      "loss": 0.552,
+      "step": 5265
+    },
+    {
+      "epoch": 1.2686567164179103,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.000170262007401312,
+      "loss": 0.5129,
+      "step": 5270
+    },
+    {
+      "epoch": 1.2698603755416467,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00017020101266133724,
+      "loss": 0.4976,
+      "step": 5275
+    },
+    {
+      "epoch": 1.2710640346653828,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00017013998020325987,
+      "loss": 0.531,
+      "step": 5280
+    },
+    {
+      "epoch": 1.2722676937891189,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00017007891008185452,
+      "loss": 0.5335,
+      "step": 5285
+    },
+    {
+      "epoch": 1.2734713529128552,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00017001780235192953,
+      "loss": 0.4964,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2746750120365913,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016995665706832702,
+      "loss": 0.5139,
+      "step": 5295
+    },
+    {
+      "epoch": 1.2758786711603274,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00016989547428592294,
+      "loss": 0.5263,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2770823302840635,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016983425405962672,
+      "loss": 0.5065,
+      "step": 5305
+    },
+    {
+      "epoch": 1.2782859894077996,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001697729964443815,
+      "loss": 0.5654,
+      "step": 5310
+    },
+    {
+      "epoch": 1.279489648531536,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00016971170149516387,
+      "loss": 0.5036,
+      "step": 5315
+    },
+    {
+      "epoch": 1.280693307655272,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00016965036926698407,
+      "loss": 0.5134,
+      "step": 5320
+    },
+    {
+      "epoch": 1.281896966779008,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016958899981488575,
+      "loss": 0.5103,
+      "step": 5325
+    },
+    {
+      "epoch": 1.2831006259027444,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00016952759319394586,
+      "loss": 0.5246,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2843042850264805,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016946614945927482,
+      "loss": 0.5287,
+      "step": 5335
+    },
+    {
+      "epoch": 1.2855079441502166,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016940466866601636,
+      "loss": 0.5348,
+      "step": 5340
+    },
+    {
+      "epoch": 1.286711603273953,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00016934315086934742,
+      "loss": 0.4994,
+      "step": 5345
+    },
+    {
+      "epoch": 1.287915262397689,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016928159612447814,
+      "loss": 0.5094,
+      "step": 5350
+    },
+    {
+      "epoch": 1.2891189215214252,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001692200044866519,
+      "loss": 0.5391,
+      "step": 5355
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016915837601114503,
+      "loss": 0.5388,
+      "step": 5360
+    },
+    {
+      "epoch": 1.2915262397688974,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00016909671075326718,
+      "loss": 0.5102,
+      "step": 5365
+    },
+    {
+      "epoch": 1.2927298988926337,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00016903500876836074,
+      "loss": 0.4788,
+      "step": 5370
+    },
+    {
+      "epoch": 1.2939335580163698,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00016897327011180122,
+      "loss": 0.4988,
+      "step": 5375
+    },
+    {
+      "epoch": 1.2951372171401059,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016891149483899695,
+      "loss": 0.5243,
+      "step": 5380
+    },
+    {
+      "epoch": 1.2963408762638422,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001688496830053893,
+      "loss": 0.5362,
+      "step": 5385
+    },
+    {
+      "epoch": 1.2975445353875783,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016878783466645221,
+      "loss": 0.5257,
+      "step": 5390
+    },
+    {
+      "epoch": 1.2987481945113144,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00016872594987769253,
+      "loss": 0.519,
+      "step": 5395
+    },
+    {
+      "epoch": 1.2999518536350505,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00016866402869464978,
+      "loss": 0.4683,
+      "step": 5400
+    },
+    {
+      "epoch": 1.3011555127587866,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016860207117289617,
+      "loss": 0.5215,
+      "step": 5405
+    },
+    {
+      "epoch": 1.302359171882523,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001685400773680365,
+      "loss": 0.5371,
+      "step": 5410
+    },
+    {
+      "epoch": 1.303562831006259,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016847804733570812,
+      "loss": 0.4945,
+      "step": 5415
+    },
+    {
+      "epoch": 1.3047664901299951,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016841598113158095,
+      "loss": 0.5054,
+      "step": 5420
+    },
+    {
+      "epoch": 1.3059701492537314,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001683538788113573,
+      "loss": 0.4893,
+      "step": 5425
+    },
+    {
+      "epoch": 1.3071738083774675,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00016829174043077192,
+      "loss": 0.5048,
+      "step": 5430
+    },
+    {
+      "epoch": 1.3083774675012037,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00016822956604559195,
+      "loss": 0.5116,
+      "step": 5435
+    },
+    {
+      "epoch": 1.3095811266249398,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00016816735571161684,
+      "loss": 0.5216,
+      "step": 5440
+    },
+    {
+      "epoch": 1.3107847857486759,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016810510948467826,
+      "loss": 0.5351,
+      "step": 5445
+    },
+    {
+      "epoch": 1.3119884448724122,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001680428274206401,
+      "loss": 0.4961,
+      "step": 5450
+    },
+    {
+      "epoch": 1.3131921039961483,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00016798050957539848,
+      "loss": 0.5049,
+      "step": 5455
+    },
+    {
+      "epoch": 1.3143957631198844,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00016791815600488154,
+      "loss": 0.5118,
+      "step": 5460
+    },
+    {
+      "epoch": 1.3155994222436207,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016785576676504955,
+      "loss": 0.5132,
+      "step": 5465
+    },
+    {
+      "epoch": 1.3168030813673568,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00016779334191189472,
+      "loss": 0.4766,
+      "step": 5470
+    },
+    {
+      "epoch": 1.318006740491093,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00016773088150144133,
+      "loss": 0.5046,
+      "step": 5475
+    },
+    {
+      "epoch": 1.319210399614829,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016766838558974554,
+      "loss": 0.5015,
+      "step": 5480
+    },
+    {
+      "epoch": 1.320414058738565,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001676058542328952,
+      "loss": 0.4881,
+      "step": 5485
+    },
+    {
+      "epoch": 1.3216177178623014,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016754328748701028,
+      "loss": 0.5218,
+      "step": 5490
+    },
+    {
+      "epoch": 1.3228213769860375,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016748068540824222,
+      "loss": 0.4982,
+      "step": 5495
+    },
+    {
+      "epoch": 1.3240250361097736,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00016741804805277434,
+      "loss": 0.5134,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3240250361097736,
+      "eval_loss": 0.4434724748134613,
+      "eval_runtime": 2.3562,
+      "eval_samples_per_second": 84.881,
+      "eval_steps_per_second": 84.881,
+      "step": 5500
+    },
+    {
+      "epoch": 1.32522869523351,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001673553754768216,
+      "loss": 0.5231,
+      "step": 5505
+    },
+    {
+      "epoch": 1.326432354357246,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00016729266773663048,
+      "loss": 0.5308,
+      "step": 5510
+    },
+    {
+      "epoch": 1.3276360134809821,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016722992488847914,
+      "loss": 0.4797,
+      "step": 5515
+    },
+    {
+      "epoch": 1.3288396726047185,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00016716714698867714,
+      "loss": 0.505,
+      "step": 5520
+    },
+    {
+      "epoch": 1.3300433317284546,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00016710433409356562,
+      "loss": 0.5195,
+      "step": 5525
+    },
+    {
+      "epoch": 1.3312469908521907,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00016704148625951696,
+      "loss": 0.4942,
+      "step": 5530
+    },
+    {
+      "epoch": 1.3324506499759268,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00016697860354293503,
+      "loss": 0.5068,
+      "step": 5535
+    },
+    {
+      "epoch": 1.3336543090996629,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.000166915686000255,
+      "loss": 0.4945,
+      "step": 5540
+    },
+    {
+      "epoch": 1.3348579682233992,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0001668527336879432,
+      "loss": 0.5158,
+      "step": 5545
+    },
+    {
+      "epoch": 1.3360616273471353,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00016678974666249727,
+      "loss": 0.5167,
+      "step": 5550
+    },
+    {
+      "epoch": 1.3372652864708714,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.000166726724980446,
+      "loss": 0.508,
+      "step": 5555
+    },
+    {
+      "epoch": 1.3384689455946077,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00016666366869834913,
+      "loss": 0.5203,
+      "step": 5560
+    },
+    {
+      "epoch": 1.3396726047183438,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00016660057787279764,
+      "loss": 0.535,
+      "step": 5565
+    },
+    {
+      "epoch": 1.34087626384208,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00016653745256041346,
+      "loss": 0.4892,
+      "step": 5570
+    },
+    {
+      "epoch": 1.342079922965816,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00016647429281784937,
+      "loss": 0.5311,
+      "step": 5575
+    },
+    {
+      "epoch": 1.3432835820895521,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001664110987017892,
+      "loss": 0.5063,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3444872412132884,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00016634787026894752,
+      "loss": 0.4854,
+      "step": 5585
+    },
+    {
+      "epoch": 1.3456909003370245,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016628460757606972,
+      "loss": 0.4942,
+      "step": 5590
+    },
+    {
+      "epoch": 1.3468945594607606,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016622131067993198,
+      "loss": 0.5466,
+      "step": 5595
+    },
+    {
+      "epoch": 1.348098218584497,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00016615797963734117,
+      "loss": 0.525,
+      "step": 5600
+    },
+    {
+      "epoch": 1.349301877708233,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00016609461450513472,
+      "loss": 0.4876,
+      "step": 5605
+    },
+    {
+      "epoch": 1.3505055368319692,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00016603121534018077,
+      "loss": 0.5088,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3517091959557053,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016596778219937796,
+      "loss": 0.5006,
+      "step": 5615
+    },
+    {
+      "epoch": 1.3529128550794414,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001659043151396554,
+      "loss": 0.4964,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3541165142031777,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001658408142179726,
+      "loss": 0.5264,
+      "step": 5625
+    },
+    {
+      "epoch": 1.3553201733269138,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016577727949131958,
+      "loss": 0.4978,
+      "step": 5630
+    },
+    {
+      "epoch": 1.35652383245065,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00016571371101671662,
+      "loss": 0.4969,
+      "step": 5635
+    },
+    {
+      "epoch": 1.3577274915743862,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00016565010885121427,
+      "loss": 0.5162,
+      "step": 5640
+    },
+    {
+      "epoch": 1.3589311506981223,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00016558647305189343,
+      "loss": 0.5403,
+      "step": 5645
+    },
+    {
+      "epoch": 1.3601348098218584,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00016552280367586499,
+      "loss": 0.5379,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3613384689455947,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016545910078027016,
+      "loss": 0.5041,
+      "step": 5655
+    },
+    {
+      "epoch": 1.3625421280693308,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00016539536442228012,
+      "loss": 0.5028,
+      "step": 5660
+    },
+    {
+      "epoch": 1.363745787193067,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001653315946590962,
+      "loss": 0.5029,
+      "step": 5665
+    },
+    {
+      "epoch": 1.364949446316803,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00016526779154794953,
+      "loss": 0.498,
+      "step": 5670
+    },
+    {
+      "epoch": 1.3661531054405391,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00016520395514610131,
+      "loss": 0.5071,
+      "step": 5675
+    },
+    {
+      "epoch": 1.3673567645642755,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00016514008551084258,
+      "loss": 0.4817,
+      "step": 5680
+    },
+    {
+      "epoch": 1.3685604236880116,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00016507618269949424,
+      "loss": 0.5214,
+      "step": 5685
+    },
+    {
+      "epoch": 1.3697640828117477,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016501224676940685,
+      "loss": 0.4726,
+      "step": 5690
+    },
+    {
+      "epoch": 1.370967741935484,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00016494827777796084,
+      "loss": 0.5419,
+      "step": 5695
+    },
+    {
+      "epoch": 1.37217140105922,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00016488427578256618,
+      "loss": 0.5134,
+      "step": 5700
+    },
+    {
+      "epoch": 1.3733750601829562,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016482024084066258,
+      "loss": 0.5023,
+      "step": 5705
+    },
+    {
+      "epoch": 1.3745787193066923,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016475617300971924,
+      "loss": 0.4796,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3757823784304284,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00016469207234723487,
+      "loss": 0.5287,
+      "step": 5715
+    },
+    {
+      "epoch": 1.3769860375541647,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0001646279389107377,
+      "loss": 0.4952,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3781896966779008,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016456377275778536,
+      "loss": 0.511,
+      "step": 5725
+    },
+    {
+      "epoch": 1.379393355801637,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00016449957394596475,
+      "loss": 0.4951,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3805970149253732,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016443534253289226,
+      "loss": 0.5266,
+      "step": 5735
+    },
+    {
+      "epoch": 1.3818006740491093,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016437107857621337,
+      "loss": 0.5175,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3830043331728454,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00016430678213360286,
+      "loss": 0.4723,
+      "step": 5745
+    },
+    {
+      "epoch": 1.3842079922965815,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016424245326276466,
+      "loss": 0.5143,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3854116514203176,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001641780920214317,
+      "loss": 0.5171,
+      "step": 5755
+    },
+    {
+      "epoch": 1.386615310544054,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00016411369846736613,
+      "loss": 0.5328,
+      "step": 5760
+    },
+    {
+      "epoch": 1.38781896966779,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016404927265835896,
+      "loss": 0.4887,
+      "step": 5765
+    },
+    {
+      "epoch": 1.3890226287915262,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00016398481465223024,
+      "loss": 0.4922,
+      "step": 5770
+    },
+    {
+      "epoch": 1.3902262879152625,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016392032450682884,
+      "loss": 0.5218,
+      "step": 5775
+    },
+    {
+      "epoch": 1.3914299470389986,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00016385580228003252,
+      "loss": 0.4743,
+      "step": 5780
+    },
+    {
+      "epoch": 1.3926336061627347,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00016379124802974783,
+      "loss": 0.5239,
+      "step": 5785
+    },
+    {
+      "epoch": 1.393837265286471,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00016372666181391004,
+      "loss": 0.5002,
+      "step": 5790
+    },
+    {
+      "epoch": 1.395040924410207,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00016366204369048316,
+      "loss": 0.4936,
+      "step": 5795
+    },
+    {
+      "epoch": 1.3962445835339432,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016359739371745973,
+      "loss": 0.5055,
+      "step": 5800
+    },
+    {
+      "epoch": 1.3974482426576793,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.000163532711952861,
+      "loss": 0.5294,
+      "step": 5805
+    },
+    {
+      "epoch": 1.3986519017814154,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016346799845473665,
+      "loss": 0.5311,
+      "step": 5810
+    },
+    {
+      "epoch": 1.3998555609051517,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001634032532811649,
+      "loss": 0.511,
+      "step": 5815
+    },
+    {
+      "epoch": 1.4010592200288878,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00016333847649025238,
+      "loss": 0.5022,
+      "step": 5820
+    },
+    {
+      "epoch": 1.402262879152624,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001632736681401341,
+      "loss": 0.5103,
+      "step": 5825
+    },
+    {
+      "epoch": 1.4034665382763603,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00016320882828897335,
+      "loss": 0.5154,
+      "step": 5830
+    },
+    {
+      "epoch": 1.4046701974000964,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00016314395699496177,
+      "loss": 0.524,
+      "step": 5835
+    },
+    {
+      "epoch": 1.4058738565238325,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016307905431631916,
+      "loss": 0.5034,
+      "step": 5840
+    },
+    {
+      "epoch": 1.4070775156475686,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016301412031129354,
+      "loss": 0.4704,
+      "step": 5845
+    },
+    {
+      "epoch": 1.4082811747713047,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00016294915503816098,
+      "loss": 0.4753,
+      "step": 5850
+    },
+    {
+      "epoch": 1.409484833895041,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001628841585552256,
+      "loss": 0.4812,
+      "step": 5855
+    },
+    {
+      "epoch": 1.410688493018777,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016281913092081965,
+      "loss": 0.4752,
+      "step": 5860
+    },
+    {
+      "epoch": 1.4118921521425132,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001627540721933032,
+      "loss": 0.4794,
+      "step": 5865
+    },
+    {
+      "epoch": 1.4130958112662495,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00016268898243106431,
+      "loss": 0.4873,
+      "step": 5870
+    },
+    {
+      "epoch": 1.4142994703899856,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016262386169251888,
+      "loss": 0.4812,
+      "step": 5875
+    },
+    {
+      "epoch": 1.4155031295137217,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016255871003611054,
+      "loss": 0.4913,
+      "step": 5880
+    },
+    {
+      "epoch": 1.4167067886374578,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00016249352752031074,
+      "loss": 0.476,
+      "step": 5885
+    },
+    {
+      "epoch": 1.417910447761194,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00016242831420361871,
+      "loss": 0.4951,
+      "step": 5890
+    },
+    {
+      "epoch": 1.4191141068849302,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001623630701445611,
+      "loss": 0.4959,
+      "step": 5895
+    },
+    {
+      "epoch": 1.4203177660086663,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001622977954016923,
+      "loss": 0.4744,
+      "step": 5900
+    },
+    {
+      "epoch": 1.4215214251324024,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00016223249003359424,
+      "loss": 0.4864,
+      "step": 5905
+    },
+    {
+      "epoch": 1.4227250842561388,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016216715409887627,
+      "loss": 0.508,
+      "step": 5910
+    },
+    {
+      "epoch": 1.4239287433798749,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00016210178765617525,
+      "loss": 0.5252,
+      "step": 5915
+    },
+    {
+      "epoch": 1.425132402503611,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00016203639076415533,
+      "loss": 0.5037,
+      "step": 5920
+    },
+    {
+      "epoch": 1.4263360616273473,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016197096348150806,
+      "loss": 0.4805,
+      "step": 5925
+    },
+    {
+      "epoch": 1.4275397207510834,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016190550586695228,
+      "loss": 0.4931,
+      "step": 5930
+    },
+    {
+      "epoch": 1.4287433798748195,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00016184001797923394,
+      "loss": 0.5004,
+      "step": 5935
+    },
+    {
+      "epoch": 1.4299470389985556,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00016177449987712626,
+      "loss": 0.5099,
+      "step": 5940
+    },
+    {
+      "epoch": 1.4311506981222917,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016170895161942954,
+      "loss": 0.5029,
+      "step": 5945
+    },
+    {
+      "epoch": 1.432354357246028,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00016164337326497118,
+      "loss": 0.4897,
+      "step": 5950
+    },
+    {
+      "epoch": 1.433558016369764,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00016157776487260545,
+      "loss": 0.5245,
+      "step": 5955
+    },
+    {
+      "epoch": 1.4347616754935002,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00016151212650121388,
+      "loss": 0.5082,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4359653346172365,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016144645820970457,
+      "loss": 0.4988,
+      "step": 5965
+    },
+    {
+      "epoch": 1.4371689937409726,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001613807600570127,
+      "loss": 0.48,
+      "step": 5970
+    },
+    {
+      "epoch": 1.4383726528647087,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00016131503210210007,
+      "loss": 0.5006,
+      "step": 5975
+    },
+    {
+      "epoch": 1.4395763119884448,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00016124927440395537,
+      "loss": 0.5205,
+      "step": 5980
+    },
+    {
+      "epoch": 1.440779971112181,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00016118348702159397,
+      "loss": 0.5068,
+      "step": 5985
+    },
+    {
+      "epoch": 1.4419836302359172,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00016111767001405784,
+      "loss": 0.4937,
+      "step": 5990
+    },
+    {
+      "epoch": 1.4431872893596533,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.0001610518234404155,
+      "loss": 0.5264,
+      "step": 5995
+    },
+    {
+      "epoch": 1.4443909484833894,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00016098594735976205,
+      "loss": 0.5072,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4443909484833894,
+      "eval_loss": 0.4375569820404053,
+      "eval_runtime": 2.3354,
+      "eval_samples_per_second": 85.637,
+      "eval_steps_per_second": 85.637,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4455946076071258,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016092004183121912,
+      "loss": 0.5059,
+      "step": 6005
+    },
+    {
+      "epoch": 1.4467982667308619,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001608541069139347,
+      "loss": 0.4873,
+      "step": 6010
+    },
+    {
+      "epoch": 1.448001925854598,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00016078814266708315,
+      "loss": 0.4887,
+      "step": 6015
+    },
+    {
+      "epoch": 1.449205584978334,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001607221491498652,
+      "loss": 0.4839,
+      "step": 6020
+    },
+    {
+      "epoch": 1.4504092441020702,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001606561264215078,
+      "loss": 0.5151,
+      "step": 6025
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00016059007454126417,
+      "loss": 0.477,
+      "step": 6030
+    },
+    {
+      "epoch": 1.4528165623495426,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00016052399356841367,
+      "loss": 0.5034,
+      "step": 6035
+    },
+    {
+      "epoch": 1.4540202214732787,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00016045788356226174,
+      "loss": 0.4674,
+      "step": 6040
+    },
+    {
+      "epoch": 1.455223880597015,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001603917445821399,
+      "loss": 0.483,
+      "step": 6045
+    },
+    {
+      "epoch": 1.4564275397207511,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00016032557668740574,
+      "loss": 0.5046,
+      "step": 6050
+    },
+    {
+      "epoch": 1.4576311988444872,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016025937993744262,
+      "loss": 0.5304,
+      "step": 6055
+    },
+    {
+      "epoch": 1.4588348579682233,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00016019315439165998,
+      "loss": 0.5065,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4600385170919594,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016012690010949306,
+      "loss": 0.4918,
+      "step": 6065
+    },
+    {
+      "epoch": 1.4612421762156957,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00016006061715040279,
+      "loss": 0.4819,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4624458353394318,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00015999430557387596,
+      "loss": 0.5011,
+      "step": 6075
+    },
+    {
+      "epoch": 1.463649494463168,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00015992796543942497,
+      "loss": 0.5366,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4648531535869043,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001598615968065879,
+      "loss": 0.4865,
+      "step": 6085
+    },
+    {
+      "epoch": 1.4660568127106404,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00015979519973492837,
+      "loss": 0.4808,
+      "step": 6090
+    },
+    {
+      "epoch": 1.4672604718343765,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00015972877428403547,
+      "loss": 0.5087,
+      "step": 6095
+    },
+    {
+      "epoch": 1.4684641309581128,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00015966232051352393,
+      "loss": 0.5038,
+      "step": 6100
+    },
+    {
+      "epoch": 1.469667790081849,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00015959583848303372,
+      "loss": 0.5011,
+      "step": 6105
+    },
+    {
+      "epoch": 1.470871449205585,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00015952932825223027,
+      "loss": 0.4811,
+      "step": 6110
+    },
+    {
+      "epoch": 1.472075108329321,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00015946278988080426,
+      "loss": 0.5142,
+      "step": 6115
+    },
+    {
+      "epoch": 1.4732787674530572,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015939622342847168,
+      "loss": 0.5151,
+      "step": 6120
+    },
+    {
+      "epoch": 1.4744824265767935,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00015932962895497368,
+      "loss": 0.4925,
+      "step": 6125
+    },
+    {
+      "epoch": 1.4756860857005296,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001592630065200766,
+      "loss": 0.5025,
+      "step": 6130
+    },
+    {
+      "epoch": 1.4768897448242657,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00015919635618357177,
+      "loss": 0.5178,
+      "step": 6135
+    },
+    {
+      "epoch": 1.478093403948002,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00015912967800527574,
+      "loss": 0.5206,
+      "step": 6140
+    },
+    {
+      "epoch": 1.4792970630717381,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001590629720450299,
+      "loss": 0.52,
+      "step": 6145
+    },
+    {
+      "epoch": 1.4805007221954742,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.00015899623836270065,
+      "loss": 0.5341,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4817043813192103,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00015892947701817914,
+      "loss": 0.4613,
+      "step": 6155
+    },
+    {
+      "epoch": 1.4829080404429464,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001588626880713816,
+      "loss": 0.4838,
+      "step": 6160
+    },
+    {
+      "epoch": 1.4841116995666828,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00015879587158224877,
+      "loss": 0.5102,
+      "step": 6165
+    },
+    {
+      "epoch": 1.4853153586904189,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001587290276107462,
+      "loss": 0.5193,
+      "step": 6170
+    },
+    {
+      "epoch": 1.486519017814155,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001586621562168642,
+      "loss": 0.4968,
+      "step": 6175
+    },
+    {
+      "epoch": 1.4877226769378913,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00015859525746061756,
+      "loss": 0.5013,
+      "step": 6180
+    },
+    {
+      "epoch": 1.4889263360616274,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001585283314020457,
+      "loss": 0.4973,
+      "step": 6185
+    },
+    {
+      "epoch": 1.4901299951853635,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00015846137810121252,
+      "loss": 0.4933,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4913336543090996,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00015839439761820633,
+      "loss": 0.5009,
+      "step": 6195
+    },
+    {
+      "epoch": 1.4925373134328357,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001583273900131399,
+      "loss": 0.5264,
+      "step": 6200
+    },
+    {
+      "epoch": 1.493740972556572,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00015826035534615037,
+      "loss": 0.4874,
+      "step": 6205
+    },
+    {
+      "epoch": 1.4949446316803081,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00015819329367739902,
+      "loss": 0.5281,
+      "step": 6210
+    },
+    {
+      "epoch": 1.4961482908040442,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00015812620506707153,
+      "loss": 0.4776,
+      "step": 6215
+    },
+    {
+      "epoch": 1.4973519499277805,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001580590895753776,
+      "loss": 0.5018,
+      "step": 6220
+    },
+    {
+      "epoch": 1.4985556090515166,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00015799194726255123,
+      "loss": 0.4809,
+      "step": 6225
+    },
+    {
+      "epoch": 1.4997592681752527,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001579247781888504,
+      "loss": 0.4903,
+      "step": 6230
+    },
+    {
+      "epoch": 1.500962927298989,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00015785758241455703,
+      "loss": 0.4824,
+      "step": 6235
+    },
+    {
+      "epoch": 1.502166586422725,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00015779035999997714,
+      "loss": 0.5151,
+      "step": 6240
+    },
+    {
+      "epoch": 1.5033702455464613,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001577231110054406,
+      "loss": 0.4715,
+      "step": 6245
+    },
+    {
+      "epoch": 1.5045739046701974,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001576558354913011,
+      "loss": 0.4438,
+      "step": 6250
+    },
+    {
+      "epoch": 1.5057775637939335,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001575885335179362,
+      "loss": 0.4964,
+      "step": 6255
+    },
+    {
+      "epoch": 1.5069812229176698,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00015752120514574716,
+      "loss": 0.4984,
+      "step": 6260
+    },
+    {
+      "epoch": 1.5081848820414059,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00015745385043515892,
+      "loss": 0.5073,
+      "step": 6265
+    },
+    {
+      "epoch": 1.509388541165142,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00015738646944662014,
+      "loss": 0.5178,
+      "step": 6270
+    },
+    {
+      "epoch": 1.5105922002888783,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015731906224060297,
+      "loss": 0.5069,
+      "step": 6275
+    },
+    {
+      "epoch": 1.5117958594126142,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00015725162887760309,
+      "loss": 0.4943,
+      "step": 6280
+    },
+    {
+      "epoch": 1.5129995185363505,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015718416941813975,
+      "loss": 0.5241,
+      "step": 6285
+    },
+    {
+      "epoch": 1.5142031776600868,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001571166839227555,
+      "loss": 0.5058,
+      "step": 6290
+    },
+    {
+      "epoch": 1.5154068367838227,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00015704917245201638,
+      "loss": 0.472,
+      "step": 6295
+    },
+    {
+      "epoch": 1.516610495907559,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015698163506651163,
+      "loss": 0.4877,
+      "step": 6300
+    },
+    {
+      "epoch": 1.5178141550312951,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00015691407182685384,
+      "loss": 0.4793,
+      "step": 6305
+    },
+    {
+      "epoch": 1.5190178141550312,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00015684648279367872,
+      "loss": 0.5021,
+      "step": 6310
+    },
+    {
+      "epoch": 1.5202214732787676,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015677886802764517,
+      "loss": 0.5212,
+      "step": 6315
+    },
+    {
+      "epoch": 1.5214251324025037,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001567112275894352,
+      "loss": 0.529,
+      "step": 6320
+    },
+    {
+      "epoch": 1.5226287915262398,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015664356153975383,
+      "loss": 0.4912,
+      "step": 6325
+    },
+    {
+      "epoch": 1.523832450649976,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001565758699393291,
+      "loss": 0.5074,
+      "step": 6330
+    },
+    {
+      "epoch": 1.525036109773712,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001565081528489119,
+      "loss": 0.5058,
+      "step": 6335
+    },
+    {
+      "epoch": 1.5262397688974483,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00015644041032927616,
+      "loss": 0.5007,
+      "step": 6340
+    },
+    {
+      "epoch": 1.5274434280211844,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00015637264244121844,
+      "loss": 0.4619,
+      "step": 6345
+    },
+    {
+      "epoch": 1.5286470871449205,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00015630484924555814,
+      "loss": 0.4976,
+      "step": 6350
+    },
+    {
+      "epoch": 1.5298507462686568,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015623703080313748,
+      "loss": 0.4578,
+      "step": 6355
+    },
+    {
+      "epoch": 1.531054405392393,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00015616918717482116,
+      "loss": 0.4891,
+      "step": 6360
+    },
+    {
+      "epoch": 1.532258064516129,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00015610131842149662,
+      "loss": 0.4603,
+      "step": 6365
+    },
+    {
+      "epoch": 1.5334617236398653,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00015603342460407377,
+      "loss": 0.4424,
+      "step": 6370
+    },
+    {
+      "epoch": 1.5346653827636012,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00015596550578348507,
+      "loss": 0.5114,
+      "step": 6375
+    },
+    {
+      "epoch": 1.5358690418873375,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.0001558975620206853,
+      "loss": 0.4539,
+      "step": 6380
+    },
+    {
+      "epoch": 1.5370727010110736,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00015582959337665187,
+      "loss": 0.4946,
+      "step": 6385
+    },
+    {
+      "epoch": 1.5382763601348097,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00015576159991238423,
+      "loss": 0.4861,
+      "step": 6390
+    },
+    {
+      "epoch": 1.539480019258546,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015569358168890436,
+      "loss": 0.5153,
+      "step": 6395
+    },
+    {
+      "epoch": 1.5406836783822822,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00015562553876725623,
+      "loss": 0.4729,
+      "step": 6400
+    },
+    {
+      "epoch": 1.5418873375060183,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00015555747120850615,
+      "loss": 0.4959,
+      "step": 6405
+    },
+    {
+      "epoch": 1.5430909966297546,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015548937907374245,
+      "loss": 0.5108,
+      "step": 6410
+    },
+    {
+      "epoch": 1.5442946557534905,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00015542126242407557,
+      "loss": 0.4787,
+      "step": 6415
+    },
+    {
+      "epoch": 1.5454983148772268,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001553531213206379,
+      "loss": 0.5225,
+      "step": 6420
+    },
+    {
+      "epoch": 1.5467019740009629,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015528495582458384,
+      "loss": 0.4815,
+      "step": 6425
+    },
+    {
+      "epoch": 1.547905633124699,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015521676599708958,
+      "loss": 0.5292,
+      "step": 6430
+    },
+    {
+      "epoch": 1.5491092922484353,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001551485518993533,
+      "loss": 0.4928,
+      "step": 6435
+    },
+    {
+      "epoch": 1.5503129513721714,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00015508031359259483,
+      "loss": 0.4872,
+      "step": 6440
+    },
+    {
+      "epoch": 1.5515166104959075,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00015501205113805573,
+      "loss": 0.4895,
+      "step": 6445
+    },
+    {
+      "epoch": 1.5527202696196438,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00015494376459699932,
+      "loss": 0.5001,
+      "step": 6450
+    },
+    {
+      "epoch": 1.55392392874338,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001548754540307105,
+      "loss": 0.493,
+      "step": 6455
+    },
+    {
+      "epoch": 1.555127587867116,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00015480711950049565,
+      "loss": 0.4657,
+      "step": 6460
+    },
+    {
+      "epoch": 1.5563312469908523,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00015473876106768283,
+      "loss": 0.4959,
+      "step": 6465
+    },
+    {
+      "epoch": 1.5575349061145882,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00015467037879362134,
+      "loss": 0.4825,
+      "step": 6470
+    },
+    {
+      "epoch": 1.5587385652383245,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015460197273968205,
+      "loss": 0.5103,
+      "step": 6475
+    },
+    {
+      "epoch": 1.5599422243620606,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00015453354296725709,
+      "loss": 0.4819,
+      "step": 6480
+    },
+    {
+      "epoch": 1.5611458834857967,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0001544650895377599,
+      "loss": 0.4982,
+      "step": 6485
+    },
+    {
+      "epoch": 1.562349542609533,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015439661251262515,
+      "loss": 0.5325,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5635532017332692,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015432811195330867,
+      "loss": 0.4814,
+      "step": 6495
+    },
+    {
+      "epoch": 1.5647568608570053,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001542595879212874,
+      "loss": 0.5118,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5647568608570053,
+      "eval_loss": 0.41606834530830383,
+      "eval_runtime": 2.3602,
+      "eval_samples_per_second": 84.737,
+      "eval_steps_per_second": 84.737,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5659605199807416,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001541910404780594,
+      "loss": 0.5079,
+      "step": 6505
+    },
+    {
+      "epoch": 1.5671641791044775,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00015412246968514375,
+      "loss": 0.4948,
+      "step": 6510
+    },
+    {
+      "epoch": 1.5683678382282138,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015405387560408037,
+      "loss": 0.4918,
+      "step": 6515
+    },
+    {
+      "epoch": 1.56957149735195,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00015398525829643013,
+      "loss": 0.4902,
+      "step": 6520
+    },
+    {
+      "epoch": 1.570775156475686,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00015391661782377487,
+      "loss": 0.4527,
+      "step": 6525
+    },
+    {
+      "epoch": 1.5719788155994223,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00015384795424771706,
+      "loss": 0.5237,
+      "step": 6530
+    },
+    {
+      "epoch": 1.5731824747231584,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00015377926762987997,
+      "loss": 0.5072,
+      "step": 6535
+    },
+    {
+      "epoch": 1.5743861338468945,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001537105580319076,
+      "loss": 0.4792,
+      "step": 6540
+    },
+    {
+      "epoch": 1.5755897929706308,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001536418255154644,
+      "loss": 0.4989,
+      "step": 6545
+    },
+    {
+      "epoch": 1.5767934520943667,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0001535730701422356,
+      "loss": 0.4838,
+      "step": 6550
+    },
+    {
+      "epoch": 1.577997111218103,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015350429197392686,
+      "loss": 0.4829,
+      "step": 6555
+    },
+    {
+      "epoch": 1.5792007703418391,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00015343549107226418,
+      "loss": 0.479,
+      "step": 6560
+    },
+    {
+      "epoch": 1.5804044294655752,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001533666674989942,
+      "loss": 0.4809,
+      "step": 6565
+    },
+    {
+      "epoch": 1.5816080885893116,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00015329782131588368,
+      "loss": 0.496,
+      "step": 6570
+    },
+    {
+      "epoch": 1.5828117477130477,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001532289525847198,
+      "loss": 0.5168,
+      "step": 6575
+    },
+    {
+      "epoch": 1.5840154068367838,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00015316006136731,
+      "loss": 0.4972,
+      "step": 6580
+    },
+    {
+      "epoch": 1.58521906596052,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015309114772548175,
+      "loss": 0.4799,
+      "step": 6585
+    },
+    {
+      "epoch": 1.5864227250842562,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001530222117210828,
+      "loss": 0.4974,
+      "step": 6590
+    },
+    {
+      "epoch": 1.5876263842079923,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00015295325341598088,
+      "loss": 0.4733,
+      "step": 6595
+    },
+    {
+      "epoch": 1.5888300433317286,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.0001528842728720638,
+      "loss": 0.4748,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5900337024554645,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00015281527015123924,
+      "loss": 0.5008,
+      "step": 6605
+    },
+    {
+      "epoch": 1.5912373615792008,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001527462453154349,
+      "loss": 0.485,
+      "step": 6610
+    },
+    {
+      "epoch": 1.592441020702937,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001526771984265982,
+      "loss": 0.5183,
+      "step": 6615
+    },
+    {
+      "epoch": 1.593644679826673,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00015260812954669646,
+      "loss": 0.4841,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5948483389504093,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015253903873771666,
+      "loss": 0.4371,
+      "step": 6625
+    },
+    {
+      "epoch": 1.5960519980741454,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00015246992606166552,
+      "loss": 0.4923,
+      "step": 6630
+    },
+    {
+      "epoch": 1.5972556571978815,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00015240079158056934,
+      "loss": 0.4436,
+      "step": 6635
+    },
+    {
+      "epoch": 1.5984593163216179,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00015233163535647402,
+      "loss": 0.4747,
+      "step": 6640
+    },
+    {
+      "epoch": 1.5996629754453537,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00015226245745144493,
+      "loss": 0.5051,
+      "step": 6645
+    },
+    {
+      "epoch": 1.60086663456909,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00015219325792756696,
+      "loss": 0.4876,
+      "step": 6650
+    },
+    {
+      "epoch": 1.6020702936928262,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00015212403684694434,
+      "loss": 0.4998,
+      "step": 6655
+    },
+    {
+      "epoch": 1.6032739528165623,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001520547942717007,
+      "loss": 0.5161,
+      "step": 6660
+    },
+    {
+      "epoch": 1.6044776119402986,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00015198553026397893,
+      "loss": 0.5203,
+      "step": 6665
+    },
+    {
+      "epoch": 1.6056812710640347,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00015191624488594116,
+      "loss": 0.5167,
+      "step": 6670
+    },
+    {
+      "epoch": 1.6068849301877708,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00015184693819976868,
+      "loss": 0.5026,
+      "step": 6675
+    },
+    {
+      "epoch": 1.608088589311507,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.000151777610267662,
+      "loss": 0.5318,
+      "step": 6680
+    },
+    {
+      "epoch": 1.609292248435243,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00015170826115184054,
+      "loss": 0.5074,
+      "step": 6685
+    },
+    {
+      "epoch": 1.6104959075589793,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00015163889091454285,
+      "loss": 0.4698,
+      "step": 6690
+    },
+    {
+      "epoch": 1.6116995666827154,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00015156949961802637,
+      "loss": 0.4793,
+      "step": 6695
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001515000873245675,
+      "loss": 0.4919,
+      "step": 6700
+    },
+    {
+      "epoch": 1.6141068849301878,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015143065409646143,
+      "loss": 0.4977,
+      "step": 6705
+    },
+    {
+      "epoch": 1.615310544053924,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001513611999960222,
+      "loss": 0.4719,
+      "step": 6710
+    },
+    {
+      "epoch": 1.61651420317766,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00015129172508558252,
+      "loss": 0.4371,
+      "step": 6715
+    },
+    {
+      "epoch": 1.6177178623013964,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00015122222942749378,
+      "loss": 0.5116,
+      "step": 6720
+    },
+    {
+      "epoch": 1.6189215214251322,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00015115271308412603,
+      "loss": 0.4868,
+      "step": 6725
+    },
+    {
+      "epoch": 1.6201251805488686,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00015108317611786784,
+      "loss": 0.4558,
+      "step": 6730
+    },
+    {
+      "epoch": 1.6213288396726049,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001510136185911264,
+      "loss": 0.4995,
+      "step": 6735
+    },
+    {
+      "epoch": 1.6225324987963408,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00015094404056632712,
+      "loss": 0.4806,
+      "step": 6740
+    },
+    {
+      "epoch": 1.623736157920077,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00015087444210591406,
+      "loss": 0.4915,
+      "step": 6745
+    },
+    {
+      "epoch": 1.6249398170438132,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015080482327234953,
+      "loss": 0.4955,
+      "step": 6750
+    },
+    {
+      "epoch": 1.6261434761675493,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00015073518412811403,
+      "loss": 0.5119,
+      "step": 6755
+    },
+    {
+      "epoch": 1.6273471352912856,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001506655247357064,
+      "loss": 0.4954,
+      "step": 6760
+    },
+    {
+      "epoch": 1.6285507944150217,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00015059584515764364,
+      "loss": 0.5023,
+      "step": 6765
+    },
+    {
+      "epoch": 1.6297544535387578,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00015052614545646076,
+      "loss": 0.4907,
+      "step": 6770
+    },
+    {
+      "epoch": 1.6309581126624941,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.000150456425694711,
+      "loss": 0.4661,
+      "step": 6775
+    },
+    {
+      "epoch": 1.63216177178623,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00015038668593496548,
+      "loss": 0.4697,
+      "step": 6780
+    },
+    {
+      "epoch": 1.6333654309099663,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00015031692623981328,
+      "loss": 0.4583,
+      "step": 6785
+    },
+    {
+      "epoch": 1.6345690900337024,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00015024714667186142,
+      "loss": 0.5061,
+      "step": 6790
+    },
+    {
+      "epoch": 1.6357727491574385,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001501773472937348,
+      "loss": 0.4883,
+      "step": 6795
+    },
+    {
+      "epoch": 1.6369764082811749,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00015010752816807586,
+      "loss": 0.5046,
+      "step": 6800
+    },
+    {
+      "epoch": 1.638180067404911,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00015003768935754511,
+      "loss": 0.4965,
+      "step": 6805
+    },
+    {
+      "epoch": 1.639383726528647,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014996783092482041,
+      "loss": 0.4708,
+      "step": 6810
+    },
+    {
+      "epoch": 1.6405873856523834,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001498979529325975,
+      "loss": 0.4768,
+      "step": 6815
+    },
+    {
+      "epoch": 1.6417910447761193,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014982805544358938,
+      "loss": 0.4742,
+      "step": 6820
+    },
+    {
+      "epoch": 1.6429947038998556,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00014975813852052683,
+      "loss": 0.4778,
+      "step": 6825
+    },
+    {
+      "epoch": 1.6441983630235917,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00014968820222615795,
+      "loss": 0.5008,
+      "step": 6830
+    },
+    {
+      "epoch": 1.6454020221473278,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00014961824662324818,
+      "loss": 0.4684,
+      "step": 6835
+    },
+    {
+      "epoch": 1.646605681271064,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00014954827177458035,
+      "loss": 0.4979,
+      "step": 6840
+    },
+    {
+      "epoch": 1.6478093403948002,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00014947827774295454,
+      "loss": 0.4698,
+      "step": 6845
+    },
+    {
+      "epoch": 1.6490129995185363,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001494082645911881,
+      "loss": 0.5106,
+      "step": 6850
+    },
+    {
+      "epoch": 1.6502166586422726,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00014933823238211542,
+      "loss": 0.4597,
+      "step": 6855
+    },
+    {
+      "epoch": 1.6514203177660085,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00014926818117858812,
+      "loss": 0.4659,
+      "step": 6860
+    },
+    {
+      "epoch": 1.6526239768897448,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001491981110434748,
+      "loss": 0.4786,
+      "step": 6865
+    },
+    {
+      "epoch": 1.653827636013481,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014912802203966103,
+      "loss": 0.5137,
+      "step": 6870
+    },
+    {
+      "epoch": 1.655031295137217,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00014905791423004946,
+      "loss": 0.4361,
+      "step": 6875
+    },
+    {
+      "epoch": 1.6562349542609534,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00014898778767755938,
+      "loss": 0.4909,
+      "step": 6880
+    },
+    {
+      "epoch": 1.6574386133846895,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00014891764244512706,
+      "loss": 0.5199,
+      "step": 6885
+    },
+    {
+      "epoch": 1.6586422725084256,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001488474785957055,
+      "loss": 0.5022,
+      "step": 6890
+    },
+    {
+      "epoch": 1.6598459316321619,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00014877729619226448,
+      "loss": 0.4733,
+      "step": 6895
+    },
+    {
+      "epoch": 1.661049590755898,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00014870709529779026,
+      "loss": 0.471,
+      "step": 6900
+    },
+    {
+      "epoch": 1.662253249879634,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014863687597528588,
+      "loss": 0.4876,
+      "step": 6905
+    },
+    {
+      "epoch": 1.6634569090033704,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00014856663828777077,
+      "loss": 0.4721,
+      "step": 6910
+    },
+    {
+      "epoch": 1.6646605681271063,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014849638229828095,
+      "loss": 0.5007,
+      "step": 6915
+    },
+    {
+      "epoch": 1.6658642272508426,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014842610806986881,
+      "loss": 0.484,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6670678863745787,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00014835581566560312,
+      "loss": 0.4783,
+      "step": 6925
+    },
+    {
+      "epoch": 1.6682715454983148,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00014828550514856895,
+      "loss": 0.5209,
+      "step": 6930
+    },
+    {
+      "epoch": 1.6694752046220511,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00014821517658186763,
+      "loss": 0.464,
+      "step": 6935
+    },
+    {
+      "epoch": 1.6706788637457872,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00014814483002861673,
+      "loss": 0.4731,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6718825228695233,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00014807446555194996,
+      "loss": 0.5083,
+      "step": 6945
+    },
+    {
+      "epoch": 1.6730861819932596,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.000148004083215017,
+      "loss": 0.4457,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6742898411169955,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014793368308098367,
+      "loss": 0.4951,
+      "step": 6955
+    },
+    {
+      "epoch": 1.6754935002407318,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001478632652130318,
+      "loss": 0.4611,
+      "step": 6960
+    },
+    {
+      "epoch": 1.676697159364468,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00014779282967435902,
+      "loss": 0.5022,
+      "step": 6965
+    },
+    {
+      "epoch": 1.677900818488204,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00014772237652817887,
+      "loss": 0.4627,
+      "step": 6970
+    },
+    {
+      "epoch": 1.6791044776119404,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001476519058377207,
+      "loss": 0.4781,
+      "step": 6975
+    },
+    {
+      "epoch": 1.6803081367356765,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001475814176662296,
+      "loss": 0.4507,
+      "step": 6980
+    },
+    {
+      "epoch": 1.6815117958594126,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001475109120769664,
+      "loss": 0.4777,
+      "step": 6985
+    },
+    {
+      "epoch": 1.682715454983149,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001474403891332074,
+      "loss": 0.5116,
+      "step": 6990
+    },
+    {
+      "epoch": 1.6839191141068848,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00014736984889824467,
+      "loss": 0.4651,
+      "step": 6995
+    },
+    {
+      "epoch": 1.685122773230621,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014729929143538569,
+      "loss": 0.4814,
+      "step": 7000
+    },
+    {
+      "epoch": 1.685122773230621,
+      "eval_loss": 0.4109489321708679,
+      "eval_runtime": 2.3671,
+      "eval_samples_per_second": 84.49,
+      "eval_steps_per_second": 84.49,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6863264323543572,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014722871680795344,
+      "loss": 0.4843,
+      "step": 7005
+    },
+    {
+      "epoch": 1.6875300914780933,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014715812507928622,
+      "loss": 0.4788,
+      "step": 7010
+    },
+    {
+      "epoch": 1.6887337506018296,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00014708751631273783,
+      "loss": 0.5014,
+      "step": 7015
+    },
+    {
+      "epoch": 1.6899374097255657,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014701689057167728,
+      "loss": 0.4386,
+      "step": 7020
+    },
+    {
+      "epoch": 1.6911410688493018,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00014694624791948874,
+      "loss": 0.465,
+      "step": 7025
+    },
+    {
+      "epoch": 1.6923447279730381,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001468755884195717,
+      "loss": 0.4641,
+      "step": 7030
+    },
+    {
+      "epoch": 1.6935483870967742,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00014680491213534065,
+      "loss": 0.497,
+      "step": 7035
+    },
+    {
+      "epoch": 1.6947520462205103,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0001467342191302252,
+      "loss": 0.4729,
+      "step": 7040
+    },
+    {
+      "epoch": 1.6959557053442467,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014666350946766998,
+      "loss": 0.4911,
+      "step": 7045
+    },
+    {
+      "epoch": 1.6971593644679825,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00014659278321113453,
+      "loss": 0.4618,
+      "step": 7050
+    },
+    {
+      "epoch": 1.6983630235917189,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00014652204042409333,
+      "loss": 0.456,
+      "step": 7055
+    },
+    {
+      "epoch": 1.699566682715455,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00014645128117003563,
+      "loss": 0.4735,
+      "step": 7060
+    },
+    {
+      "epoch": 1.700770341839191,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00014638050551246554,
+      "loss": 0.4519,
+      "step": 7065
+    },
+    {
+      "epoch": 1.7019740009629274,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014630971351490177,
+      "loss": 0.4905,
+      "step": 7070
+    },
+    {
+      "epoch": 1.7031776600866635,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00014623890524087787,
+      "loss": 0.5093,
+      "step": 7075
+    },
+    {
+      "epoch": 1.7043813192103996,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00014616808075394185,
+      "loss": 0.4425,
+      "step": 7080
+    },
+    {
+      "epoch": 1.705584978334136,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00014609724011765626,
+      "loss": 0.4641,
+      "step": 7085
+    },
+    {
+      "epoch": 1.7067886374578718,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00014602638339559834,
+      "loss": 0.4622,
+      "step": 7090
+    },
+    {
+      "epoch": 1.7079922965816081,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001459555106513595,
+      "loss": 0.4513,
+      "step": 7095
+    },
+    {
+      "epoch": 1.7091959557053442,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001458846219485457,
+      "loss": 0.4843,
+      "step": 7100
+    },
+    {
+      "epoch": 1.7103996148290803,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00014581371735077726,
+      "loss": 0.4659,
+      "step": 7105
+    },
+    {
+      "epoch": 1.7116032739528166,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00014574279692168858,
+      "loss": 0.4964,
+      "step": 7110
+    },
+    {
+      "epoch": 1.7128069330765527,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00014567186072492842,
+      "loss": 0.4776,
+      "step": 7115
+    },
+    {
+      "epoch": 1.7140105922002888,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00014560090882415963,
+      "loss": 0.4844,
+      "step": 7120
+    },
+    {
+      "epoch": 1.7152142513240252,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00014552994128305918,
+      "loss": 0.4485,
+      "step": 7125
+    },
+    {
+      "epoch": 1.716417910447761,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00014545895816531806,
+      "loss": 0.5152,
+      "step": 7130
+    },
+    {
+      "epoch": 1.7176215695714974,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00014538795953464127,
+      "loss": 0.4929,
+      "step": 7135
+    },
+    {
+      "epoch": 1.7188252286952335,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014531694545474768,
+      "loss": 0.4854,
+      "step": 7140
+    },
+    {
+      "epoch": 1.7200288878189696,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00014524591598937006,
+      "loss": 0.4483,
+      "step": 7145
+    },
+    {
+      "epoch": 1.7212325469427059,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00014517487120225498,
+      "loss": 0.4767,
+      "step": 7150
+    },
+    {
+      "epoch": 1.722436206066442,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00014510381115716278,
+      "loss": 0.4812,
+      "step": 7155
+    },
+    {
+      "epoch": 1.723639865190178,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00014503273591786743,
+      "loss": 0.4936,
+      "step": 7160
+    },
+    {
+      "epoch": 1.7248435243139144,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014496164554815666,
+      "loss": 0.4479,
+      "step": 7165
+    },
+    {
+      "epoch": 1.7260471834376505,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00014489054011183162,
+      "loss": 0.4691,
+      "step": 7170
+    },
+    {
+      "epoch": 1.7272508425613866,
+      "grad_norm": 1.125,
+      "learning_rate": 0.0001448194196727071,
+      "loss": 0.522,
+      "step": 7175
+    },
+    {
+      "epoch": 1.728454501685123,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.0001447482842946113,
+      "loss": 0.4568,
+      "step": 7180
+    },
+    {
+      "epoch": 1.7296581608088588,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00014467713404138588,
+      "loss": 0.5027,
+      "step": 7185
+    },
+    {
+      "epoch": 1.7308618199325951,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00014460596897688578,
+      "loss": 0.5168,
+      "step": 7190
+    },
+    {
+      "epoch": 1.7320654790563312,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014453478916497925,
+      "loss": 0.4555,
+      "step": 7195
+    },
+    {
+      "epoch": 1.7332691381800673,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014446359466954787,
+      "loss": 0.4819,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7344727973038037,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00014439238555448623,
+      "loss": 0.4617,
+      "step": 7205
+    },
+    {
+      "epoch": 1.7356764564275398,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00014432116188370218,
+      "loss": 0.4842,
+      "step": 7210
+    },
+    {
+      "epoch": 1.7368801155512759,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001442499237211166,
+      "loss": 0.4424,
+      "step": 7215
+    },
+    {
+      "epoch": 1.7380837746750122,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.0001441786711306633,
+      "loss": 0.4828,
+      "step": 7220
+    },
+    {
+      "epoch": 1.739287433798748,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00014410740417628912,
+      "loss": 0.5024,
+      "step": 7225
+    },
+    {
+      "epoch": 1.7404910929224844,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001440361229219538,
+      "loss": 0.4626,
+      "step": 7230
+    },
+    {
+      "epoch": 1.7416947520462205,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00014396482743162986,
+      "loss": 0.4882,
+      "step": 7235
+    },
+    {
+      "epoch": 1.7428984111699566,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00014389351776930264,
+      "loss": 0.4854,
+      "step": 7240
+    },
+    {
+      "epoch": 1.744102070293693,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00014382219399897015,
+      "loss": 0.4766,
+      "step": 7245
+    },
+    {
+      "epoch": 1.745305729417429,
+      "grad_norm": 1.375,
+      "learning_rate": 0.0001437508561846431,
+      "loss": 0.4777,
+      "step": 7250
+    },
+    {
+      "epoch": 1.746509388541165,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001436795043903448,
+      "loss": 0.4624,
+      "step": 7255
+    },
+    {
+      "epoch": 1.7477130476649014,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0001436081386801111,
+      "loss": 0.4742,
+      "step": 7260
+    },
+    {
+      "epoch": 1.7489167067886373,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00014353675911799033,
+      "loss": 0.4871,
+      "step": 7265
+    },
+    {
+      "epoch": 1.7501203659123736,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00014346536576804328,
+      "loss": 0.4724,
+      "step": 7270
+    },
+    {
+      "epoch": 1.7513240250361097,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00014339395869434306,
+      "loss": 0.4546,
+      "step": 7275
+    },
+    {
+      "epoch": 1.7525276841598458,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00014332253796097517,
+      "loss": 0.4565,
+      "step": 7280
+    },
+    {
+      "epoch": 1.7537313432835822,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00014325110363203733,
+      "loss": 0.4603,
+      "step": 7285
+    },
+    {
+      "epoch": 1.7549350024073183,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00014317965577163942,
+      "loss": 0.4641,
+      "step": 7290
+    },
+    {
+      "epoch": 1.7561386615310544,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00014310819444390352,
+      "loss": 0.4569,
+      "step": 7295
+    },
+    {
+      "epoch": 1.7573423206547907,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00014303671971296384,
+      "loss": 0.4512,
+      "step": 7300
+    },
+    {
+      "epoch": 1.7585459797785266,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0001429652316429665,
+      "loss": 0.4693,
+      "step": 7305
+    },
+    {
+      "epoch": 1.7597496389022629,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00014289373029806966,
+      "loss": 0.5019,
+      "step": 7310
+    },
+    {
+      "epoch": 1.7609532980259992,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001428222157424434,
+      "loss": 0.4469,
+      "step": 7315
+    },
+    {
+      "epoch": 1.762156957149735,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00014275068804026967,
+      "loss": 0.4632,
+      "step": 7320
+    },
+    {
+      "epoch": 1.7633606162734714,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00014267914725574218,
+      "loss": 0.4967,
+      "step": 7325
+    },
+    {
+      "epoch": 1.7645642753972075,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00014260759345306634,
+      "loss": 0.4606,
+      "step": 7330
+    },
+    {
+      "epoch": 1.7657679345209436,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014253602669645933,
+      "loss": 0.5006,
+      "step": 7335
+    },
+    {
+      "epoch": 1.76697159364468,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00014246444705014996,
+      "loss": 0.4755,
+      "step": 7340
+    },
+    {
+      "epoch": 1.768175252768416,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.0001423928545783785,
+      "loss": 0.4979,
+      "step": 7345
+    },
+    {
+      "epoch": 1.7693789118921521,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00014232124934539686,
+      "loss": 0.4587,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7705825710158885,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00014224963141546833,
+      "loss": 0.4827,
+      "step": 7355
+    },
+    {
+      "epoch": 1.7717862301396243,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001421780008528676,
+      "loss": 0.4649,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7729898892633607,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00014210635772188065,
+      "loss": 0.4997,
+      "step": 7365
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001420347020868049,
+      "loss": 0.4525,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7753972075108329,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001419630340119488,
+      "loss": 0.4857,
+      "step": 7375
+    },
+    {
+      "epoch": 1.7766008666345692,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00014189135356163212,
+      "loss": 0.4892,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7778045257583053,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00014181966080018568,
+      "loss": 0.4653,
+      "step": 7385
+    },
+    {
+      "epoch": 1.7790081848820414,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.0001417479557919512,
+      "loss": 0.4671,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7802118440057777,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0001416762386012817,
+      "loss": 0.4641,
+      "step": 7395
+    },
+    {
+      "epoch": 1.7814155031295136,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00014160450929254082,
+      "loss": 0.4733,
+      "step": 7400
+    },
+    {
+      "epoch": 1.78261916225325,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00014153276793010334,
+      "loss": 0.4886,
+      "step": 7405
+    },
+    {
+      "epoch": 1.783822821376986,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014146101457835464,
+      "loss": 0.4662,
+      "step": 7410
+    },
+    {
+      "epoch": 1.785026480500722,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00014138924930169098,
+      "loss": 0.4709,
+      "step": 7415
+    },
+    {
+      "epoch": 1.7862301396244584,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.0001413174721645193,
+      "loss": 0.4808,
+      "step": 7420
+    },
+    {
+      "epoch": 1.7874337987481945,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00014124568323125724,
+      "loss": 0.4524,
+      "step": 7425
+    },
+    {
+      "epoch": 1.7886374578719306,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0001411738825663329,
+      "loss": 0.4687,
+      "step": 7430
+    },
+    {
+      "epoch": 1.789841116995667,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00014110207023418497,
+      "loss": 0.4937,
+      "step": 7435
+    },
+    {
+      "epoch": 1.7910447761194028,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00014103024629926264,
+      "loss": 0.4673,
+      "step": 7440
+    },
+    {
+      "epoch": 1.7922484352431391,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00014095841082602552,
+      "loss": 0.4704,
+      "step": 7445
+    },
+    {
+      "epoch": 1.7934520943668752,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.0001408865638789435,
+      "loss": 0.4909,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7946557534906113,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00014081470552249685,
+      "loss": 0.4633,
+      "step": 7455
+    },
+    {
+      "epoch": 1.7958594126143477,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.000140742835821176,
+      "loss": 0.452,
+      "step": 7460
+    },
+    {
+      "epoch": 1.7970630717380838,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00014067095483948167,
+      "loss": 0.4742,
+      "step": 7465
+    },
+    {
+      "epoch": 1.7982667308618199,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00014059906264192457,
+      "loss": 0.4509,
+      "step": 7470
+    },
+    {
+      "epoch": 1.7994703899855562,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00014052715929302557,
+      "loss": 0.4924,
+      "step": 7475
+    },
+    {
+      "epoch": 1.8006740491092923,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00014045524485731554,
+      "loss": 0.4548,
+      "step": 7480
+    },
+    {
+      "epoch": 1.8018777082330284,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00014038331939933527,
+      "loss": 0.4899,
+      "step": 7485
+    },
+    {
+      "epoch": 1.8030813673567647,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00014031138298363542,
+      "loss": 0.4613,
+      "step": 7490
+    },
+    {
+      "epoch": 1.8042850264805006,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001402394356747766,
+      "loss": 0.4537,
+      "step": 7495
+    },
+    {
+      "epoch": 1.805488685604237,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00014016747753732903,
+      "loss": 0.4488,
+      "step": 7500
+    },
+    {
+      "epoch": 1.805488685604237,
+      "eval_loss": 0.40007805824279785,
+      "eval_runtime": 2.3802,
+      "eval_samples_per_second": 84.026,
+      "eval_steps_per_second": 84.026,
+      "step": 7500
+    },
+    {
+      "epoch": 1.806692344727973,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00014009550863587278,
+      "loss": 0.4985,
+      "step": 7505
+    },
+    {
+      "epoch": 1.8078960038517091,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00014002352903499753,
+      "loss": 0.4562,
+      "step": 7510
+    },
+    {
+      "epoch": 1.8090996629754454,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00013995153879930254,
+      "loss": 0.4752,
+      "step": 7515
+    },
+    {
+      "epoch": 1.8103033220991815,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013987953799339667,
+      "loss": 0.4579,
+      "step": 7520
+    },
+    {
+      "epoch": 1.8115069812229176,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013980752668189822,
+      "loss": 0.4484,
+      "step": 7525
+    },
+    {
+      "epoch": 1.812710640346654,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00013973550492943495,
+      "loss": 0.4812,
+      "step": 7530
+    },
+    {
+      "epoch": 1.8139142994703898,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013966347280064395,
+      "loss": 0.507,
+      "step": 7535
+    },
+    {
+      "epoch": 1.8151179585941262,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00013959143036017168,
+      "loss": 0.4822,
+      "step": 7540
+    },
+    {
+      "epoch": 1.8163216177178623,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001395193776726738,
+      "loss": 0.4889,
+      "step": 7545
+    },
+    {
+      "epoch": 1.8175252768415984,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0001394473148028152,
+      "loss": 0.4593,
+      "step": 7550
+    },
+    {
+      "epoch": 1.8187289359653347,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001393752418152699,
+      "loss": 0.5091,
+      "step": 7555
+    },
+    {
+      "epoch": 1.8199325950890708,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013930315877472093,
+      "loss": 0.4611,
+      "step": 7560
+    },
+    {
+      "epoch": 1.821136254212807,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013923106574586055,
+      "loss": 0.4749,
+      "step": 7565
+    },
+    {
+      "epoch": 1.8223399133365432,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001391589627933897,
+      "loss": 0.4552,
+      "step": 7570
+    },
+    {
+      "epoch": 1.823543572460279,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013908684998201846,
+      "loss": 0.476,
+      "step": 7575
+    },
+    {
+      "epoch": 1.8247472315840154,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013901472737646564,
+      "loss": 0.468,
+      "step": 7580
+    },
+    {
+      "epoch": 1.8259508907077515,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00013894259504145887,
+      "loss": 0.4631,
+      "step": 7585
+    },
+    {
+      "epoch": 1.8271545498314876,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00013887045304173447,
+      "loss": 0.4736,
+      "step": 7590
+    },
+    {
+      "epoch": 1.828358208955224,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00013879830144203757,
+      "loss": 0.4895,
+      "step": 7595
+    },
+    {
+      "epoch": 1.82956186807896,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001387261403071217,
+      "loss": 0.4471,
+      "step": 7600
+    },
+    {
+      "epoch": 1.8307655272026961,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013865396970174918,
+      "loss": 0.4647,
+      "step": 7605
+    },
+    {
+      "epoch": 1.8319691863264325,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00013858178969069067,
+      "loss": 0.4823,
+      "step": 7610
+    },
+    {
+      "epoch": 1.8331728454501686,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00013850960033872528,
+      "loss": 0.4635,
+      "step": 7615
+    },
+    {
+      "epoch": 1.8343765045739047,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013843740171064057,
+      "loss": 0.4798,
+      "step": 7620
+    },
+    {
+      "epoch": 1.835580163697641,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00013836519387123243,
+      "loss": 0.4532,
+      "step": 7625
+    },
+    {
+      "epoch": 1.8367838228213769,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001382929768853049,
+      "loss": 0.4852,
+      "step": 7630
+    },
+    {
+      "epoch": 1.8379874819451132,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001382207508176704,
+      "loss": 0.4542,
+      "step": 7635
+    },
+    {
+      "epoch": 1.8391911410688493,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013814851573314932,
+      "loss": 0.4939,
+      "step": 7640
+    },
+    {
+      "epoch": 1.8403948001925854,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00013807627169657032,
+      "loss": 0.4772,
+      "step": 7645
+    },
+    {
+      "epoch": 1.8415984593163217,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00013800401877276996,
+      "loss": 0.4939,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8428021184400578,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013793175702659285,
+      "loss": 0.4659,
+      "step": 7655
+    },
+    {
+      "epoch": 1.844005777563794,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013785948652289147,
+      "loss": 0.4671,
+      "step": 7660
+    },
+    {
+      "epoch": 1.8452094366875302,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001377872073265262,
+      "loss": 0.4965,
+      "step": 7665
+    },
+    {
+      "epoch": 1.8464130958112661,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001377149195023652,
+      "loss": 0.4812,
+      "step": 7670
+    },
+    {
+      "epoch": 1.8476167549350024,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00013764262311528435,
+      "loss": 0.4793,
+      "step": 7675
+    },
+    {
+      "epoch": 1.8488204140587385,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00013757031823016726,
+      "loss": 0.4397,
+      "step": 7680
+    },
+    {
+      "epoch": 1.8500240731824746,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00013749800491190517,
+      "loss": 0.4793,
+      "step": 7685
+    },
+    {
+      "epoch": 1.851227732306211,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00013742568322539682,
+      "loss": 0.4907,
+      "step": 7690
+    },
+    {
+      "epoch": 1.852431391429947,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00013735335323554853,
+      "loss": 0.4401,
+      "step": 7695
+    },
+    {
+      "epoch": 1.8536350505536832,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013728101500727407,
+      "loss": 0.4786,
+      "step": 7700
+    },
+    {
+      "epoch": 1.8548387096774195,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00013720866860549456,
+      "loss": 0.4559,
+      "step": 7705
+    },
+    {
+      "epoch": 1.8560423688011554,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00013713631409513846,
+      "loss": 0.4573,
+      "step": 7710
+    },
+    {
+      "epoch": 1.8572460279248917,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013706395154114152,
+      "loss": 0.479,
+      "step": 7715
+    },
+    {
+      "epoch": 1.8584496870486278,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00013699158100844677,
+      "loss": 0.4797,
+      "step": 7720
+    },
+    {
+      "epoch": 1.8596533461723639,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001369192025620043,
+      "loss": 0.4549,
+      "step": 7725
+    },
+    {
+      "epoch": 1.8608570052961002,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00013684681626677136,
+      "loss": 0.4647,
+      "step": 7730
+    },
+    {
+      "epoch": 1.8620606644198363,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013677442218771227,
+      "loss": 0.4867,
+      "step": 7735
+    },
+    {
+      "epoch": 1.8632643235435724,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001367020203897982,
+      "loss": 0.468,
+      "step": 7740
+    },
+    {
+      "epoch": 1.8644679826673087,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0001366296109380074,
+      "loss": 0.4344,
+      "step": 7745
+    },
+    {
+      "epoch": 1.8656716417910446,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013655719389732494,
+      "loss": 0.4853,
+      "step": 7750
+    },
+    {
+      "epoch": 1.866875300914781,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00013648476933274267,
+      "loss": 0.4542,
+      "step": 7755
+    },
+    {
+      "epoch": 1.8680789600385173,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.00013641233730925925,
+      "loss": 0.4683,
+      "step": 7760
+    },
+    {
+      "epoch": 1.8692826191622531,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013633989789187998,
+      "loss": 0.4593,
+      "step": 7765
+    },
+    {
+      "epoch": 1.8704862782859895,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00013626745114561682,
+      "loss": 0.4625,
+      "step": 7770
+    },
+    {
+      "epoch": 1.8716899374097256,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001361949971354883,
+      "loss": 0.4598,
+      "step": 7775
+    },
+    {
+      "epoch": 1.8728935965334617,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001361225359265194,
+      "loss": 0.452,
+      "step": 7780
+    },
+    {
+      "epoch": 1.874097255657198,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001360500675837418,
+      "loss": 0.4659,
+      "step": 7785
+    },
+    {
+      "epoch": 1.875300914780934,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013597759217219327,
+      "loss": 0.4534,
+      "step": 7790
+    },
+    {
+      "epoch": 1.8765045739046702,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001359051097569181,
+      "loss": 0.5062,
+      "step": 7795
+    },
+    {
+      "epoch": 1.8777082330284065,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013583262040296688,
+      "loss": 0.4732,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8789118921521424,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001357601241753964,
+      "loss": 0.4987,
+      "step": 7805
+    },
+    {
+      "epoch": 1.8801155512758787,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001356876211392695,
+      "loss": 0.4723,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8813192103996148,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00013561511135965528,
+      "loss": 0.4462,
+      "step": 7815
+    },
+    {
+      "epoch": 1.882522869523351,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001355425949016289,
+      "loss": 0.4976,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8837265286470872,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00013547007183027136,
+      "loss": 0.4795,
+      "step": 7825
+    },
+    {
+      "epoch": 1.8849301877708233,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00013539754221066977,
+      "loss": 0.4897,
+      "step": 7830
+    },
+    {
+      "epoch": 1.8861338468945594,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.000135325006107917,
+      "loss": 0.4789,
+      "step": 7835
+    },
+    {
+      "epoch": 1.8873375060182958,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001352524635871118,
+      "loss": 0.4801,
+      "step": 7840
+    },
+    {
+      "epoch": 1.8885411651420316,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00013517991471335866,
+      "loss": 0.4723,
+      "step": 7845
+    },
+    {
+      "epoch": 1.889744824265768,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00013510735955176773,
+      "loss": 0.4408,
+      "step": 7850
+    },
+    {
+      "epoch": 1.890948483389504,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00013503479816745487,
+      "loss": 0.4645,
+      "step": 7855
+    },
+    {
+      "epoch": 1.8921521425132402,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001349622306255415,
+      "loss": 0.4407,
+      "step": 7860
+    },
+    {
+      "epoch": 1.8933558016369765,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00013488965699115456,
+      "loss": 0.47,
+      "step": 7865
+    },
+    {
+      "epoch": 1.8945594607607126,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00013481707732942647,
+      "loss": 0.4473,
+      "step": 7870
+    },
+    {
+      "epoch": 1.8957631198844487,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013474449170549501,
+      "loss": 0.4851,
+      "step": 7875
+    },
+    {
+      "epoch": 1.896966779008185,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0001346719001845034,
+      "loss": 0.4167,
+      "step": 7880
+    },
+    {
+      "epoch": 1.8981704381319209,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001345993028316001,
+      "loss": 0.4626,
+      "step": 7885
+    },
+    {
+      "epoch": 1.8993740972556572,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0001345266997119388,
+      "loss": 0.4798,
+      "step": 7890
+    },
+    {
+      "epoch": 1.9005777563793933,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0001344540908906783,
+      "loss": 0.4781,
+      "step": 7895
+    },
+    {
+      "epoch": 1.9017814155031294,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001343814764329827,
+      "loss": 0.4333,
+      "step": 7900
+    },
+    {
+      "epoch": 1.9029850746268657,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013430885640402102,
+      "loss": 0.4794,
+      "step": 7905
+    },
+    {
+      "epoch": 1.9041887337506018,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001342362308689673,
+      "loss": 0.4537,
+      "step": 7910
+    },
+    {
+      "epoch": 1.905392392874338,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00013416359989300045,
+      "loss": 0.4523,
+      "step": 7915
+    },
+    {
+      "epoch": 1.9065960519980742,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001340909635413045,
+      "loss": 0.4772,
+      "step": 7920
+    },
+    {
+      "epoch": 1.9077997111218103,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013401832187906804,
+      "loss": 0.5114,
+      "step": 7925
+    },
+    {
+      "epoch": 1.9090033702455464,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00013394567497148455,
+      "loss": 0.4662,
+      "step": 7930
+    },
+    {
+      "epoch": 1.9102070293692828,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001338730228837522,
+      "loss": 0.4646,
+      "step": 7935
+    },
+    {
+      "epoch": 1.9114106884930187,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013380036568107383,
+      "loss": 0.4487,
+      "step": 7940
+    },
+    {
+      "epoch": 1.912614347616755,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013372770342865686,
+      "loss": 0.4789,
+      "step": 7945
+    },
+    {
+      "epoch": 1.913818006740491,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00013365503619171322,
+      "loss": 0.4429,
+      "step": 7950
+    },
+    {
+      "epoch": 1.9150216658642272,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013358236403545928,
+      "loss": 0.4706,
+      "step": 7955
+    },
+    {
+      "epoch": 1.9162253249879635,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013350968702511594,
+      "loss": 0.4776,
+      "step": 7960
+    },
+    {
+      "epoch": 1.9174289841116996,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00013343700522590833,
+      "loss": 0.4636,
+      "step": 7965
+    },
+    {
+      "epoch": 1.9186326432354357,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.000133364318703066,
+      "loss": 0.4238,
+      "step": 7970
+    },
+    {
+      "epoch": 1.919836302359172,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00013329162752182263,
+      "loss": 0.4501,
+      "step": 7975
+    },
+    {
+      "epoch": 1.921039961482908,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00013321893174741614,
+      "loss": 0.4599,
+      "step": 7980
+    },
+    {
+      "epoch": 1.9222436206066442,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00013314623144508855,
+      "loss": 0.497,
+      "step": 7985
+    },
+    {
+      "epoch": 1.9234472797303803,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00013307352668008593,
+      "loss": 0.4634,
+      "step": 7990
+    },
+    {
+      "epoch": 1.9246509388541164,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00013300081751765842,
+      "loss": 0.4743,
+      "step": 7995
+    },
+    {
+      "epoch": 1.9258545979778527,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00013292810402306003,
+      "loss": 0.4509,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9258545979778527,
+      "eval_loss": 0.39278820157051086,
+      "eval_runtime": 2.3538,
+      "eval_samples_per_second": 84.968,
+      "eval_steps_per_second": 84.968,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9270582571015888,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001328553862615487,
+      "loss": 0.4794,
+      "step": 8005
+    },
+    {
+      "epoch": 1.928261916225325,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00013278266429838623,
+      "loss": 0.4426,
+      "step": 8010
+    },
+    {
+      "epoch": 1.9294655753490613,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001327099381988381,
+      "loss": 0.472,
+      "step": 8015
+    },
+    {
+      "epoch": 1.9306692344727971,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00013263720802817356,
+      "loss": 0.4785,
+      "step": 8020
+    },
+    {
+      "epoch": 1.9318728935965335,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0001325644738516655,
+      "loss": 0.4977,
+      "step": 8025
+    },
+    {
+      "epoch": 1.9330765527202696,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001324917357345905,
+      "loss": 0.5,
+      "step": 8030
+    },
+    {
+      "epoch": 1.9342802118440057,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00013241899374222843,
+      "loss": 0.425,
+      "step": 8035
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00013234624793986293,
+      "loss": 0.4957,
+      "step": 8040
+    },
+    {
+      "epoch": 1.936687530091478,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.0001322734983927809,
+      "loss": 0.4527,
+      "step": 8045
+    },
+    {
+      "epoch": 1.9378911892152142,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0001322007451662726,
+      "loss": 0.4652,
+      "step": 8050
+    },
+    {
+      "epoch": 1.9390948483389505,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013212798832563157,
+      "loss": 0.4574,
+      "step": 8055
+    },
+    {
+      "epoch": 1.9402985074626866,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00013205522793615474,
+      "loss": 0.4662,
+      "step": 8060
+    },
+    {
+      "epoch": 1.9415021665864227,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00013198246406314208,
+      "loss": 0.4573,
+      "step": 8065
+    },
+    {
+      "epoch": 1.942705825710159,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00013190969677189673,
+      "loss": 0.4387,
+      "step": 8070
+    },
+    {
+      "epoch": 1.943909484833895,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00013183692612772494,
+      "loss": 0.4536,
+      "step": 8075
+    },
+    {
+      "epoch": 1.9451131439576312,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.0001317641521959359,
+      "loss": 0.4405,
+      "step": 8080
+    },
+    {
+      "epoch": 1.9463168030813673,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001316913750418418,
+      "loss": 0.4146,
+      "step": 8085
+    },
+    {
+      "epoch": 1.9475204622051034,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00013161859473075763,
+      "loss": 0.4324,
+      "step": 8090
+    },
+    {
+      "epoch": 1.9487241213288398,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00013154581132800137,
+      "loss": 0.4418,
+      "step": 8095
+    },
+    {
+      "epoch": 1.9499277804525759,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00013147302489889366,
+      "loss": 0.473,
+      "step": 8100
+    },
+    {
+      "epoch": 1.951131439576312,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00013140023550875792,
+      "loss": 0.5026,
+      "step": 8105
+    },
+    {
+      "epoch": 1.9523350987000483,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00013132744322292016,
+      "loss": 0.4494,
+      "step": 8110
+    },
+    {
+      "epoch": 1.9535387578237842,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00013125464810670903,
+      "loss": 0.4592,
+      "step": 8115
+    },
+    {
+      "epoch": 1.9547424169475205,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001311818502254557,
+      "loss": 0.438,
+      "step": 8120
+    },
+    {
+      "epoch": 1.9559460760712566,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013110904964449386,
+      "loss": 0.4633,
+      "step": 8125
+    },
+    {
+      "epoch": 1.9571497351949927,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.0001310362464291596,
+      "loss": 0.4591,
+      "step": 8130
+    },
+    {
+      "epoch": 1.958353394318729,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00013096344064479135,
+      "loss": 0.4875,
+      "step": 8135
+    },
+    {
+      "epoch": 1.9595570534424651,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00013089063235672985,
+      "loss": 0.4554,
+      "step": 8140
+    },
+    {
+      "epoch": 1.9607607125662012,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00013081782163031815,
+      "loss": 0.4585,
+      "step": 8145
+    },
+    {
+      "epoch": 1.9619643716899375,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00013074500853090147,
+      "loss": 0.428,
+      "step": 8150
+    },
+    {
+      "epoch": 1.9631680308136734,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00013067219312382701,
+      "loss": 0.4728,
+      "step": 8155
+    },
+    {
+      "epoch": 1.9643716899374097,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001305993754744443,
+      "loss": 0.4498,
+      "step": 8160
+    },
+    {
+      "epoch": 1.9655753490611458,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00013052655564810468,
+      "loss": 0.4499,
+      "step": 8165
+    },
+    {
+      "epoch": 1.966779008184882,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00013045373371016152,
+      "loss": 0.4514,
+      "step": 8170
+    },
+    {
+      "epoch": 1.9679826673086183,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00013038090972597007,
+      "loss": 0.469,
+      "step": 8175
+    },
+    {
+      "epoch": 1.9691863264323544,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00013030808376088743,
+      "loss": 0.49,
+      "step": 8180
+    },
+    {
+      "epoch": 1.9703899855560905,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00013023525588027246,
+      "loss": 0.4705,
+      "step": 8185
+    },
+    {
+      "epoch": 1.9715936446798268,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00013016242614948578,
+      "loss": 0.464,
+      "step": 8190
+    },
+    {
+      "epoch": 1.9727973038035629,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00013008959463388958,
+      "loss": 0.4513,
+      "step": 8195
+    },
+    {
+      "epoch": 1.974000962927299,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00013001676139884775,
+      "loss": 0.4491,
+      "step": 8200
+    },
+    {
+      "epoch": 1.9752046220510353,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001299439265097257,
+      "loss": 0.4901,
+      "step": 8205
+    },
+    {
+      "epoch": 1.9764082811747712,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00012987109003189027,
+      "loss": 0.4848,
+      "step": 8210
+    },
+    {
+      "epoch": 1.9776119402985075,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00012979825203070979,
+      "loss": 0.4355,
+      "step": 8215
+    },
+    {
+      "epoch": 1.9788155994222436,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001297254125715539,
+      "loss": 0.4644,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9800192585459797,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00012965257171979367,
+      "loss": 0.434,
+      "step": 8225
+    },
+    {
+      "epoch": 1.981222917669716,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001295797295408012,
+      "loss": 0.4595,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9824265767934521,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00012950688609994997,
+      "loss": 0.4504,
+      "step": 8235
+    },
+    {
+      "epoch": 1.9836302359171882,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00012943404146261459,
+      "loss": 0.4579,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9848338950409246,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00012936119569417057,
+      "loss": 0.4539,
+      "step": 8245
+    },
+    {
+      "epoch": 1.9860375541646604,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012928834885999466,
+      "loss": 0.4857,
+      "step": 8250
+    },
+    {
+      "epoch": 1.9872412132883968,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.0001292155010254644,
+      "loss": 0.453,
+      "step": 8255
+    },
+    {
+      "epoch": 1.9884448724121329,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00012914265225595831,
+      "loss": 0.4573,
+      "step": 8260
+    },
+    {
+      "epoch": 1.989648531535869,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0001290698026168557,
+      "loss": 0.459,
+      "step": 8265
+    },
+    {
+      "epoch": 1.9908521906596053,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00012899695217353675,
+      "loss": 0.4746,
+      "step": 8270
+    },
+    {
+      "epoch": 1.9920558497833414,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0001289241009913822,
+      "loss": 0.4564,
+      "step": 8275
+    },
+    {
+      "epoch": 1.9932595089070775,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00012885124913577362,
+      "loss": 0.4649,
+      "step": 8280
+    },
+    {
+      "epoch": 1.9944631680308138,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00012877839667209311,
+      "loss": 0.4803,
+      "step": 8285
+    },
+    {
+      "epoch": 1.9956668271545497,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00012870554366572332,
+      "loss": 0.4651,
+      "step": 8290
+    },
+    {
+      "epoch": 1.996870486278286,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00012863269018204733,
+      "loss": 0.4555,
+      "step": 8295
+    },
+    {
+      "epoch": 1.998074145402022,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00012855983628644882,
+      "loss": 0.4275,
+      "step": 8300
+    },
+    {
+      "epoch": 1.9992778045257582,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001284869820443116,
+      "loss": 0.4397,
+      "step": 8305
+    },
+    {
+      "epoch": 1.9995185363505055,
+      "eval_loss": 0.3866819739341736,
+      "eval_runtime": 2.3881,
+      "eval_samples_per_second": 83.748,
+      "eval_steps_per_second": 83.748,
+      "step": 8306
+    },
+    {
+      "epoch": 2.0004814636494945,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00012841412752102,
+      "loss": 0.4412,
+      "step": 8310
+    },
+    {
+      "epoch": 2.0016851227732304,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.00012834127278195853,
+      "loss": 0.3943,
+      "step": 8315
+    },
+    {
+      "epoch": 2.0028887818969667,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00012826841789251179,
+      "loss": 0.4436,
+      "step": 8320
+    },
+    {
+      "epoch": 2.004092441020703,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.0001281955629180647,
+      "loss": 0.4118,
+      "step": 8325
+    },
+    {
+      "epoch": 2.005296100144439,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00012812270792400212,
+      "loss": 0.3988,
+      "step": 8330
+    },
+    {
+      "epoch": 2.0064997592681753,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00012804985297570898,
+      "loss": 0.4198,
+      "step": 8335
+    },
+    {
+      "epoch": 2.0077034183919116,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012797699813857022,
+      "loss": 0.4483,
+      "step": 8340
+    },
+    {
+      "epoch": 2.0089070775156475,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.0001279041434779705,
+      "loss": 0.4045,
+      "step": 8345
+    },
+    {
+      "epoch": 2.0101107366393838,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00012783128905929453,
+      "loss": 0.404,
+      "step": 8350
+    },
+    {
+      "epoch": 2.01131439576312,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00012775843494792672,
+      "loss": 0.4213,
+      "step": 8355
+    },
+    {
+      "epoch": 2.012518054886856,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00012768558120925116,
+      "loss": 0.4015,
+      "step": 8360
+    },
+    {
+      "epoch": 2.0137217140105923,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00012761272790865163,
+      "loss": 0.4103,
+      "step": 8365
+    },
+    {
+      "epoch": 2.014925373134328,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00012753987511151158,
+      "loss": 0.398,
+      "step": 8370
+    },
+    {
+      "epoch": 2.0161290322580645,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012746702288321396,
+      "loss": 0.4256,
+      "step": 8375
+    },
+    {
+      "epoch": 2.017332691381801,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0001273941712891412,
+      "loss": 0.3739,
+      "step": 8380
+    },
+    {
+      "epoch": 2.0185363505055367,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00012732132039467516,
+      "loss": 0.4267,
+      "step": 8385
+    },
+    {
+      "epoch": 2.019740009629273,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00012724847026519706,
+      "loss": 0.406,
+      "step": 8390
+    },
+    {
+      "epoch": 2.0209436687530093,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00012717562096608748,
+      "loss": 0.3906,
+      "step": 8395
+    },
+    {
+      "epoch": 2.0221473278767452,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012710277256272625,
+      "loss": 0.4145,
+      "step": 8400
+    },
+    {
+      "epoch": 2.0233509870004815,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00012702992512049236,
+      "loss": 0.4302,
+      "step": 8405
+    },
+    {
+      "epoch": 2.0245546461242174,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00012695707870476395,
+      "loss": 0.4055,
+      "step": 8410
+    },
+    {
+      "epoch": 2.0257583052479537,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00012688423338091823,
+      "loss": 0.3927,
+      "step": 8415
+    },
+    {
+      "epoch": 2.02696196437169,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00012681138921433153,
+      "loss": 0.3975,
+      "step": 8420
+    },
+    {
+      "epoch": 2.028165623495426,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0001267385462703789,
+      "loss": 0.4427,
+      "step": 8425
+    },
+    {
+      "epoch": 2.0293692826191623,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001266657046144346,
+      "loss": 0.4027,
+      "step": 8430
+    },
+    {
+      "epoch": 2.0305729417428986,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00012659286431187147,
+      "loss": 0.4161,
+      "step": 8435
+    },
+    {
+      "epoch": 2.0317766008666345,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00012652002542806132,
+      "loss": 0.4367,
+      "step": 8440
+    },
+    {
+      "epoch": 2.032980259990371,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00012644718802837457,
+      "loss": 0.4297,
+      "step": 8445
+    },
+    {
+      "epoch": 2.0341839191141067,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00012637435217818037,
+      "loss": 0.4216,
+      "step": 8450
+    },
+    {
+      "epoch": 2.035387578237843,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00012630151794284646,
+      "loss": 0.4061,
+      "step": 8455
+    },
+    {
+      "epoch": 2.0365912373615793,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00012622868538773914,
+      "loss": 0.4072,
+      "step": 8460
+    },
+    {
+      "epoch": 2.037794896485315,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00012615585457822318,
+      "loss": 0.4141,
+      "step": 8465
+    },
+    {
+      "epoch": 2.0389985556090515,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001260830255796618,
+      "loss": 0.4174,
+      "step": 8470
+    },
+    {
+      "epoch": 2.040202214732788,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0001260101984574166,
+      "loss": 0.4019,
+      "step": 8475
+    },
+    {
+      "epoch": 2.0414058738565237,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012593737327684743,
+      "loss": 0.3992,
+      "step": 8480
+    },
+    {
+      "epoch": 2.04260953298026,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00012586455010331254,
+      "loss": 0.4175,
+      "step": 8485
+    },
+    {
+      "epoch": 2.043813192103996,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00012579172900216827,
+      "loss": 0.4268,
+      "step": 8490
+    },
+    {
+      "epoch": 2.0450168512277322,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00012571891003876912,
+      "loss": 0.4015,
+      "step": 8495
+    },
+    {
+      "epoch": 2.0462205103514686,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00012564609327846764,
+      "loss": 0.4222,
+      "step": 8500
+    },
+    {
+      "epoch": 2.0462205103514686,
+      "eval_loss": 0.3873463571071625,
+      "eval_runtime": 2.4132,
+      "eval_samples_per_second": 82.879,
+      "eval_steps_per_second": 82.879,
+      "step": 8500
+    },
+    {
+      "epoch": 2.0474241694752044,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00012557327878661453,
+      "loss": 0.4061,
+      "step": 8505
+    },
+    {
+      "epoch": 2.0486278285989408,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00012550046662855826,
+      "loss": 0.4288,
+      "step": 8510
+    },
+    {
+      "epoch": 2.049831487722677,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00012542765686964532,
+      "loss": 0.413,
+      "step": 8515
+    },
+    {
+      "epoch": 2.051035146846413,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00012535484957522012,
+      "loss": 0.4285,
+      "step": 8520
+    },
+    {
+      "epoch": 2.0522388059701493,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001252820448106247,
+      "loss": 0.3946,
+      "step": 8525
+    },
+    {
+      "epoch": 2.0534424650938856,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.0001252092426411989,
+      "loss": 0.3727,
+      "step": 8530
+    },
+    {
+      "epoch": 2.0546461242176215,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00012513644313228028,
+      "loss": 0.4193,
+      "step": 8535
+    },
+    {
+      "epoch": 2.055849783341358,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00012506364634920391,
+      "loss": 0.4009,
+      "step": 8540
+    },
+    {
+      "epoch": 2.0570534424650937,
+      "grad_norm": 1.1796875,
+      "learning_rate": 0.00012499085235730256,
+      "loss": 0.4007,
+      "step": 8545
+    },
+    {
+      "epoch": 2.05825710158883,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012491806122190627,
+      "loss": 0.378,
+      "step": 8550
+    },
+    {
+      "epoch": 2.0594607607125663,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012484527300834277,
+      "loss": 0.436,
+      "step": 8555
+    },
+    {
+      "epoch": 2.060664419836302,
+      "grad_norm": 1.0,
+      "learning_rate": 0.000124772487781937,
+      "loss": 0.4276,
+      "step": 8560
+    },
+    {
+      "epoch": 2.0618680789600385,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012469970560801126,
+      "loss": 0.4108,
+      "step": 8565
+    },
+    {
+      "epoch": 2.063071738083775,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00012462692655188515,
+      "loss": 0.4106,
+      "step": 8570
+    },
+    {
+      "epoch": 2.0642753972075107,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001245541506788754,
+      "loss": 0.3826,
+      "step": 8575
+    },
+    {
+      "epoch": 2.065479056331247,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00012448137805429595,
+      "loss": 0.423,
+      "step": 8580
+    },
+    {
+      "epoch": 2.066682715454983,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001244086087434578,
+      "loss": 0.4087,
+      "step": 8585
+    },
+    {
+      "epoch": 2.0678863745787193,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00012433584281166897,
+      "loss": 0.4181,
+      "step": 8590
+    },
+    {
+      "epoch": 2.0690900337024556,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00012426308032423443,
+      "loss": 0.3956,
+      "step": 8595
+    },
+    {
+      "epoch": 2.0702936928261915,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001241903213464561,
+      "loss": 0.4212,
+      "step": 8600
+    },
+    {
+      "epoch": 2.071497351949928,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00012411756594363276,
+      "loss": 0.4217,
+      "step": 8605
+    },
+    {
+      "epoch": 2.072701011073664,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012404481418105989,
+      "loss": 0.4157,
+      "step": 8610
+    },
+    {
+      "epoch": 2.0739046701974,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00012397206612402978,
+      "loss": 0.4061,
+      "step": 8615
+    },
+    {
+      "epoch": 2.0751083293211363,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001238993218378314,
+      "loss": 0.3925,
+      "step": 8620
+    },
+    {
+      "epoch": 2.076311988444872,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001238265813877504,
+      "loss": 0.4481,
+      "step": 8625
+    },
+    {
+      "epoch": 2.0775156475686085,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00012375384483906876,
+      "loss": 0.4055,
+      "step": 8630
+    },
+    {
+      "epoch": 2.078719306692345,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012368111225706513,
+      "loss": 0.4255,
+      "step": 8635
+    },
+    {
+      "epoch": 2.0799229658160807,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00012360838370701458,
+      "loss": 0.4198,
+      "step": 8640
+    },
+    {
+      "epoch": 2.081126624939817,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001235356592541886,
+      "loss": 0.391,
+      "step": 8645
+    },
+    {
+      "epoch": 2.0823302840635534,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00012346293896385493,
+      "loss": 0.4005,
+      "step": 8650
+    },
+    {
+      "epoch": 2.0835339431872892,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001233902229012776,
+      "loss": 0.3931,
+      "step": 8655
+    },
+    {
+      "epoch": 2.0847376023110256,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00012331751113171682,
+      "loss": 0.4166,
+      "step": 8660
+    },
+    {
+      "epoch": 2.085941261434762,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012324480372042902,
+      "loss": 0.3932,
+      "step": 8665
+    },
+    {
+      "epoch": 2.0871449205584978,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00012317210073266663,
+      "loss": 0.4082,
+      "step": 8670
+    },
+    {
+      "epoch": 2.088348579682234,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00012309940223367816,
+      "loss": 0.4052,
+      "step": 8675
+    },
+    {
+      "epoch": 2.08955223880597,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0001230267082887081,
+      "loss": 0.4007,
+      "step": 8680
+    },
+    {
+      "epoch": 2.0907558979297063,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0001229540189629968,
+      "loss": 0.3983,
+      "step": 8685
+    },
+    {
+      "epoch": 2.0919595570534426,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00012288133432178055,
+      "loss": 0.4255,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0931632161771785,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00012280865443029136,
+      "loss": 0.4028,
+      "step": 8695
+    },
+    {
+      "epoch": 2.094366875300915,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.000122735979353757,
+      "loss": 0.406,
+      "step": 8700
+    },
+    {
+      "epoch": 2.095570534424651,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00012266330915740095,
+      "loss": 0.3728,
+      "step": 8705
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00012259064390644225,
+      "loss": 0.3881,
+      "step": 8710
+    },
+    {
+      "epoch": 2.0979778526721233,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001225179836660955,
+      "loss": 0.4264,
+      "step": 8715
+    },
+    {
+      "epoch": 2.099181511795859,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00012244532850157094,
+      "loss": 0.4353,
+      "step": 8720
+    },
+    {
+      "epoch": 2.1003851709195955,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00012237267847807406,
+      "loss": 0.4396,
+      "step": 8725
+    },
+    {
+      "epoch": 2.101588830043332,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00012230003366080585,
+      "loss": 0.4089,
+      "step": 8730
+    },
+    {
+      "epoch": 2.1027924891670677,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00012222739411496265,
+      "loss": 0.4109,
+      "step": 8735
+    },
+    {
+      "epoch": 2.103996148290804,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00012215475990573596,
+      "loss": 0.4052,
+      "step": 8740
+    },
+    {
+      "epoch": 2.1051998074145404,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.0001220821310983126,
+      "loss": 0.3983,
+      "step": 8745
+    },
+    {
+      "epoch": 2.1064034665382763,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001220095077578745,
+      "loss": 0.4028,
+      "step": 8750
+    },
+    {
+      "epoch": 2.1076071256620126,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00012193688994959861,
+      "loss": 0.3848,
+      "step": 8755
+    },
+    {
+      "epoch": 2.1088107847857485,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00012186427773865705,
+      "loss": 0.3722,
+      "step": 8760
+    },
+    {
+      "epoch": 2.110014443909485,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00012179167119021688,
+      "loss": 0.4045,
+      "step": 8765
+    },
+    {
+      "epoch": 2.111218103033221,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00012171907036943996,
+      "loss": 0.4107,
+      "step": 8770
+    },
+    {
+      "epoch": 2.112421762156957,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00012164647534148318,
+      "loss": 0.4243,
+      "step": 8775
+    },
+    {
+      "epoch": 2.1136254212806933,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00012157388617149812,
+      "loss": 0.4188,
+      "step": 8780
+    },
+    {
+      "epoch": 2.1148290804044296,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00012150130292463117,
+      "loss": 0.426,
+      "step": 8785
+    },
+    {
+      "epoch": 2.1160327395281655,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00012142872566602326,
+      "loss": 0.4372,
+      "step": 8790
+    },
+    {
+      "epoch": 2.117236398651902,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00012135615446081019,
+      "loss": 0.4109,
+      "step": 8795
+    },
+    {
+      "epoch": 2.118440057775638,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00012128358937412208,
+      "loss": 0.3847,
+      "step": 8800
+    },
+    {
+      "epoch": 2.119643716899374,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00012121103047108372,
+      "loss": 0.4057,
+      "step": 8805
+    },
+    {
+      "epoch": 2.1208473760231104,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00012113847781681432,
+      "loss": 0.4312,
+      "step": 8810
+    },
+    {
+      "epoch": 2.1220510351468462,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00012106593147642746,
+      "loss": 0.4389,
+      "step": 8815
+    },
+    {
+      "epoch": 2.1232546942705826,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00012099339151503099,
+      "loss": 0.4136,
+      "step": 8820
+    },
+    {
+      "epoch": 2.124458353394319,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00012092085799772719,
+      "loss": 0.4158,
+      "step": 8825
+    },
+    {
+      "epoch": 2.1256620125180548,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001208483309896124,
+      "loss": 0.4244,
+      "step": 8830
+    },
+    {
+      "epoch": 2.126865671641791,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00012077581055577722,
+      "loss": 0.3688,
+      "step": 8835
+    },
+    {
+      "epoch": 2.1280693307655274,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00012070329676130628,
+      "loss": 0.4213,
+      "step": 8840
+    },
+    {
+      "epoch": 2.1292729898892633,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00012063078967127836,
+      "loss": 0.4086,
+      "step": 8845
+    },
+    {
+      "epoch": 2.1304766490129996,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00012055828935076606,
+      "loss": 0.3829,
+      "step": 8850
+    },
+    {
+      "epoch": 2.1316803081367355,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00012048579586483603,
+      "loss": 0.4027,
+      "step": 8855
+    },
+    {
+      "epoch": 2.132883967260472,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0001204133092785488,
+      "loss": 0.3939,
+      "step": 8860
+    },
+    {
+      "epoch": 2.134087626384208,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0001203408296569586,
+      "loss": 0.4197,
+      "step": 8865
+    },
+    {
+      "epoch": 2.135291285507944,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00012026835706511346,
+      "loss": 0.4008,
+      "step": 8870
+    },
+    {
+      "epoch": 2.1364949446316803,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00012019589156805517,
+      "loss": 0.3959,
+      "step": 8875
+    },
+    {
+      "epoch": 2.1376986037554166,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.000120123433230819,
+      "loss": 0.3929,
+      "step": 8880
+    },
+    {
+      "epoch": 2.1389022628791525,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00012005098211843397,
+      "loss": 0.4064,
+      "step": 8885
+    },
+    {
+      "epoch": 2.140105922002889,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00011997853829592251,
+      "loss": 0.3928,
+      "step": 8890
+    },
+    {
+      "epoch": 2.1413095811266247,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011990610182830048,
+      "loss": 0.3999,
+      "step": 8895
+    },
+    {
+      "epoch": 2.142513240250361,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011983367278057726,
+      "loss": 0.4234,
+      "step": 8900
+    },
+    {
+      "epoch": 2.1437168993740974,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011976125121775548,
+      "loss": 0.3948,
+      "step": 8905
+    },
+    {
+      "epoch": 2.1449205584978333,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00011968883720483105,
+      "loss": 0.417,
+      "step": 8910
+    },
+    {
+      "epoch": 2.1461242176215696,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00011961643080679311,
+      "loss": 0.4292,
+      "step": 8915
+    },
+    {
+      "epoch": 2.147327876745306,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00011954403208862404,
+      "loss": 0.3968,
+      "step": 8920
+    },
+    {
+      "epoch": 2.1485315358690418,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011947164111529922,
+      "loss": 0.4056,
+      "step": 8925
+    },
+    {
+      "epoch": 2.149735194992778,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011939925795178715,
+      "loss": 0.3954,
+      "step": 8930
+    },
+    {
+      "epoch": 2.1509388541165144,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011932688266304929,
+      "loss": 0.4046,
+      "step": 8935
+    },
+    {
+      "epoch": 2.1521425132402503,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00011925451531404004,
+      "loss": 0.4105,
+      "step": 8940
+    },
+    {
+      "epoch": 2.1533461723639866,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00011918215596970674,
+      "loss": 0.4205,
+      "step": 8945
+    },
+    {
+      "epoch": 2.1545498314877225,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001191098046949894,
+      "loss": 0.4017,
+      "step": 8950
+    },
+    {
+      "epoch": 2.155753490611459,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00011903746155482086,
+      "loss": 0.3812,
+      "step": 8955
+    },
+    {
+      "epoch": 2.156957149735195,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00011896512661412676,
+      "loss": 0.4132,
+      "step": 8960
+    },
+    {
+      "epoch": 2.158160808858931,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011889279993782521,
+      "loss": 0.3805,
+      "step": 8965
+    },
+    {
+      "epoch": 2.1593644679826673,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00011882048159082702,
+      "loss": 0.4274,
+      "step": 8970
+    },
+    {
+      "epoch": 2.1605681271064037,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00011874817163803549,
+      "loss": 0.3998,
+      "step": 8975
+    },
+    {
+      "epoch": 2.1617717862301395,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00011867587014434639,
+      "loss": 0.3893,
+      "step": 8980
+    },
+    {
+      "epoch": 2.162975445353876,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011860357717464788,
+      "loss": 0.386,
+      "step": 8985
+    },
+    {
+      "epoch": 2.1641791044776117,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001185312927938205,
+      "loss": 0.4287,
+      "step": 8990
+    },
+    {
+      "epoch": 2.165382763601348,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00011845901706673703,
+      "loss": 0.3968,
+      "step": 8995
+    },
+    {
+      "epoch": 2.1665864227250844,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00011838675005826253,
+      "loss": 0.415,
+      "step": 9000
+    },
+    {
+      "epoch": 2.1665864227250844,
+      "eval_loss": 0.3820147216320038,
+      "eval_runtime": 2.3573,
+      "eval_samples_per_second": 84.844,
+      "eval_steps_per_second": 84.844,
+      "step": 9000
+    },
+    {
+      "epoch": 2.1677900818488203,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00011831449183325428,
+      "loss": 0.3953,
+      "step": 9005
+    },
+    {
+      "epoch": 2.1689937409725566,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00011824224245656154,
+      "loss": 0.4264,
+      "step": 9010
+    },
+    {
+      "epoch": 2.170197400096293,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001181700019930258,
+      "loss": 0.3985,
+      "step": 9015
+    },
+    {
+      "epoch": 2.171401059220029,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00011809777050748037,
+      "loss": 0.4049,
+      "step": 9020
+    },
+    {
+      "epoch": 2.172604718343765,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011802554806475067,
+      "loss": 0.402,
+      "step": 9025
+    },
+    {
+      "epoch": 2.173808377467501,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00011795333472965388,
+      "loss": 0.425,
+      "step": 9030
+    },
+    {
+      "epoch": 2.1750120365912373,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00011788113056699906,
+      "loss": 0.4106,
+      "step": 9035
+    },
+    {
+      "epoch": 2.1762156957149736,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00011780893564158704,
+      "loss": 0.4154,
+      "step": 9040
+    },
+    {
+      "epoch": 2.1774193548387095,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00011773675001821037,
+      "loss": 0.4097,
+      "step": 9045
+    },
+    {
+      "epoch": 2.178623013962446,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00011766457376165317,
+      "loss": 0.4114,
+      "step": 9050
+    },
+    {
+      "epoch": 2.179826673086182,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011759240693669132,
+      "loss": 0.395,
+      "step": 9055
+    },
+    {
+      "epoch": 2.181030332209918,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00011752024960809206,
+      "loss": 0.388,
+      "step": 9060
+    },
+    {
+      "epoch": 2.1822339913336544,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001174481018406142,
+      "loss": 0.4199,
+      "step": 9065
+    },
+    {
+      "epoch": 2.1834376504573907,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00011737596369900796,
+      "loss": 0.3993,
+      "step": 9070
+    },
+    {
+      "epoch": 2.1846413095811266,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001173038352480149,
+      "loss": 0.4056,
+      "step": 9075
+    },
+    {
+      "epoch": 2.185844968704863,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00011723171655236787,
+      "loss": 0.4201,
+      "step": 9080
+    },
+    {
+      "epoch": 2.1870486278285988,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00011715960767679106,
+      "loss": 0.3978,
+      "step": 9085
+    },
+    {
+      "epoch": 2.188252286952335,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011708750868599965,
+      "loss": 0.41,
+      "step": 9090
+    },
+    {
+      "epoch": 2.1894559460760714,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001170154196447002,
+      "loss": 0.4116,
+      "step": 9095
+    },
+    {
+      "epoch": 2.1906596051998073,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011694334061759016,
+      "loss": 0.4226,
+      "step": 9100
+    },
+    {
+      "epoch": 2.1918632643235436,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011687127166935807,
+      "loss": 0.4147,
+      "step": 9105
+    },
+    {
+      "epoch": 2.19306692344728,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.0001167992128646834,
+      "loss": 0.3978,
+      "step": 9110
+    },
+    {
+      "epoch": 2.194270582571016,
+      "grad_norm": 1.0,
+      "learning_rate": 0.0001167271642682365,
+      "loss": 0.4203,
+      "step": 9115
+    },
+    {
+      "epoch": 2.195474241694752,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011665512594467856,
+      "loss": 0.439,
+      "step": 9120
+    },
+    {
+      "epoch": 2.196677900818488,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00011658309795866159,
+      "loss": 0.4113,
+      "step": 9125
+    },
+    {
+      "epoch": 2.1978815599422243,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00011651108037482833,
+      "loss": 0.4124,
+      "step": 9130
+    },
+    {
+      "epoch": 2.1990852190659607,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001164390732578121,
+      "loss": 0.4198,
+      "step": 9135
+    },
+    {
+      "epoch": 2.2002888781896965,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011636707667223691,
+      "loss": 0.422,
+      "step": 9140
+    },
+    {
+      "epoch": 2.201492537313433,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00011629509068271729,
+      "loss": 0.4166,
+      "step": 9145
+    },
+    {
+      "epoch": 2.202696196437169,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00011622311535385828,
+      "loss": 0.4341,
+      "step": 9150
+    },
+    {
+      "epoch": 2.203899855560905,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00011615115075025525,
+      "loss": 0.4154,
+      "step": 9155
+    },
+    {
+      "epoch": 2.2051035146846414,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00011607919693649414,
+      "loss": 0.4112,
+      "step": 9160
+    },
+    {
+      "epoch": 2.2063071738083773,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.000116007253977151,
+      "loss": 0.3878,
+      "step": 9165
+    },
+    {
+      "epoch": 2.2075108329321136,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0001159353219367923,
+      "loss": 0.3968,
+      "step": 9170
+    },
+    {
+      "epoch": 2.20871449205585,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011586340087997456,
+      "loss": 0.4013,
+      "step": 9175
+    },
+    {
+      "epoch": 2.209918151179586,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00011579149087124459,
+      "loss": 0.3849,
+      "step": 9180
+    },
+    {
+      "epoch": 2.211121810303322,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00011571959197513922,
+      "loss": 0.414,
+      "step": 9185
+    },
+    {
+      "epoch": 2.2123254694270584,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00011564770425618531,
+      "loss": 0.3969,
+      "step": 9190
+    },
+    {
+      "epoch": 2.2135291285507943,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00011557582777889963,
+      "loss": 0.3943,
+      "step": 9195
+    },
+    {
+      "epoch": 2.2147327876745306,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00011550396260778895,
+      "loss": 0.4187,
+      "step": 9200
+    },
+    {
+      "epoch": 2.215936446798267,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00011543210880734988,
+      "loss": 0.392,
+      "step": 9205
+    },
+    {
+      "epoch": 2.217140105922003,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00011536026644206876,
+      "loss": 0.3892,
+      "step": 9210
+    },
+    {
+      "epoch": 2.218343765045739,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011528843557642173,
+      "loss": 0.4663,
+      "step": 9215
+    },
+    {
+      "epoch": 2.219547424169475,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00011521661627487461,
+      "loss": 0.3975,
+      "step": 9220
+    },
+    {
+      "epoch": 2.2207510832932114,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011514480860188279,
+      "loss": 0.4113,
+      "step": 9225
+    },
+    {
+      "epoch": 2.2219547424169477,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011507301262189133,
+      "loss": 0.4319,
+      "step": 9230
+    },
+    {
+      "epoch": 2.2231584015406836,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00011500122839933462,
+      "loss": 0.4144,
+      "step": 9235
+    },
+    {
+      "epoch": 2.22436206066442,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00011492945599863666,
+      "loss": 0.3937,
+      "step": 9240
+    },
+    {
+      "epoch": 2.2255657197881558,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00011485769548421075,
+      "loss": 0.3988,
+      "step": 9245
+    },
+    {
+      "epoch": 2.226769378911892,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00011478594692045958,
+      "loss": 0.3955,
+      "step": 9250
+    },
+    {
+      "epoch": 2.2279730380356284,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011471421037177508,
+      "loss": 0.3914,
+      "step": 9255
+    },
+    {
+      "epoch": 2.2291766971593643,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001146424859025384,
+      "loss": 0.4198,
+      "step": 9260
+    },
+    {
+      "epoch": 2.2303803562831006,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00011457077357711986,
+      "loss": 0.3931,
+      "step": 9265
+    },
+    {
+      "epoch": 2.231584015406837,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0001144990734598789,
+      "loss": 0.3969,
+      "step": 9270
+    },
+    {
+      "epoch": 2.232787674530573,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00011442738561516391,
+      "loss": 0.3841,
+      "step": 9275
+    },
+    {
+      "epoch": 2.233991333654309,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00011435571010731239,
+      "loss": 0.4217,
+      "step": 9280
+    },
+    {
+      "epoch": 2.2351949927780455,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.0001142840470006507,
+      "loss": 0.3936,
+      "step": 9285
+    },
+    {
+      "epoch": 2.2363986519017813,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011421239635949406,
+      "loss": 0.4178,
+      "step": 9290
+    },
+    {
+      "epoch": 2.2376023110255177,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001141407582481466,
+      "loss": 0.4418,
+      "step": 9295
+    },
+    {
+      "epoch": 2.2388059701492535,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00011406913273090103,
+      "loss": 0.4408,
+      "step": 9300
+    },
+    {
+      "epoch": 2.24000962927299,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.0001139975198720389,
+      "loss": 0.3957,
+      "step": 9305
+    },
+    {
+      "epoch": 2.241213288396726,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00011392591973583041,
+      "loss": 0.4021,
+      "step": 9310
+    },
+    {
+      "epoch": 2.242416947520462,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00011385433238653421,
+      "loss": 0.4005,
+      "step": 9315
+    },
+    {
+      "epoch": 2.2436206066441984,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011378275788839758,
+      "loss": 0.4248,
+      "step": 9320
+    },
+    {
+      "epoch": 2.2448242657679347,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00011371119630565626,
+      "loss": 0.4037,
+      "step": 9325
+    },
+    {
+      "epoch": 2.2460279248916706,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00011363964770253433,
+      "loss": 0.4233,
+      "step": 9330
+    },
+    {
+      "epoch": 2.247231584015407,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00011356811214324431,
+      "loss": 0.3855,
+      "step": 9335
+    },
+    {
+      "epoch": 2.2484352431391432,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.000113496589691987,
+      "loss": 0.433,
+      "step": 9340
+    },
+    {
+      "epoch": 2.249638902262879,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011342508041295133,
+      "loss": 0.4197,
+      "step": 9345
+    },
+    {
+      "epoch": 2.2508425613866154,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00011335358437031457,
+      "loss": 0.4289,
+      "step": 9350
+    },
+    {
+      "epoch": 2.2520462205103513,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.000113282101628242,
+      "loss": 0.3911,
+      "step": 9355
+    },
+    {
+      "epoch": 2.2532498796340876,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00011321063225088695,
+      "loss": 0.4378,
+      "step": 9360
+    },
+    {
+      "epoch": 2.254453538757824,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011313917630239082,
+      "loss": 0.3964,
+      "step": 9365
+    },
+    {
+      "epoch": 2.25565719788156,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00011306773384688298,
+      "loss": 0.4,
+      "step": 9370
+    },
+    {
+      "epoch": 2.256860857005296,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0001129963049484806,
+      "loss": 0.3987,
+      "step": 9375
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00011292488967128874,
+      "loss": 0.4173,
+      "step": 9380
+    },
+    {
+      "epoch": 2.2592681752527684,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00011285348807940022,
+      "loss": 0.4203,
+      "step": 9385
+    },
+    {
+      "epoch": 2.2604718343765047,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011278210023689561,
+      "loss": 0.3924,
+      "step": 9390
+    },
+    {
+      "epoch": 2.2616754935002406,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00011271072620784304,
+      "loss": 0.3965,
+      "step": 9395
+    },
+    {
+      "epoch": 2.262879152623977,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00011263936605629842,
+      "loss": 0.3944,
+      "step": 9400
+    },
+    {
+      "epoch": 2.264082811747713,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00011256801984630502,
+      "loss": 0.3896,
+      "step": 9405
+    },
+    {
+      "epoch": 2.265286470871449,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011249668764189369,
+      "loss": 0.3876,
+      "step": 9410
+    },
+    {
+      "epoch": 2.2664901299951854,
+      "grad_norm": 1.1328125,
+      "learning_rate": 0.00011242536950708272,
+      "loss": 0.4062,
+      "step": 9415
+    },
+    {
+      "epoch": 2.2676937891189217,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00011235406550587776,
+      "loss": 0.4153,
+      "step": 9420
+    },
+    {
+      "epoch": 2.2688974482426576,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011228277570227175,
+      "loss": 0.4118,
+      "step": 9425
+    },
+    {
+      "epoch": 2.270101107366394,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00011221150016024493,
+      "loss": 0.3852,
+      "step": 9430
+    },
+    {
+      "epoch": 2.27130476649013,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00011214023894376464,
+      "loss": 0.4141,
+      "step": 9435
+    },
+    {
+      "epoch": 2.272508425613866,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00011206899211678556,
+      "loss": 0.4061,
+      "step": 9440
+    },
+    {
+      "epoch": 2.2737120847376024,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00011199775974324922,
+      "loss": 0.3992,
+      "step": 9445
+    },
+    {
+      "epoch": 2.2749157438613383,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00011192654188708437,
+      "loss": 0.4264,
+      "step": 9450
+    },
+    {
+      "epoch": 2.2761194029850746,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00011185533861220663,
+      "loss": 0.4014,
+      "step": 9455
+    },
+    {
+      "epoch": 2.277323062108811,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00011178414998251854,
+      "loss": 0.3882,
+      "step": 9460
+    },
+    {
+      "epoch": 2.278526721232547,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00011171297606190955,
+      "loss": 0.3994,
+      "step": 9465
+    },
+    {
+      "epoch": 2.279730380356283,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011164181691425591,
+      "loss": 0.4129,
+      "step": 9470
+    },
+    {
+      "epoch": 2.2809340394800195,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00011157067260342047,
+      "loss": 0.4218,
+      "step": 9475
+    },
+    {
+      "epoch": 2.2821376986037554,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00011149954319325295,
+      "loss": 0.4282,
+      "step": 9480
+    },
+    {
+      "epoch": 2.2833413577274917,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001114284287475896,
+      "loss": 0.4001,
+      "step": 9485
+    },
+    {
+      "epoch": 2.2845450168512276,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00011135732933025322,
+      "loss": 0.4099,
+      "step": 9490
+    },
+    {
+      "epoch": 2.285748675974964,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00011128624500505322,
+      "loss": 0.4388,
+      "step": 9495
+    },
+    {
+      "epoch": 2.2869523350987,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00011121517583578538,
+      "loss": 0.425,
+      "step": 9500
+    },
+    {
+      "epoch": 2.2869523350987,
+      "eval_loss": 0.3737788498401642,
+      "eval_runtime": 2.3555,
+      "eval_samples_per_second": 84.907,
+      "eval_steps_per_second": 84.907,
+      "step": 9500
+    },
+    {
+      "epoch": 2.288155994222436,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00011114412188623192,
+      "loss": 0.3965,
+      "step": 9505
+    },
+    {
+      "epoch": 2.2893596533461724,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00011107308322016139,
+      "loss": 0.3889,
+      "step": 9510
+    },
+    {
+      "epoch": 2.2905633124699083,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00011100205990132853,
+      "loss": 0.3988,
+      "step": 9515
+    },
+    {
+      "epoch": 2.2917669715936446,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001109310519934745,
+      "loss": 0.4219,
+      "step": 9520
+    },
+    {
+      "epoch": 2.292970630717381,
+      "grad_norm": 0.9140625,
+      "learning_rate": 0.0001108600595603265,
+      "loss": 0.3981,
+      "step": 9525
+    },
+    {
+      "epoch": 2.294174289841117,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00011078908266559781,
+      "loss": 0.4256,
+      "step": 9530
+    },
+    {
+      "epoch": 2.295377948964853,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00011071812137298792,
+      "loss": 0.4239,
+      "step": 9535
+    },
+    {
+      "epoch": 2.2965816080885895,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00011064717574618213,
+      "loss": 0.4134,
+      "step": 9540
+    },
+    {
+      "epoch": 2.2977852672123253,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00011057624584885178,
+      "loss": 0.4064,
+      "step": 9545
+    },
+    {
+      "epoch": 2.2989889263360617,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00011050533174465415,
+      "loss": 0.4028,
+      "step": 9550
+    },
+    {
+      "epoch": 2.300192585459798,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00011043443349723222,
+      "loss": 0.3983,
+      "step": 9555
+    },
+    {
+      "epoch": 2.301396244583534,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00011036355117021479,
+      "loss": 0.4169,
+      "step": 9560
+    },
+    {
+      "epoch": 2.30259990370727,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00011029268482721641,
+      "loss": 0.3789,
+      "step": 9565
+    },
+    {
+      "epoch": 2.303803562831006,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00011022183453183724,
+      "loss": 0.4294,
+      "step": 9570
+    },
+    {
+      "epoch": 2.3050072219547424,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00011015100034766303,
+      "loss": 0.3813,
+      "step": 9575
+    },
+    {
+      "epoch": 2.3062108810784787,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00011008018233826515,
+      "loss": 0.4321,
+      "step": 9580
+    },
+    {
+      "epoch": 2.3074145402022146,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00011000938056720036,
+      "loss": 0.3977,
+      "step": 9585
+    },
+    {
+      "epoch": 2.308618199325951,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010993859509801087,
+      "loss": 0.4167,
+      "step": 9590
+    },
+    {
+      "epoch": 2.3098218584496872,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010986782599422432,
+      "loss": 0.4403,
+      "step": 9595
+    },
+    {
+      "epoch": 2.311025517573423,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00010979707331935356,
+      "loss": 0.4135,
+      "step": 9600
+    },
+    {
+      "epoch": 2.3122291766971594,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00010972633713689674,
+      "loss": 0.4014,
+      "step": 9605
+    },
+    {
+      "epoch": 2.3134328358208958,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00010965561751033732,
+      "loss": 0.3802,
+      "step": 9610
+    },
+    {
+      "epoch": 2.3146364949446316,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010958491450314369,
+      "loss": 0.4037,
+      "step": 9615
+    },
+    {
+      "epoch": 2.315840154068368,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00010951422817876948,
+      "loss": 0.4409,
+      "step": 9620
+    },
+    {
+      "epoch": 2.317043813192104,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00010944355860065331,
+      "loss": 0.398,
+      "step": 9625
+    },
+    {
+      "epoch": 2.31824747231584,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00010937290583221877,
+      "loss": 0.4227,
+      "step": 9630
+    },
+    {
+      "epoch": 2.3194511314395765,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010930226993687437,
+      "loss": 0.3891,
+      "step": 9635
+    },
+    {
+      "epoch": 2.3206547905633124,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010923165097801341,
+      "loss": 0.3923,
+      "step": 9640
+    },
+    {
+      "epoch": 2.3218584496870487,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00010916104901901406,
+      "loss": 0.3993,
+      "step": 9645
+    },
+    {
+      "epoch": 2.3230621088107846,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00010909046412323926,
+      "loss": 0.4074,
+      "step": 9650
+    },
+    {
+      "epoch": 2.324265767934521,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010901989635403658,
+      "loss": 0.4037,
+      "step": 9655
+    },
+    {
+      "epoch": 2.325469427058257,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001089493457747382,
+      "loss": 0.3959,
+      "step": 9660
+    },
+    {
+      "epoch": 2.326673086181993,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00010887881244866092,
+      "loss": 0.3863,
+      "step": 9665
+    },
+    {
+      "epoch": 2.3278767453057294,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00010880829643910606,
+      "loss": 0.4277,
+      "step": 9670
+    },
+    {
+      "epoch": 2.3290804044294657,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010873779780935943,
+      "loss": 0.4081,
+      "step": 9675
+    },
+    {
+      "epoch": 2.3302840635532016,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010866731662269109,
+      "loss": 0.4039,
+      "step": 9680
+    },
+    {
+      "epoch": 2.331487722676938,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001085968529423556,
+      "loss": 0.4221,
+      "step": 9685
+    },
+    {
+      "epoch": 2.3326913818006743,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00010852640683159177,
+      "loss": 0.3882,
+      "step": 9690
+    },
+    {
+      "epoch": 2.33389504092441,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00010845597835362258,
+      "loss": 0.4002,
+      "step": 9695
+    },
+    {
+      "epoch": 2.3350987000481465,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00010838556757165528,
+      "loss": 0.4113,
+      "step": 9700
+    },
+    {
+      "epoch": 2.3363023591718823,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00010831517454888122,
+      "loss": 0.3922,
+      "step": 9705
+    },
+    {
+      "epoch": 2.3375060182956187,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001082447993484757,
+      "loss": 0.402,
+      "step": 9710
+    },
+    {
+      "epoch": 2.338709677419355,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00010817444203359823,
+      "loss": 0.3954,
+      "step": 9715
+    },
+    {
+      "epoch": 2.339913336543091,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00010810410266739204,
+      "loss": 0.3903,
+      "step": 9720
+    },
+    {
+      "epoch": 2.341116995666827,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00010803378131298438,
+      "loss": 0.4084,
+      "step": 9725
+    },
+    {
+      "epoch": 2.3423206547905635,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010796347803348634,
+      "loss": 0.4101,
+      "step": 9730
+    },
+    {
+      "epoch": 2.3435243139142994,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00010789319289199275,
+      "loss": 0.4125,
+      "step": 9735
+    },
+    {
+      "epoch": 2.3447279730380357,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.0001078229259515822,
+      "loss": 0.4058,
+      "step": 9740
+    },
+    {
+      "epoch": 2.345931632161772,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00010775267727531688,
+      "loss": 0.3906,
+      "step": 9745
+    },
+    {
+      "epoch": 2.347135291285508,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010768244692624267,
+      "loss": 0.4411,
+      "step": 9750
+    },
+    {
+      "epoch": 2.3483389504092442,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00010761223496738896,
+      "loss": 0.3768,
+      "step": 9755
+    },
+    {
+      "epoch": 2.34954260953298,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.0001075420414617686,
+      "loss": 0.3837,
+      "step": 9760
+    },
+    {
+      "epoch": 2.3507462686567164,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00010747186647237793,
+      "loss": 0.4019,
+      "step": 9765
+    },
+    {
+      "epoch": 2.3519499277804528,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00010740171006219663,
+      "loss": 0.3842,
+      "step": 9770
+    },
+    {
+      "epoch": 2.3531535869041886,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00010733157229418781,
+      "loss": 0.4019,
+      "step": 9775
+    },
+    {
+      "epoch": 2.354357246027925,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010726145323129768,
+      "loss": 0.3976,
+      "step": 9780
+    },
+    {
+      "epoch": 2.355560905151661,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00010719135293645583,
+      "loss": 0.4061,
+      "step": 9785
+    },
+    {
+      "epoch": 2.356764564275397,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00010712127147257491,
+      "loss": 0.4393,
+      "step": 9790
+    },
+    {
+      "epoch": 2.3579682233991335,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00010705120890255071,
+      "loss": 0.3997,
+      "step": 9795
+    },
+    {
+      "epoch": 2.3591718825228694,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00010698116528926197,
+      "loss": 0.3949,
+      "step": 9800
+    },
+    {
+      "epoch": 2.3603755416466057,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00010691114069557054,
+      "loss": 0.3888,
+      "step": 9805
+    },
+    {
+      "epoch": 2.361579200770342,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010684113518432115,
+      "loss": 0.3958,
+      "step": 9810
+    },
+    {
+      "epoch": 2.362782859894078,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00010677114881834144,
+      "loss": 0.4197,
+      "step": 9815
+    },
+    {
+      "epoch": 2.363986519017814,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0001067011816604418,
+      "loss": 0.4055,
+      "step": 9820
+    },
+    {
+      "epoch": 2.3651901781415505,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.0001066312337734154,
+      "loss": 0.4217,
+      "step": 9825
+    },
+    {
+      "epoch": 2.3663938372652864,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00010656130522003814,
+      "loss": 0.3787,
+      "step": 9830
+    },
+    {
+      "epoch": 2.3675974963890227,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010649139606306862,
+      "loss": 0.4474,
+      "step": 9835
+    },
+    {
+      "epoch": 2.3688011555127586,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00010642150636524789,
+      "loss": 0.4077,
+      "step": 9840
+    },
+    {
+      "epoch": 2.370004814636495,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00010635163618929966,
+      "loss": 0.3959,
+      "step": 9845
+    },
+    {
+      "epoch": 2.3712084737602313,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00010628178559793005,
+      "loss": 0.4345,
+      "step": 9850
+    },
+    {
+      "epoch": 2.372412132883967,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00010621195465382762,
+      "loss": 0.3856,
+      "step": 9855
+    },
+    {
+      "epoch": 2.3736157920077035,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.0001061421434196634,
+      "loss": 0.3847,
+      "step": 9860
+    },
+    {
+      "epoch": 2.3748194511314393,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00010607235195809052,
+      "loss": 0.3991,
+      "step": 9865
+    },
+    {
+      "epoch": 2.3760231102551757,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00010600258033174455,
+      "loss": 0.4006,
+      "step": 9870
+    },
+    {
+      "epoch": 2.377226769378912,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00010593282860324323,
+      "loss": 0.4036,
+      "step": 9875
+    },
+    {
+      "epoch": 2.3784304285026483,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010586309683518631,
+      "loss": 0.4463,
+      "step": 9880
+    },
+    {
+      "epoch": 2.379634087626384,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010579338509015576,
+      "loss": 0.4012,
+      "step": 9885
+    },
+    {
+      "epoch": 2.3808377467501205,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010572369343071558,
+      "loss": 0.3909,
+      "step": 9890
+    },
+    {
+      "epoch": 2.3820414058738564,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00010565402191941163,
+      "loss": 0.4349,
+      "step": 9895
+    },
+    {
+      "epoch": 2.3832450649975927,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00010558437061877183,
+      "loss": 0.391,
+      "step": 9900
+    },
+    {
+      "epoch": 2.384448724121329,
+      "grad_norm": 1.0078125,
+      "learning_rate": 0.00010551473959130588,
+      "loss": 0.3932,
+      "step": 9905
+    },
+    {
+      "epoch": 2.385652383245065,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010544512889950525,
+      "loss": 0.3994,
+      "step": 9910
+    },
+    {
+      "epoch": 2.3868560423688012,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.0001053755386058433,
+      "loss": 0.4208,
+      "step": 9915
+    },
+    {
+      "epoch": 2.388059701492537,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010530596877277495,
+      "loss": 0.3968,
+      "step": 9920
+    },
+    {
+      "epoch": 2.3892633606162734,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010523641946273675,
+      "loss": 0.3979,
+      "step": 9925
+    },
+    {
+      "epoch": 2.3904670197400097,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010516689073814694,
+      "loss": 0.3819,
+      "step": 9930
+    },
+    {
+      "epoch": 2.3916706788637456,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.0001050973826614052,
+      "loss": 0.3867,
+      "step": 9935
+    },
+    {
+      "epoch": 2.392874337987482,
+      "grad_norm": 1.1640625,
+      "learning_rate": 0.00010502789529489273,
+      "loss": 0.3827,
+      "step": 9940
+    },
+    {
+      "epoch": 2.3940779971112183,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010495842870097208,
+      "loss": 0.3925,
+      "step": 9945
+    },
+    {
+      "epoch": 2.395281656234954,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0001048889829419872,
+      "loss": 0.4186,
+      "step": 9950
+    },
+    {
+      "epoch": 2.3964853153586905,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00010481955808026344,
+      "loss": 0.3871,
+      "step": 9955
+    },
+    {
+      "epoch": 2.397688974482427,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00010475015417810716,
+      "loss": 0.421,
+      "step": 9960
+    },
+    {
+      "epoch": 2.3988926336061627,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00010468077129780608,
+      "loss": 0.4179,
+      "step": 9965
+    },
+    {
+      "epoch": 2.400096292729899,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00010461140950162902,
+      "loss": 0.4046,
+      "step": 9970
+    },
+    {
+      "epoch": 2.401299951853635,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010454206885182592,
+      "loss": 0.4405,
+      "step": 9975
+    },
+    {
+      "epoch": 2.402503610977371,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00010447274941062763,
+      "loss": 0.3921,
+      "step": 9980
+    },
+    {
+      "epoch": 2.4037072701011075,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010440345124024607,
+      "loss": 0.4091,
+      "step": 9985
+    },
+    {
+      "epoch": 2.4049109292248434,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.000104334174402874,
+      "loss": 0.3719,
+      "step": 9990
+    },
+    {
+      "epoch": 2.4061145883485797,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010426491896068513,
+      "loss": 0.3683,
+      "step": 9995
+    },
+    {
+      "epoch": 2.4073182474723156,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00010419568497583383,
+      "loss": 0.4107,
+      "step": 10000
+    },
+    {
+      "epoch": 2.4073182474723156,
+      "eval_loss": 0.36819595098495483,
+      "eval_runtime": 2.3514,
+      "eval_samples_per_second": 85.057,
+      "eval_steps_per_second": 85.057,
+      "step": 10000
+    },
+    {
+      "epoch": 2.408521906596052,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010412647251045531,
+      "loss": 0.3848,
+      "step": 10005
+    },
+    {
+      "epoch": 2.4097255657197882,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00010405728162666547,
+      "loss": 0.3846,
+      "step": 10010
+    },
+    {
+      "epoch": 2.410929224843524,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00010398811238656079,
+      "loss": 0.3957,
+      "step": 10015
+    },
+    {
+      "epoch": 2.4121328839672604,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00010391896485221835,
+      "loss": 0.3971,
+      "step": 10020
+    },
+    {
+      "epoch": 2.4133365430909968,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00010384983908569576,
+      "loss": 0.3867,
+      "step": 10025
+    },
+    {
+      "epoch": 2.4145402022147326,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00010378073514903111,
+      "loss": 0.4201,
+      "step": 10030
+    },
+    {
+      "epoch": 2.415743861338469,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010371165310424286,
+      "loss": 0.3967,
+      "step": 10035
+    },
+    {
+      "epoch": 2.4169475204622053,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010364259301332979,
+      "loss": 0.4013,
+      "step": 10040
+    },
+    {
+      "epoch": 2.418151179585941,
+      "grad_norm": 1.140625,
+      "learning_rate": 0.00010357355493827108,
+      "loss": 0.4278,
+      "step": 10045
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.00010350453894102603,
+      "loss": 0.4207,
+      "step": 10050
+    },
+    {
+      "epoch": 2.4205584978334134,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010343554508353427,
+      "loss": 0.3962,
+      "step": 10055
+    },
+    {
+      "epoch": 2.4217621569571497,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00010336657342771544,
+      "loss": 0.4175,
+      "step": 10060
+    },
+    {
+      "epoch": 2.422965816080886,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010329762403546929,
+      "loss": 0.3889,
+      "step": 10065
+    },
+    {
+      "epoch": 2.424169475204622,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00010322869696867563,
+      "loss": 0.3866,
+      "step": 10070
+    },
+    {
+      "epoch": 2.425373134328358,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00010315979228919418,
+      "loss": 0.3924,
+      "step": 10075
+    },
+    {
+      "epoch": 2.4265767934520945,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010309091005886452,
+      "loss": 0.4005,
+      "step": 10080
+    },
+    {
+      "epoch": 2.4277804525758304,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010302205033950628,
+      "loss": 0.4052,
+      "step": 10085
+    },
+    {
+      "epoch": 2.4289841116995667,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00010295321319291862,
+      "loss": 0.3789,
+      "step": 10090
+    },
+    {
+      "epoch": 2.430187770823303,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00010288439868088067,
+      "loss": 0.3771,
+      "step": 10095
+    },
+    {
+      "epoch": 2.431391429947039,
+      "grad_norm": 1.0,
+      "learning_rate": 0.00010281560686515113,
+      "loss": 0.4043,
+      "step": 10100
+    },
+    {
+      "epoch": 2.4325950890707753,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00010274683780746836,
+      "loss": 0.4279,
+      "step": 10105
+    },
+    {
+      "epoch": 2.433798748194511,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00010267809156955029,
+      "loss": 0.3857,
+      "step": 10110
+    },
+    {
+      "epoch": 2.4350024073182475,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001026093682130944,
+      "loss": 0.4142,
+      "step": 10115
+    },
+    {
+      "epoch": 2.436206066441984,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001025406677997775,
+      "loss": 0.4047,
+      "step": 10120
+    },
+    {
+      "epoch": 2.4374097255657197,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010247199039125601,
+      "loss": 0.3986,
+      "step": 10125
+    },
+    {
+      "epoch": 2.438613384689456,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00010240333604916559,
+      "loss": 0.4001,
+      "step": 10130
+    },
+    {
+      "epoch": 2.439817043813192,
+      "grad_norm": 1.15625,
+      "learning_rate": 0.0001023347048351212,
+      "loss": 0.401,
+      "step": 10135
+    },
+    {
+      "epoch": 2.441020702936928,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010226609681071709,
+      "loss": 0.3961,
+      "step": 10140
+    },
+    {
+      "epoch": 2.4422243620606645,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010219751203752663,
+      "loss": 0.4151,
+      "step": 10145
+    },
+    {
+      "epoch": 2.4434280211844004,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00010212895057710238,
+      "loss": 0.4057,
+      "step": 10150
+    },
+    {
+      "epoch": 2.4446316803081367,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010206041249097601,
+      "loss": 0.3807,
+      "step": 10155
+    },
+    {
+      "epoch": 2.445835339431873,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010199189784065808,
+      "loss": 0.4226,
+      "step": 10160
+    },
+    {
+      "epoch": 2.447038998555609,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00010192340668763823,
+      "loss": 0.3955,
+      "step": 10165
+    },
+    {
+      "epoch": 2.4482426576793452,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00010185493909338498,
+      "loss": 0.4077,
+      "step": 10170
+    },
+    {
+      "epoch": 2.4494463168030816,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.0001017864951193457,
+      "loss": 0.41,
+      "step": 10175
+    },
+    {
+      "epoch": 2.4506499759268174,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010171807482694658,
+      "loss": 0.4075,
+      "step": 10180
+    },
+    {
+      "epoch": 2.4518536350505538,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.00010164967827759255,
+      "loss": 0.3915,
+      "step": 10185
+    },
+    {
+      "epoch": 2.4530572941742896,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00010158130553266722,
+      "loss": 0.3908,
+      "step": 10190
+    },
+    {
+      "epoch": 2.454260953298026,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00010151295665353287,
+      "loss": 0.4049,
+      "step": 10195
+    },
+    {
+      "epoch": 2.4554646124217623,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.0001014446317015303,
+      "loss": 0.3893,
+      "step": 10200
+    },
+    {
+      "epoch": 2.456668271545498,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010137633073797885,
+      "loss": 0.3975,
+      "step": 10205
+    },
+    {
+      "epoch": 2.4578719306692345,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00010130805382417638,
+      "loss": 0.4042,
+      "step": 10210
+    },
+    {
+      "epoch": 2.459075589792971,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.00010123980102139918,
+      "loss": 0.4282,
+      "step": 10215
+    },
+    {
+      "epoch": 2.4602792489167067,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00010117157239090176,
+      "loss": 0.4187,
+      "step": 10220
+    },
+    {
+      "epoch": 2.461482908040443,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010110336799391714,
+      "loss": 0.397,
+      "step": 10225
+    },
+    {
+      "epoch": 2.4626865671641793,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00010103518789165641,
+      "loss": 0.4012,
+      "step": 10230
+    },
+    {
+      "epoch": 2.463890226287915,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010096703214530897,
+      "loss": 0.4001,
+      "step": 10235
+    },
+    {
+      "epoch": 2.4650938854116515,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00010089890081604235,
+      "loss": 0.3821,
+      "step": 10240
+    },
+    {
+      "epoch": 2.4662975445353874,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00010083079396500208,
+      "loss": 0.4222,
+      "step": 10245
+    },
+    {
+      "epoch": 2.4675012036591237,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010076271165331181,
+      "loss": 0.3884,
+      "step": 10250
+    },
+    {
+      "epoch": 2.46870486278286,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010069465394207315,
+      "loss": 0.4078,
+      "step": 10255
+    },
+    {
+      "epoch": 2.469908521906596,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00010062662089236559,
+      "loss": 0.4043,
+      "step": 10260
+    },
+    {
+      "epoch": 2.4711121810303323,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00010055861256524654,
+      "loss": 0.4076,
+      "step": 10265
+    },
+    {
+      "epoch": 2.472315840154068,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00010049062902175123,
+      "loss": 0.404,
+      "step": 10270
+    },
+    {
+      "epoch": 2.4735194992778045,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00010042267032289255,
+      "loss": 0.4178,
+      "step": 10275
+    },
+    {
+      "epoch": 2.474723158401541,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00010035473652966124,
+      "loss": 0.3973,
+      "step": 10280
+    },
+    {
+      "epoch": 2.4759268175252767,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00010028682770302555,
+      "loss": 0.3977,
+      "step": 10285
+    },
+    {
+      "epoch": 2.477130476649013,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00010021894390393138,
+      "loss": 0.39,
+      "step": 10290
+    },
+    {
+      "epoch": 2.4783341357727493,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.0001001510851933022,
+      "loss": 0.3888,
+      "step": 10295
+    },
+    {
+      "epoch": 2.479537794896485,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.00010008325163203892,
+      "loss": 0.394,
+      "step": 10300
+    },
+    {
+      "epoch": 2.4807414540202215,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00010001544328101988,
+      "loss": 0.4204,
+      "step": 10305
+    },
+    {
+      "epoch": 2.481945113143958,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.994766020110084e-05,
+      "loss": 0.376,
+      "step": 10310
+    },
+    {
+      "epoch": 2.4831487722676937,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.987990245311485e-05,
+      "loss": 0.4057,
+      "step": 10315
+    },
+    {
+      "epoch": 2.48435243139143,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.981217009787222e-05,
+      "loss": 0.3924,
+      "step": 10320
+    },
+    {
+      "epoch": 2.485556090515166,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.974446319616045e-05,
+      "loss": 0.4139,
+      "step": 10325
+    },
+    {
+      "epoch": 2.4867597496389022,
+      "grad_norm": 0.9296875,
+      "learning_rate": 9.967678180874422e-05,
+      "loss": 0.3848,
+      "step": 10330
+    },
+    {
+      "epoch": 2.4879634087626386,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.960912599636532e-05,
+      "loss": 0.4026,
+      "step": 10335
+    },
+    {
+      "epoch": 2.4891670678863744,
+      "grad_norm": 1.21875,
+      "learning_rate": 9.954149581974263e-05,
+      "loss": 0.4009,
+      "step": 10340
+    },
+    {
+      "epoch": 2.4903707270101108,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.94738913395719e-05,
+      "loss": 0.4379,
+      "step": 10345
+    },
+    {
+      "epoch": 2.491574386133847,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.940631261652592e-05,
+      "loss": 0.3752,
+      "step": 10350
+    },
+    {
+      "epoch": 2.492778045257583,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.933875971125432e-05,
+      "loss": 0.3975,
+      "step": 10355
+    },
+    {
+      "epoch": 2.4939817043813193,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.927123268438365e-05,
+      "loss": 0.4052,
+      "step": 10360
+    },
+    {
+      "epoch": 2.4951853635050556,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.920373159651707e-05,
+      "loss": 0.4254,
+      "step": 10365
+    },
+    {
+      "epoch": 2.4963890226287915,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.913625650823458e-05,
+      "loss": 0.421,
+      "step": 10370
+    },
+    {
+      "epoch": 2.497592681752528,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.906880748009281e-05,
+      "loss": 0.4111,
+      "step": 10375
+    },
+    {
+      "epoch": 2.4987963408762637,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.900138457262503e-05,
+      "loss": 0.3946,
+      "step": 10380
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.893398784634103e-05,
+      "loss": 0.4282,
+      "step": 10385
+    },
+    {
+      "epoch": 2.5012036591237363,
+      "grad_norm": 1.0,
+      "learning_rate": 9.886661736172711e-05,
+      "loss": 0.3733,
+      "step": 10390
+    },
+    {
+      "epoch": 2.502407318247472,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.879927317924606e-05,
+      "loss": 0.4407,
+      "step": 10395
+    },
+    {
+      "epoch": 2.5036109773712085,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.873195535933702e-05,
+      "loss": 0.3855,
+      "step": 10400
+    },
+    {
+      "epoch": 2.5048146364949444,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.866466396241545e-05,
+      "loss": 0.3845,
+      "step": 10405
+    },
+    {
+      "epoch": 2.5060182956186807,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.859739904887317e-05,
+      "loss": 0.3955,
+      "step": 10410
+    },
+    {
+      "epoch": 2.507221954742417,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.853016067907816e-05,
+      "loss": 0.3971,
+      "step": 10415
+    },
+    {
+      "epoch": 2.5084256138661534,
+      "grad_norm": 1.0,
+      "learning_rate": 9.846294891337463e-05,
+      "loss": 0.4062,
+      "step": 10420
+    },
+    {
+      "epoch": 2.5096292729898892,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.83957638120829e-05,
+      "loss": 0.4147,
+      "step": 10425
+    },
+    {
+      "epoch": 2.5108329321136256,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.832860543549933e-05,
+      "loss": 0.4149,
+      "step": 10430
+    },
+    {
+      "epoch": 2.5120365912373614,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.826147384389636e-05,
+      "loss": 0.3858,
+      "step": 10435
+    },
+    {
+      "epoch": 2.5132402503610978,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.819436909752232e-05,
+      "loss": 0.3821,
+      "step": 10440
+    },
+    {
+      "epoch": 2.514443909484834,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.812729125660146e-05,
+      "loss": 0.4158,
+      "step": 10445
+    },
+    {
+      "epoch": 2.51564756860857,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.806024038133389e-05,
+      "loss": 0.3795,
+      "step": 10450
+    },
+    {
+      "epoch": 2.5168512277323063,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.79932165318956e-05,
+      "loss": 0.4036,
+      "step": 10455
+    },
+    {
+      "epoch": 2.518054886856042,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.792621976843817e-05,
+      "loss": 0.392,
+      "step": 10460
+    },
+    {
+      "epoch": 2.5192585459797785,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.785925015108904e-05,
+      "loss": 0.3991,
+      "step": 10465
+    },
+    {
+      "epoch": 2.520462205103515,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.779230773995113e-05,
+      "loss": 0.3894,
+      "step": 10470
+    },
+    {
+      "epoch": 2.5216658642272507,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.772539259510305e-05,
+      "loss": 0.4063,
+      "step": 10475
+    },
+    {
+      "epoch": 2.522869523350987,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.765850477659892e-05,
+      "loss": 0.3827,
+      "step": 10480
+    },
+    {
+      "epoch": 2.524073182474723,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.759164434446832e-05,
+      "loss": 0.3864,
+      "step": 10485
+    },
+    {
+      "epoch": 2.525276841598459,
+      "grad_norm": 1.0,
+      "learning_rate": 9.75248113587162e-05,
+      "loss": 0.3797,
+      "step": 10490
+    },
+    {
+      "epoch": 2.5264805007221955,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.745800587932298e-05,
+      "loss": 0.3853,
+      "step": 10495
+    },
+    {
+      "epoch": 2.527684159845932,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.739122796624431e-05,
+      "loss": 0.3866,
+      "step": 10500
+    },
+    {
+      "epoch": 2.527684159845932,
+      "eval_loss": 0.3630833029747009,
+      "eval_runtime": 2.3394,
+      "eval_samples_per_second": 85.493,
+      "eval_steps_per_second": 85.493,
+      "step": 10500
+    },
+    {
+      "epoch": 2.5288878189696677,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.73244776794112e-05,
+      "loss": 0.397,
+      "step": 10505
+    },
+    {
+      "epoch": 2.530091478093404,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.725775507872973e-05,
+      "loss": 0.4032,
+      "step": 10510
+    },
+    {
+      "epoch": 2.53129513721714,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.719106022408123e-05,
+      "loss": 0.3882,
+      "step": 10515
+    },
+    {
+      "epoch": 2.5324987963408763,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.712439317532215e-05,
+      "loss": 0.3823,
+      "step": 10520
+    },
+    {
+      "epoch": 2.5337024554646126,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.705775399228387e-05,
+      "loss": 0.3973,
+      "step": 10525
+    },
+    {
+      "epoch": 2.5349061145883485,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.699114273477283e-05,
+      "loss": 0.3695,
+      "step": 10530
+    },
+    {
+      "epoch": 2.536109773712085,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.692455946257045e-05,
+      "loss": 0.3957,
+      "step": 10535
+    },
+    {
+      "epoch": 2.5373134328358207,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.685800423543297e-05,
+      "loss": 0.3967,
+      "step": 10540
+    },
+    {
+      "epoch": 2.538517091959557,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.679147711309146e-05,
+      "loss": 0.3955,
+      "step": 10545
+    },
+    {
+      "epoch": 2.5397207510832933,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.672497815525183e-05,
+      "loss": 0.3963,
+      "step": 10550
+    },
+    {
+      "epoch": 2.5409244102070296,
+      "grad_norm": 1.125,
+      "learning_rate": 9.665850742159466e-05,
+      "loss": 0.4036,
+      "step": 10555
+    },
+    {
+      "epoch": 2.5421280693307655,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.659206497177522e-05,
+      "loss": 0.3887,
+      "step": 10560
+    },
+    {
+      "epoch": 2.543331728454502,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.652565086542337e-05,
+      "loss": 0.3876,
+      "step": 10565
+    },
+    {
+      "epoch": 2.5445353875782377,
+      "grad_norm": 1.125,
+      "learning_rate": 9.645926516214354e-05,
+      "loss": 0.3917,
+      "step": 10570
+    },
+    {
+      "epoch": 2.545739046701974,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.639290792151467e-05,
+      "loss": 0.3657,
+      "step": 10575
+    },
+    {
+      "epoch": 2.5469427058257104,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.632657920309022e-05,
+      "loss": 0.4036,
+      "step": 10580
+    },
+    {
+      "epoch": 2.5481463649494462,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.626027906639795e-05,
+      "loss": 0.4352,
+      "step": 10585
+    },
+    {
+      "epoch": 2.5493500240731826,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.619400757094004e-05,
+      "loss": 0.3994,
+      "step": 10590
+    },
+    {
+      "epoch": 2.5505536831969184,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.612776477619295e-05,
+      "loss": 0.4286,
+      "step": 10595
+    },
+    {
+      "epoch": 2.5517573423206548,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.606155074160736e-05,
+      "loss": 0.3913,
+      "step": 10600
+    },
+    {
+      "epoch": 2.552961001444391,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.599536552660815e-05,
+      "loss": 0.419,
+      "step": 10605
+    },
+    {
+      "epoch": 2.554164660568127,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.592920919059434e-05,
+      "loss": 0.3813,
+      "step": 10610
+    },
+    {
+      "epoch": 2.5553683196918633,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.586308179293904e-05,
+      "loss": 0.3909,
+      "step": 10615
+    },
+    {
+      "epoch": 2.556571978815599,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.579698339298932e-05,
+      "loss": 0.3785,
+      "step": 10620
+    },
+    {
+      "epoch": 2.5577756379393355,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.573091405006638e-05,
+      "loss": 0.3863,
+      "step": 10625
+    },
+    {
+      "epoch": 2.558979297063072,
+      "grad_norm": 1.0,
+      "learning_rate": 9.56648738234652e-05,
+      "loss": 0.4144,
+      "step": 10630
+    },
+    {
+      "epoch": 2.560182956186808,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.559886277245465e-05,
+      "loss": 0.4071,
+      "step": 10635
+    },
+    {
+      "epoch": 2.561386615310544,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.553288095627754e-05,
+      "loss": 0.3946,
+      "step": 10640
+    },
+    {
+      "epoch": 2.5625902744342803,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.546692843415022e-05,
+      "loss": 0.4195,
+      "step": 10645
+    },
+    {
+      "epoch": 2.563793933558016,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.540100526526292e-05,
+      "loss": 0.3856,
+      "step": 10650
+    },
+    {
+      "epoch": 2.5649975926817525,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.53351115087795e-05,
+      "loss": 0.3962,
+      "step": 10655
+    },
+    {
+      "epoch": 2.566201251805489,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.52692472238374e-05,
+      "loss": 0.398,
+      "step": 10660
+    },
+    {
+      "epoch": 2.5674049109292247,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.520341246954757e-05,
+      "loss": 0.4203,
+      "step": 10665
+    },
+    {
+      "epoch": 2.568608570052961,
+      "grad_norm": 1.0,
+      "learning_rate": 9.51376073049946e-05,
+      "loss": 0.402,
+      "step": 10670
+    },
+    {
+      "epoch": 2.569812229176697,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.507183178923635e-05,
+      "loss": 0.3876,
+      "step": 10675
+    },
+    {
+      "epoch": 2.5710158883004333,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.500608598130415e-05,
+      "loss": 0.3616,
+      "step": 10680
+    },
+    {
+      "epoch": 2.5722195474241696,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.49403699402027e-05,
+      "loss": 0.3868,
+      "step": 10685
+    },
+    {
+      "epoch": 2.573423206547906,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.487468372490991e-05,
+      "loss": 0.4075,
+      "step": 10690
+    },
+    {
+      "epoch": 2.574626865671642,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.480902739437695e-05,
+      "loss": 0.415,
+      "step": 10695
+    },
+    {
+      "epoch": 2.575830524795378,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.474340100752823e-05,
+      "loss": 0.3818,
+      "step": 10700
+    },
+    {
+      "epoch": 2.577034183919114,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.467780462326124e-05,
+      "loss": 0.3997,
+      "step": 10705
+    },
+    {
+      "epoch": 2.5782378430428503,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.461223830044651e-05,
+      "loss": 0.4047,
+      "step": 10710
+    },
+    {
+      "epoch": 2.5794415021665866,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.454670209792762e-05,
+      "loss": 0.399,
+      "step": 10715
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.448119607452115e-05,
+      "loss": 0.4255,
+      "step": 10720
+    },
+    {
+      "epoch": 2.581848820414059,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.441572028901658e-05,
+      "loss": 0.3836,
+      "step": 10725
+    },
+    {
+      "epoch": 2.5830524795377947,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.43502748001762e-05,
+      "loss": 0.4085,
+      "step": 10730
+    },
+    {
+      "epoch": 2.584256138661531,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.428485966673516e-05,
+      "loss": 0.4165,
+      "step": 10735
+    },
+    {
+      "epoch": 2.5854597977852674,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.421947494740135e-05,
+      "loss": 0.3771,
+      "step": 10740
+    },
+    {
+      "epoch": 2.5866634569090032,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.415412070085535e-05,
+      "loss": 0.415,
+      "step": 10745
+    },
+    {
+      "epoch": 2.5878671160327396,
+      "grad_norm": 1.0,
+      "learning_rate": 9.408879698575049e-05,
+      "loss": 0.3786,
+      "step": 10750
+    },
+    {
+      "epoch": 2.5890707751564754,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.402350386071254e-05,
+      "loss": 0.3987,
+      "step": 10755
+    },
+    {
+      "epoch": 2.5902744342802118,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.395824138433992e-05,
+      "loss": 0.3924,
+      "step": 10760
+    },
+    {
+      "epoch": 2.591478093403948,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.389300961520358e-05,
+      "loss": 0.3643,
+      "step": 10765
+    },
+    {
+      "epoch": 2.5926817525276844,
+      "grad_norm": 1.1171875,
+      "learning_rate": 9.382780861184676e-05,
+      "loss": 0.3906,
+      "step": 10770
+    },
+    {
+      "epoch": 2.5938854116514203,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.37626384327852e-05,
+      "loss": 0.3635,
+      "step": 10775
+    },
+    {
+      "epoch": 2.5950890707751566,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.369749913650702e-05,
+      "loss": 0.4274,
+      "step": 10780
+    },
+    {
+      "epoch": 2.5962927298988925,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.363239078147253e-05,
+      "loss": 0.3744,
+      "step": 10785
+    },
+    {
+      "epoch": 2.597496389022629,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.356731342611423e-05,
+      "loss": 0.401,
+      "step": 10790
+    },
+    {
+      "epoch": 2.598700048146365,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.350226712883697e-05,
+      "loss": 0.3947,
+      "step": 10795
+    },
+    {
+      "epoch": 2.599903707270101,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.34372519480176e-05,
+      "loss": 0.3952,
+      "step": 10800
+    },
+    {
+      "epoch": 2.6011073663938373,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.337226794200508e-05,
+      "loss": 0.3771,
+      "step": 10805
+    },
+    {
+      "epoch": 2.602311025517573,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.330731516912035e-05,
+      "loss": 0.4102,
+      "step": 10810
+    },
+    {
+      "epoch": 2.6035146846413095,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.324239368765638e-05,
+      "loss": 0.3882,
+      "step": 10815
+    },
+    {
+      "epoch": 2.604718343765046,
+      "grad_norm": 1.0,
+      "learning_rate": 9.317750355587799e-05,
+      "loss": 0.3778,
+      "step": 10820
+    },
+    {
+      "epoch": 2.605922002888782,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.311264483202195e-05,
+      "loss": 0.4163,
+      "step": 10825
+    },
+    {
+      "epoch": 2.607125662012518,
+      "grad_norm": 0.9765625,
+      "learning_rate": 9.30478175742968e-05,
+      "loss": 0.4155,
+      "step": 10830
+    },
+    {
+      "epoch": 2.6083293211362544,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.29830218408828e-05,
+      "loss": 0.4223,
+      "step": 10835
+    },
+    {
+      "epoch": 2.6095329802599903,
+      "grad_norm": 0.98046875,
+      "learning_rate": 9.291825768993194e-05,
+      "loss": 0.3742,
+      "step": 10840
+    },
+    {
+      "epoch": 2.6107366393837266,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.285352517956799e-05,
+      "loss": 0.4121,
+      "step": 10845
+    },
+    {
+      "epoch": 2.611940298507463,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.278882436788607e-05,
+      "loss": 0.4198,
+      "step": 10850
+    },
+    {
+      "epoch": 2.6131439576311988,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.27241553129531e-05,
+      "loss": 0.3995,
+      "step": 10855
+    },
+    {
+      "epoch": 2.614347616754935,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.265951807280735e-05,
+      "loss": 0.4032,
+      "step": 10860
+    },
+    {
+      "epoch": 2.615551275878671,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.259491270545859e-05,
+      "loss": 0.3969,
+      "step": 10865
+    },
+    {
+      "epoch": 2.6167549350024073,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.253033926888795e-05,
+      "loss": 0.3794,
+      "step": 10870
+    },
+    {
+      "epoch": 2.6179585941261436,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.2465797821048e-05,
+      "loss": 0.4146,
+      "step": 10875
+    },
+    {
+      "epoch": 2.6191622532498795,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.240128841986247e-05,
+      "loss": 0.414,
+      "step": 10880
+    },
+    {
+      "epoch": 2.620365912373616,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.233681112322644e-05,
+      "loss": 0.4071,
+      "step": 10885
+    },
+    {
+      "epoch": 2.6215695714973517,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.227236598900608e-05,
+      "loss": 0.3857,
+      "step": 10890
+    },
+    {
+      "epoch": 2.622773230621088,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.220795307503878e-05,
+      "loss": 0.3946,
+      "step": 10895
+    },
+    {
+      "epoch": 2.6239768897448243,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.214357243913294e-05,
+      "loss": 0.3633,
+      "step": 10900
+    },
+    {
+      "epoch": 2.6251805488685607,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.207922413906806e-05,
+      "loss": 0.3983,
+      "step": 10905
+    },
+    {
+      "epoch": 2.6263842079922965,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.201490823259459e-05,
+      "loss": 0.4072,
+      "step": 10910
+    },
+    {
+      "epoch": 2.627587867116033,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.195062477743388e-05,
+      "loss": 0.3836,
+      "step": 10915
+    },
+    {
+      "epoch": 2.6287915262397687,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.188637383127822e-05,
+      "loss": 0.3868,
+      "step": 10920
+    },
+    {
+      "epoch": 2.629995185363505,
+      "grad_norm": 0.98828125,
+      "learning_rate": 9.182215545179071e-05,
+      "loss": 0.3938,
+      "step": 10925
+    },
+    {
+      "epoch": 2.6311988444872414,
+      "grad_norm": 1.0,
+      "learning_rate": 9.175796969660507e-05,
+      "loss": 0.3903,
+      "step": 10930
+    },
+    {
+      "epoch": 2.6324025036109773,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.1693816623326e-05,
+      "loss": 0.4061,
+      "step": 10935
+    },
+    {
+      "epoch": 2.6336061627347136,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.162969628952869e-05,
+      "loss": 0.4212,
+      "step": 10940
+    },
+    {
+      "epoch": 2.6348098218584495,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.156560875275898e-05,
+      "loss": 0.3982,
+      "step": 10945
+    },
+    {
+      "epoch": 2.636013480982186,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.150155407053333e-05,
+      "loss": 0.4195,
+      "step": 10950
+    },
+    {
+      "epoch": 2.637217140105922,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.143753230033865e-05,
+      "loss": 0.3932,
+      "step": 10955
+    },
+    {
+      "epoch": 2.638420799229658,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.137354349963234e-05,
+      "loss": 0.388,
+      "step": 10960
+    },
+    {
+      "epoch": 2.6396244583533943,
+      "grad_norm": 1.125,
+      "learning_rate": 9.130958772584227e-05,
+      "loss": 0.4106,
+      "step": 10965
+    },
+    {
+      "epoch": 2.64082811747713,
+      "grad_norm": 1.140625,
+      "learning_rate": 9.124566503636652e-05,
+      "loss": 0.4392,
+      "step": 10970
+    },
+    {
+      "epoch": 2.6420317766008665,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.118177548857359e-05,
+      "loss": 0.3709,
+      "step": 10975
+    },
+    {
+      "epoch": 2.643235435724603,
+      "grad_norm": 0.9609375,
+      "learning_rate": 9.111791913980223e-05,
+      "loss": 0.3966,
+      "step": 10980
+    },
+    {
+      "epoch": 2.644439094848339,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.10540960473614e-05,
+      "loss": 0.3767,
+      "step": 10985
+    },
+    {
+      "epoch": 2.645642753972075,
+      "grad_norm": 1.0,
+      "learning_rate": 9.099030626853014e-05,
+      "loss": 0.4308,
+      "step": 10990
+    },
+    {
+      "epoch": 2.6468464130958114,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.092654986055768e-05,
+      "loss": 0.4138,
+      "step": 10995
+    },
+    {
+      "epoch": 2.6480500722195472,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.086282688066329e-05,
+      "loss": 0.404,
+      "step": 11000
+    },
+    {
+      "epoch": 2.6480500722195472,
+      "eval_loss": 0.35517600178718567,
+      "eval_runtime": 2.3728,
+      "eval_samples_per_second": 84.287,
+      "eval_steps_per_second": 84.287,
+      "step": 11000
+    },
+    {
+      "epoch": 2.6492537313432836,
+      "grad_norm": 1.1015625,
+      "learning_rate": 9.079913738603618e-05,
+      "loss": 0.3896,
+      "step": 11005
+    },
+    {
+      "epoch": 2.65045739046702,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.073548143383552e-05,
+      "loss": 0.3777,
+      "step": 11010
+    },
+    {
+      "epoch": 2.6516610495907558,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.067185908119042e-05,
+      "loss": 0.3989,
+      "step": 11015
+    },
+    {
+      "epoch": 2.652864708714492,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.060827038519982e-05,
+      "loss": 0.4111,
+      "step": 11020
+    },
+    {
+      "epoch": 2.654068367838228,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.054471540293246e-05,
+      "loss": 0.4184,
+      "step": 11025
+    },
+    {
+      "epoch": 2.6552720269619643,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.048119419142682e-05,
+      "loss": 0.3772,
+      "step": 11030
+    },
+    {
+      "epoch": 2.6564756860857006,
+      "grad_norm": 1.0390625,
+      "learning_rate": 9.041770680769101e-05,
+      "loss": 0.3998,
+      "step": 11035
+    },
+    {
+      "epoch": 2.657679345209437,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.035425330870288e-05,
+      "loss": 0.3967,
+      "step": 11040
+    },
+    {
+      "epoch": 2.658883004333173,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.029083375140985e-05,
+      "loss": 0.3764,
+      "step": 11045
+    },
+    {
+      "epoch": 2.660086663456909,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.02274481927288e-05,
+      "loss": 0.3922,
+      "step": 11050
+    },
+    {
+      "epoch": 2.661290322580645,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.016409668954613e-05,
+      "loss": 0.3777,
+      "step": 11055
+    },
+    {
+      "epoch": 2.6624939817043813,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.010077929871774e-05,
+      "loss": 0.4263,
+      "step": 11060
+    },
+    {
+      "epoch": 2.6636976408281177,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.00374960770689e-05,
+      "loss": 0.3757,
+      "step": 11065
+    },
+    {
+      "epoch": 2.6649012999518535,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.997424708139413e-05,
+      "loss": 0.4283,
+      "step": 11070
+    },
+    {
+      "epoch": 2.66610495907559,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.991103236845732e-05,
+      "loss": 0.3888,
+      "step": 11075
+    },
+    {
+      "epoch": 2.6673086181993257,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.984785199499156e-05,
+      "loss": 0.4085,
+      "step": 11080
+    },
+    {
+      "epoch": 2.668512277323062,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.978470601769912e-05,
+      "loss": 0.4253,
+      "step": 11085
+    },
+    {
+      "epoch": 2.6697159364467984,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.97215944932514e-05,
+      "loss": 0.4471,
+      "step": 11090
+    },
+    {
+      "epoch": 2.6709195955705343,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.965851747828886e-05,
+      "loss": 0.3811,
+      "step": 11095
+    },
+    {
+      "epoch": 2.6721232546942706,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.959547502942104e-05,
+      "loss": 0.3741,
+      "step": 11100
+    },
+    {
+      "epoch": 2.6733269138180065,
+      "grad_norm": 0.90625,
+      "learning_rate": 8.953246720322645e-05,
+      "loss": 0.3827,
+      "step": 11105
+    },
+    {
+      "epoch": 2.674530572941743,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.946949405625249e-05,
+      "loss": 0.3894,
+      "step": 11110
+    },
+    {
+      "epoch": 2.675734232065479,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.940655564501541e-05,
+      "loss": 0.4108,
+      "step": 11115
+    },
+    {
+      "epoch": 2.6769378911892154,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.93436520260004e-05,
+      "loss": 0.4081,
+      "step": 11120
+    },
+    {
+      "epoch": 2.6781415503129513,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.928078325566136e-05,
+      "loss": 0.3856,
+      "step": 11125
+    },
+    {
+      "epoch": 2.6793452094366876,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.921794939042084e-05,
+      "loss": 0.3938,
+      "step": 11130
+    },
+    {
+      "epoch": 2.6805488685604235,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.915515048667017e-05,
+      "loss": 0.3602,
+      "step": 11135
+    },
+    {
+      "epoch": 2.68175252768416,
+      "grad_norm": 0.94921875,
+      "learning_rate": 8.909238660076928e-05,
+      "loss": 0.3779,
+      "step": 11140
+    },
+    {
+      "epoch": 2.682956186807896,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.902965778904661e-05,
+      "loss": 0.4019,
+      "step": 11145
+    },
+    {
+      "epoch": 2.684159845931632,
+      "grad_norm": 2.1875,
+      "learning_rate": 8.896696410779925e-05,
+      "loss": 0.3877,
+      "step": 11150
+    },
+    {
+      "epoch": 2.6853635050553684,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.890430561329264e-05,
+      "loss": 0.3855,
+      "step": 11155
+    },
+    {
+      "epoch": 2.6865671641791042,
+      "grad_norm": 1.0,
+      "learning_rate": 8.884168236176068e-05,
+      "loss": 0.3857,
+      "step": 11160
+    },
+    {
+      "epoch": 2.6877708233028406,
+      "grad_norm": 1.0,
+      "learning_rate": 8.877909440940567e-05,
+      "loss": 0.3958,
+      "step": 11165
+    },
+    {
+      "epoch": 2.688974482426577,
+      "grad_norm": 0.88671875,
+      "learning_rate": 8.871654181239817e-05,
+      "loss": 0.41,
+      "step": 11170
+    },
+    {
+      "epoch": 2.690178141550313,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.865402462687704e-05,
+      "loss": 0.4004,
+      "step": 11175
+    },
+    {
+      "epoch": 2.691381800674049,
+      "grad_norm": 1.0546875,
+      "learning_rate": 8.859154290894942e-05,
+      "loss": 0.3717,
+      "step": 11180
+    },
+    {
+      "epoch": 2.6925854597977854,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.852909671469052e-05,
+      "loss": 0.3802,
+      "step": 11185
+    },
+    {
+      "epoch": 2.6937891189215213,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.846668610014373e-05,
+      "loss": 0.388,
+      "step": 11190
+    },
+    {
+      "epoch": 2.6949927780452576,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.840431112132045e-05,
+      "loss": 0.4203,
+      "step": 11195
+    },
+    {
+      "epoch": 2.696196437168994,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.83419718342002e-05,
+      "loss": 0.3832,
+      "step": 11200
+    },
+    {
+      "epoch": 2.69740009629273,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.827966829473039e-05,
+      "loss": 0.374,
+      "step": 11205
+    },
+    {
+      "epoch": 2.698603755416466,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.821740055882634e-05,
+      "loss": 0.3974,
+      "step": 11210
+    },
+    {
+      "epoch": 2.699807414540202,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.815516868237128e-05,
+      "loss": 0.3399,
+      "step": 11215
+    },
+    {
+      "epoch": 2.7010110736639383,
+      "grad_norm": 1.0,
+      "learning_rate": 8.809297272121622e-05,
+      "loss": 0.3781,
+      "step": 11220
+    },
+    {
+      "epoch": 2.7022147327876747,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.803081273117995e-05,
+      "loss": 0.3578,
+      "step": 11225
+    },
+    {
+      "epoch": 2.7034183919114105,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.796868876804903e-05,
+      "loss": 0.383,
+      "step": 11230
+    },
+    {
+      "epoch": 2.704622051035147,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.790660088757762e-05,
+      "loss": 0.3848,
+      "step": 11235
+    },
+    {
+      "epoch": 2.7058257101588827,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.784454914548747e-05,
+      "loss": 0.4085,
+      "step": 11240
+    },
+    {
+      "epoch": 2.707029369282619,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.778253359746804e-05,
+      "loss": 0.3852,
+      "step": 11245
+    },
+    {
+      "epoch": 2.7082330284063554,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.772055429917616e-05,
+      "loss": 0.4188,
+      "step": 11250
+    },
+    {
+      "epoch": 2.7094366875300917,
+      "grad_norm": 0.8828125,
+      "learning_rate": 8.765861130623611e-05,
+      "loss": 0.3809,
+      "step": 11255
+    },
+    {
+      "epoch": 2.7106403466538276,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.759670467423976e-05,
+      "loss": 0.3768,
+      "step": 11260
+    },
+    {
+      "epoch": 2.711844005777564,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.753483445874621e-05,
+      "loss": 0.3776,
+      "step": 11265
+    },
+    {
+      "epoch": 2.7130476649013,
+      "grad_norm": 1.0,
+      "learning_rate": 8.747300071528187e-05,
+      "loss": 0.3808,
+      "step": 11270
+    },
+    {
+      "epoch": 2.714251324025036,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.741120349934053e-05,
+      "loss": 0.4156,
+      "step": 11275
+    },
+    {
+      "epoch": 2.7154549831487724,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.734944286638307e-05,
+      "loss": 0.3832,
+      "step": 11280
+    },
+    {
+      "epoch": 2.7166586422725083,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.728771887183763e-05,
+      "loss": 0.3803,
+      "step": 11285
+    },
+    {
+      "epoch": 2.7178623013962446,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.722603157109941e-05,
+      "loss": 0.3753,
+      "step": 11290
+    },
+    {
+      "epoch": 2.7190659605199805,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.71643810195307e-05,
+      "loss": 0.4058,
+      "step": 11295
+    },
+    {
+      "epoch": 2.720269619643717,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.710276727246078e-05,
+      "loss": 0.3838,
+      "step": 11300
+    },
+    {
+      "epoch": 2.721473278767453,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.704119038518599e-05,
+      "loss": 0.4128,
+      "step": 11305
+    },
+    {
+      "epoch": 2.7226769378911895,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.697965041296952e-05,
+      "loss": 0.3971,
+      "step": 11310
+    },
+    {
+      "epoch": 2.7238805970149254,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.691814741104139e-05,
+      "loss": 0.3698,
+      "step": 11315
+    },
+    {
+      "epoch": 2.7250842561386617,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.685668143459852e-05,
+      "loss": 0.395,
+      "step": 11320
+    },
+    {
+      "epoch": 2.7262879152623976,
+      "grad_norm": 1.0,
+      "learning_rate": 8.679525253880459e-05,
+      "loss": 0.3841,
+      "step": 11325
+    },
+    {
+      "epoch": 2.727491574386134,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.673386077878993e-05,
+      "loss": 0.3734,
+      "step": 11330
+    },
+    {
+      "epoch": 2.72869523350987,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.667250620965166e-05,
+      "loss": 0.4006,
+      "step": 11335
+    },
+    {
+      "epoch": 2.729898892633606,
+      "grad_norm": 0.99609375,
+      "learning_rate": 8.661118888645338e-05,
+      "loss": 0.3828,
+      "step": 11340
+    },
+    {
+      "epoch": 2.7311025517573424,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.654990886422538e-05,
+      "loss": 0.3761,
+      "step": 11345
+    },
+    {
+      "epoch": 2.7323062108810783,
+      "grad_norm": 0.93359375,
+      "learning_rate": 8.648866619796439e-05,
+      "loss": 0.3668,
+      "step": 11350
+    },
+    {
+      "epoch": 2.7335098700048146,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.642746094263373e-05,
+      "loss": 0.3705,
+      "step": 11355
+    },
+    {
+      "epoch": 2.734713529128551,
+      "grad_norm": 0.99609375,
+      "learning_rate": 8.636629315316303e-05,
+      "loss": 0.3938,
+      "step": 11360
+    },
+    {
+      "epoch": 2.735917188252287,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.630516288444831e-05,
+      "loss": 0.4106,
+      "step": 11365
+    },
+    {
+      "epoch": 2.737120847376023,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.624407019135204e-05,
+      "loss": 0.392,
+      "step": 11370
+    },
+    {
+      "epoch": 2.738324506499759,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.618301512870276e-05,
+      "loss": 0.4202,
+      "step": 11375
+    },
+    {
+      "epoch": 2.7395281656234953,
+      "grad_norm": 0.91796875,
+      "learning_rate": 8.612199775129536e-05,
+      "loss": 0.3734,
+      "step": 11380
+    },
+    {
+      "epoch": 2.7407318247472316,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.606101811389095e-05,
+      "loss": 0.3714,
+      "step": 11385
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 0.98828125,
+      "learning_rate": 8.60000762712167e-05,
+      "loss": 0.401,
+      "step": 11390
+    },
+    {
+      "epoch": 2.743139142994704,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.593917227796585e-05,
+      "loss": 0.3891,
+      "step": 11395
+    },
+    {
+      "epoch": 2.74434280211844,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.587830618879773e-05,
+      "loss": 0.389,
+      "step": 11400
+    },
+    {
+      "epoch": 2.745546461242176,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.581747805833761e-05,
+      "loss": 0.4075,
+      "step": 11405
+    },
+    {
+      "epoch": 2.7467501203659124,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.575668794117669e-05,
+      "loss": 0.4192,
+      "step": 11410
+    },
+    {
+      "epoch": 2.7479537794896487,
+      "grad_norm": 1.0,
+      "learning_rate": 8.569593589187211e-05,
+      "loss": 0.3831,
+      "step": 11415
+    },
+    {
+      "epoch": 2.7491574386133846,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.563522196494674e-05,
+      "loss": 0.4003,
+      "step": 11420
+    },
+    {
+      "epoch": 2.750361097737121,
+      "grad_norm": 1.0,
+      "learning_rate": 8.557454621488932e-05,
+      "loss": 0.379,
+      "step": 11425
+    },
+    {
+      "epoch": 2.7515647568608568,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.551390869615435e-05,
+      "loss": 0.4004,
+      "step": 11430
+    },
+    {
+      "epoch": 2.752768415984593,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.545330946316195e-05,
+      "loss": 0.401,
+      "step": 11435
+    },
+    {
+      "epoch": 2.7539720751083294,
+      "grad_norm": 1.0,
+      "learning_rate": 8.539274857029791e-05,
+      "loss": 0.3789,
+      "step": 11440
+    },
+    {
+      "epoch": 2.7551757342320657,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.533222607191355e-05,
+      "loss": 0.3897,
+      "step": 11445
+    },
+    {
+      "epoch": 2.7563793933558016,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.527174202232593e-05,
+      "loss": 0.398,
+      "step": 11450
+    },
+    {
+      "epoch": 2.757583052479538,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.521129647581729e-05,
+      "loss": 0.3888,
+      "step": 11455
+    },
+    {
+      "epoch": 2.758786711603274,
+      "grad_norm": 0.93359375,
+      "learning_rate": 8.51508894866356e-05,
+      "loss": 0.3683,
+      "step": 11460
+    },
+    {
+      "epoch": 2.75999037072701,
+      "grad_norm": 1.0,
+      "learning_rate": 8.509052110899409e-05,
+      "loss": 0.3711,
+      "step": 11465
+    },
+    {
+      "epoch": 2.7611940298507465,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.503019139707136e-05,
+      "loss": 0.3944,
+      "step": 11470
+    },
+    {
+      "epoch": 2.7623976889744823,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.496990040501128e-05,
+      "loss": 0.3923,
+      "step": 11475
+    },
+    {
+      "epoch": 2.7636013480982187,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.490964818692304e-05,
+      "loss": 0.3809,
+      "step": 11480
+    },
+    {
+      "epoch": 2.7648050072219545,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.484943479688098e-05,
+      "loss": 0.3877,
+      "step": 11485
+    },
+    {
+      "epoch": 2.766008666345691,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.478926028892461e-05,
+      "loss": 0.4207,
+      "step": 11490
+    },
+    {
+      "epoch": 2.767212325469427,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.472912471705853e-05,
+      "loss": 0.3729,
+      "step": 11495
+    },
+    {
+      "epoch": 2.768415984593163,
+      "grad_norm": 0.92578125,
+      "learning_rate": 8.46690281352524e-05,
+      "loss": 0.3772,
+      "step": 11500
+    },
+    {
+      "epoch": 2.768415984593163,
+      "eval_loss": 0.3504635989665985,
+      "eval_runtime": 2.365,
+      "eval_samples_per_second": 84.565,
+      "eval_steps_per_second": 84.565,
+      "step": 11500
+    },
+    {
+      "epoch": 2.7696196437168994,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.46089705974409e-05,
+      "loss": 0.3634,
+      "step": 11505
+    },
+    {
+      "epoch": 2.7708233028406353,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.454895215752366e-05,
+      "loss": 0.3889,
+      "step": 11510
+    },
+    {
+      "epoch": 2.7720269619643716,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.448897286936522e-05,
+      "loss": 0.3705,
+      "step": 11515
+    },
+    {
+      "epoch": 2.773230621088108,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.4429032786795e-05,
+      "loss": 0.3879,
+      "step": 11520
+    },
+    {
+      "epoch": 2.7744342802118442,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.436913196360722e-05,
+      "loss": 0.3815,
+      "step": 11525
+    },
+    {
+      "epoch": 2.77563793933558,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.43092704535609e-05,
+      "loss": 0.3777,
+      "step": 11530
+    },
+    {
+      "epoch": 2.7768415984593164,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.424944831037964e-05,
+      "loss": 0.3965,
+      "step": 11535
+    },
+    {
+      "epoch": 2.7780452575830523,
+      "grad_norm": 0.93359375,
+      "learning_rate": 8.418966558775191e-05,
+      "loss": 0.3988,
+      "step": 11540
+    },
+    {
+      "epoch": 2.7792489167067886,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.412992233933069e-05,
+      "loss": 0.3823,
+      "step": 11545
+    },
+    {
+      "epoch": 2.780452575830525,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.407021861873348e-05,
+      "loss": 0.4054,
+      "step": 11550
+    },
+    {
+      "epoch": 2.781656234954261,
+      "grad_norm": 1.125,
+      "learning_rate": 8.401055447954247e-05,
+      "loss": 0.3903,
+      "step": 11555
+    },
+    {
+      "epoch": 2.782859894077997,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.39509299753042e-05,
+      "loss": 0.4003,
+      "step": 11560
+    },
+    {
+      "epoch": 2.784063553201733,
+      "grad_norm": 0.890625,
+      "learning_rate": 8.38913451595296e-05,
+      "loss": 0.3866,
+      "step": 11565
+    },
+    {
+      "epoch": 2.7852672123254694,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.383180008569418e-05,
+      "loss": 0.3793,
+      "step": 11570
+    },
+    {
+      "epoch": 2.7864708714492057,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.377229480723755e-05,
+      "loss": 0.3959,
+      "step": 11575
+    },
+    {
+      "epoch": 2.787674530572942,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.371282937756373e-05,
+      "loss": 0.3714,
+      "step": 11580
+    },
+    {
+      "epoch": 2.788878189696678,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.365340385004097e-05,
+      "loss": 0.3968,
+      "step": 11585
+    },
+    {
+      "epoch": 2.790081848820414,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.359401827800168e-05,
+      "loss": 0.3889,
+      "step": 11590
+    },
+    {
+      "epoch": 2.79128550794415,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.353467271474245e-05,
+      "loss": 0.3991,
+      "step": 11595
+    },
+    {
+      "epoch": 2.7924891670678864,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.347536721352392e-05,
+      "loss": 0.4005,
+      "step": 11600
+    },
+    {
+      "epoch": 2.7936928261916227,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.341610182757082e-05,
+      "loss": 0.3871,
+      "step": 11605
+    },
+    {
+      "epoch": 2.7948964853153586,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.335687661007182e-05,
+      "loss": 0.3835,
+      "step": 11610
+    },
+    {
+      "epoch": 2.796100144439095,
+      "grad_norm": 0.921875,
+      "learning_rate": 8.32976916141796e-05,
+      "loss": 0.374,
+      "step": 11615
+    },
+    {
+      "epoch": 2.797303803562831,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.323854689301069e-05,
+      "loss": 0.3964,
+      "step": 11620
+    },
+    {
+      "epoch": 2.798507462686567,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.317944249964548e-05,
+      "loss": 0.4016,
+      "step": 11625
+    },
+    {
+      "epoch": 2.7997111218103035,
+      "grad_norm": 1.0546875,
+      "learning_rate": 8.312037848712827e-05,
+      "loss": 0.3821,
+      "step": 11630
+    },
+    {
+      "epoch": 2.8009147809340393,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.306135490846698e-05,
+      "loss": 0.377,
+      "step": 11635
+    },
+    {
+      "epoch": 2.8021184400577757,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.300237181663328e-05,
+      "loss": 0.3933,
+      "step": 11640
+    },
+    {
+      "epoch": 2.8033220991815115,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.294342926456254e-05,
+      "loss": 0.3621,
+      "step": 11645
+    },
+    {
+      "epoch": 2.804525758305248,
+      "grad_norm": 0.98828125,
+      "learning_rate": 8.288452730515381e-05,
+      "loss": 0.3964,
+      "step": 11650
+    },
+    {
+      "epoch": 2.805729417428984,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.282566599126949e-05,
+      "loss": 0.3903,
+      "step": 11655
+    },
+    {
+      "epoch": 2.8069330765527205,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.276684537573574e-05,
+      "loss": 0.38,
+      "step": 11660
+    },
+    {
+      "epoch": 2.8081367356764564,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.270806551134206e-05,
+      "loss": 0.3812,
+      "step": 11665
+    },
+    {
+      "epoch": 2.8093403948001927,
+      "grad_norm": 0.98046875,
+      "learning_rate": 8.264932645084144e-05,
+      "loss": 0.4091,
+      "step": 11670
+    },
+    {
+      "epoch": 2.8105440539239286,
+      "grad_norm": 0.99609375,
+      "learning_rate": 8.259062824695016e-05,
+      "loss": 0.3996,
+      "step": 11675
+    },
+    {
+      "epoch": 2.811747713047665,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.253197095234797e-05,
+      "loss": 0.3889,
+      "step": 11680
+    },
+    {
+      "epoch": 2.8129513721714012,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.247335461967781e-05,
+      "loss": 0.4001,
+      "step": 11685
+    },
+    {
+      "epoch": 2.814155031295137,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.241477930154588e-05,
+      "loss": 0.4161,
+      "step": 11690
+    },
+    {
+      "epoch": 2.8153586904188734,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.235624505052155e-05,
+      "loss": 0.3668,
+      "step": 11695
+    },
+    {
+      "epoch": 2.8165623495426093,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.229775191913736e-05,
+      "loss": 0.4022,
+      "step": 11700
+    },
+    {
+      "epoch": 2.8177660086663456,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.223929995988893e-05,
+      "loss": 0.3847,
+      "step": 11705
+    },
+    {
+      "epoch": 2.818969667790082,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.218088922523498e-05,
+      "loss": 0.4047,
+      "step": 11710
+    },
+    {
+      "epoch": 2.8201733269138183,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.212251976759716e-05,
+      "loss": 0.3711,
+      "step": 11715
+    },
+    {
+      "epoch": 2.821376986037554,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.206419163936011e-05,
+      "loss": 0.3761,
+      "step": 11720
+    },
+    {
+      "epoch": 2.8225806451612905,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.200590489287138e-05,
+      "loss": 0.3882,
+      "step": 11725
+    },
+    {
+      "epoch": 2.8237843042850264,
+      "grad_norm": 0.93359375,
+      "learning_rate": 8.19476595804414e-05,
+      "loss": 0.3845,
+      "step": 11730
+    },
+    {
+      "epoch": 2.8249879634087627,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.188945575434334e-05,
+      "loss": 0.409,
+      "step": 11735
+    },
+    {
+      "epoch": 2.826191622532499,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.183129346681325e-05,
+      "loss": 0.392,
+      "step": 11740
+    },
+    {
+      "epoch": 2.827395281656235,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.177317277004978e-05,
+      "loss": 0.3755,
+      "step": 11745
+    },
+    {
+      "epoch": 2.828598940779971,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.171509371621433e-05,
+      "loss": 0.3835,
+      "step": 11750
+    },
+    {
+      "epoch": 2.829802599903707,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.165705635743094e-05,
+      "loss": 0.3687,
+      "step": 11755
+    },
+    {
+      "epoch": 2.8310062590274434,
+      "grad_norm": 0.9921875,
+      "learning_rate": 8.159906074578616e-05,
+      "loss": 0.3892,
+      "step": 11760
+    },
+    {
+      "epoch": 2.8322099181511797,
+      "grad_norm": 0.98828125,
+      "learning_rate": 8.154110693332913e-05,
+      "loss": 0.4034,
+      "step": 11765
+    },
+    {
+      "epoch": 2.8334135772749156,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.148319497207148e-05,
+      "loss": 0.4024,
+      "step": 11770
+    },
+    {
+      "epoch": 2.834617236398652,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.142532491398725e-05,
+      "loss": 0.3841,
+      "step": 11775
+    },
+    {
+      "epoch": 2.835820895522388,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.136749681101283e-05,
+      "loss": 0.3683,
+      "step": 11780
+    },
+    {
+      "epoch": 2.837024554646124,
+      "grad_norm": 0.921875,
+      "learning_rate": 8.130971071504707e-05,
+      "loss": 0.3895,
+      "step": 11785
+    },
+    {
+      "epoch": 2.8382282137698605,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.125196667795105e-05,
+      "loss": 0.3933,
+      "step": 11790
+    },
+    {
+      "epoch": 2.8394318728935968,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.11942647515481e-05,
+      "loss": 0.4035,
+      "step": 11795
+    },
+    {
+      "epoch": 2.8406355320173327,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.113660498762375e-05,
+      "loss": 0.3824,
+      "step": 11800
+    },
+    {
+      "epoch": 2.841839191141069,
+      "grad_norm": 0.93359375,
+      "learning_rate": 8.107898743792579e-05,
+      "loss": 0.3805,
+      "step": 11805
+    },
+    {
+      "epoch": 2.843042850264805,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.102141215416396e-05,
+      "loss": 0.3917,
+      "step": 11810
+    },
+    {
+      "epoch": 2.844246509388541,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.096387918801019e-05,
+      "loss": 0.3794,
+      "step": 11815
+    },
+    {
+      "epoch": 2.8454501685122775,
+      "grad_norm": 0.99609375,
+      "learning_rate": 8.090638859109839e-05,
+      "loss": 0.4024,
+      "step": 11820
+    },
+    {
+      "epoch": 2.8466538276360134,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.084894041502444e-05,
+      "loss": 0.4483,
+      "step": 11825
+    },
+    {
+      "epoch": 2.8478574867597497,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.079153471134614e-05,
+      "loss": 0.384,
+      "step": 11830
+    },
+    {
+      "epoch": 2.8490611458834856,
+      "grad_norm": 1.0546875,
+      "learning_rate": 8.073417153158325e-05,
+      "loss": 0.3759,
+      "step": 11835
+    },
+    {
+      "epoch": 2.850264805007222,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.067685092721726e-05,
+      "loss": 0.3825,
+      "step": 11840
+    },
+    {
+      "epoch": 2.8514684641309582,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.061957294969148e-05,
+      "loss": 0.3708,
+      "step": 11845
+    },
+    {
+      "epoch": 2.8526721232546945,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.056233765041105e-05,
+      "loss": 0.387,
+      "step": 11850
+    },
+    {
+      "epoch": 2.8538757823784304,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.050514508074272e-05,
+      "loss": 0.41,
+      "step": 11855
+    },
+    {
+      "epoch": 2.8550794415021667,
+      "grad_norm": 0.984375,
+      "learning_rate": 8.044799529201483e-05,
+      "loss": 0.3748,
+      "step": 11860
+    },
+    {
+      "epoch": 2.8562831006259026,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.03908883355175e-05,
+      "loss": 0.3829,
+      "step": 11865
+    },
+    {
+      "epoch": 2.857486759749639,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.033382426250227e-05,
+      "loss": 0.395,
+      "step": 11870
+    },
+    {
+      "epoch": 2.8586904188733753,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.02768031241822e-05,
+      "loss": 0.3944,
+      "step": 11875
+    },
+    {
+      "epoch": 2.859894077997111,
+      "grad_norm": 0.91796875,
+      "learning_rate": 8.021982497173193e-05,
+      "loss": 0.3946,
+      "step": 11880
+    },
+    {
+      "epoch": 2.8610977371208475,
+      "grad_norm": 1.0390625,
+      "learning_rate": 8.016288985628743e-05,
+      "loss": 0.4306,
+      "step": 11885
+    },
+    {
+      "epoch": 2.8623013962445834,
+      "grad_norm": 0.94921875,
+      "learning_rate": 8.010599782894605e-05,
+      "loss": 0.4042,
+      "step": 11890
+    },
+    {
+      "epoch": 2.8635050553683197,
+      "grad_norm": 1.046875,
+      "learning_rate": 8.004914894076649e-05,
+      "loss": 0.3906,
+      "step": 11895
+    },
+    {
+      "epoch": 2.864708714492056,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.99923432427687e-05,
+      "loss": 0.4043,
+      "step": 11900
+    },
+    {
+      "epoch": 2.865912373615792,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.99355807859339e-05,
+      "loss": 0.3804,
+      "step": 11905
+    },
+    {
+      "epoch": 2.867116032739528,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.987886162120453e-05,
+      "loss": 0.374,
+      "step": 11910
+    },
+    {
+      "epoch": 2.868319691863264,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.982218579948415e-05,
+      "loss": 0.3614,
+      "step": 11915
+    },
+    {
+      "epoch": 2.8695233509870004,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.976555337163738e-05,
+      "loss": 0.3907,
+      "step": 11920
+    },
+    {
+      "epoch": 2.8707270101107367,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.970896438848991e-05,
+      "loss": 0.3611,
+      "step": 11925
+    },
+    {
+      "epoch": 2.871930669234473,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.965241890082854e-05,
+      "loss": 0.3854,
+      "step": 11930
+    },
+    {
+      "epoch": 2.873134328358209,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.959591695940088e-05,
+      "loss": 0.3927,
+      "step": 11935
+    },
+    {
+      "epoch": 2.8743379874819452,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.95394586149156e-05,
+      "loss": 0.3551,
+      "step": 11940
+    },
+    {
+      "epoch": 2.875541646605681,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.948304391804213e-05,
+      "loss": 0.384,
+      "step": 11945
+    },
+    {
+      "epoch": 2.8767453057294174,
+      "grad_norm": 0.890625,
+      "learning_rate": 7.942667291941077e-05,
+      "loss": 0.3702,
+      "step": 11950
+    },
+    {
+      "epoch": 2.8779489648531538,
+      "grad_norm": 1.0,
+      "learning_rate": 7.937034566961261e-05,
+      "loss": 0.3534,
+      "step": 11955
+    },
+    {
+      "epoch": 2.8791526239768896,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.931406221919951e-05,
+      "loss": 0.4097,
+      "step": 11960
+    },
+    {
+      "epoch": 2.880356283100626,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.925782261868398e-05,
+      "loss": 0.3739,
+      "step": 11965
+    },
+    {
+      "epoch": 2.881559942224362,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.920162691853912e-05,
+      "loss": 0.4148,
+      "step": 11970
+    },
+    {
+      "epoch": 2.882763601348098,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.91454751691988e-05,
+      "loss": 0.3768,
+      "step": 11975
+    },
+    {
+      "epoch": 2.8839672604718345,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.908936742105723e-05,
+      "loss": 0.4186,
+      "step": 11980
+    },
+    {
+      "epoch": 2.8851709195955704,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.903330372446928e-05,
+      "loss": 0.3519,
+      "step": 11985
+    },
+    {
+      "epoch": 2.8863745787193067,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.897728412975026e-05,
+      "loss": 0.3889,
+      "step": 11990
+    },
+    {
+      "epoch": 2.887578237843043,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.892130868717588e-05,
+      "loss": 0.3637,
+      "step": 11995
+    },
+    {
+      "epoch": 2.888781896966779,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.88653774469822e-05,
+      "loss": 0.3651,
+      "step": 12000
+    },
+    {
+      "epoch": 2.888781896966779,
+      "eval_loss": 0.3513660430908203,
+      "eval_runtime": 2.3807,
+      "eval_samples_per_second": 84.009,
+      "eval_steps_per_second": 84.009,
+      "step": 12000
+    },
+    {
+      "epoch": 2.889985556090515,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.880949045936566e-05,
+      "loss": 0.3601,
+      "step": 12005
+    },
+    {
+      "epoch": 2.8911892152142515,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.875364777448298e-05,
+      "loss": 0.4051,
+      "step": 12010
+    },
+    {
+      "epoch": 2.8923928743379874,
+      "grad_norm": 1.0,
+      "learning_rate": 7.869784944245108e-05,
+      "loss": 0.3925,
+      "step": 12015
+    },
+    {
+      "epoch": 2.8935965334617237,
+      "grad_norm": 1.0,
+      "learning_rate": 7.86420955133471e-05,
+      "loss": 0.3714,
+      "step": 12020
+    },
+    {
+      "epoch": 2.8948001925854596,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.858638603720834e-05,
+      "loss": 0.3792,
+      "step": 12025
+    },
+    {
+      "epoch": 2.896003851709196,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.853072106403215e-05,
+      "loss": 0.3951,
+      "step": 12030
+    },
+    {
+      "epoch": 2.8972075108329323,
+      "grad_norm": 0.91796875,
+      "learning_rate": 7.847510064377607e-05,
+      "loss": 0.3867,
+      "step": 12035
+    },
+    {
+      "epoch": 2.898411169956668,
+      "grad_norm": 0.9140625,
+      "learning_rate": 7.841952482635747e-05,
+      "loss": 0.3871,
+      "step": 12040
+    },
+    {
+      "epoch": 2.8996148290804045,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.836399366165384e-05,
+      "loss": 0.3761,
+      "step": 12045
+    },
+    {
+      "epoch": 2.9008184882041403,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.830850719950255e-05,
+      "loss": 0.4006,
+      "step": 12050
+    },
+    {
+      "epoch": 2.9020221473278767,
+      "grad_norm": 0.98046875,
+      "learning_rate": 7.825306548970087e-05,
+      "loss": 0.403,
+      "step": 12055
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.819766858200578e-05,
+      "loss": 0.3751,
+      "step": 12060
+    },
+    {
+      "epoch": 2.9044294655753493,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.814231652613424e-05,
+      "loss": 0.3819,
+      "step": 12065
+    },
+    {
+      "epoch": 2.905633124699085,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.808700937176286e-05,
+      "loss": 0.384,
+      "step": 12070
+    },
+    {
+      "epoch": 2.9068367838228215,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.803174716852792e-05,
+      "loss": 0.3689,
+      "step": 12075
+    },
+    {
+      "epoch": 2.9080404429465574,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.797652996602545e-05,
+      "loss": 0.3828,
+      "step": 12080
+    },
+    {
+      "epoch": 2.9092441020702937,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.792135781381103e-05,
+      "loss": 0.3655,
+      "step": 12085
+    },
+    {
+      "epoch": 2.91044776119403,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.78662307613998e-05,
+      "loss": 0.4076,
+      "step": 12090
+    },
+    {
+      "epoch": 2.911651420317766,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.781114885826648e-05,
+      "loss": 0.3714,
+      "step": 12095
+    },
+    {
+      "epoch": 2.9128550794415022,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.775611215384523e-05,
+      "loss": 0.3798,
+      "step": 12100
+    },
+    {
+      "epoch": 2.914058738565238,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.770112069752963e-05,
+      "loss": 0.3963,
+      "step": 12105
+    },
+    {
+      "epoch": 2.9152623976889744,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.764617453867265e-05,
+      "loss": 0.3808,
+      "step": 12110
+    },
+    {
+      "epoch": 2.9164660568127108,
+      "grad_norm": 0.97265625,
+      "learning_rate": 7.759127372658672e-05,
+      "loss": 0.3681,
+      "step": 12115
+    },
+    {
+      "epoch": 2.9176697159364466,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.75364183105434e-05,
+      "loss": 0.4188,
+      "step": 12120
+    },
+    {
+      "epoch": 2.918873375060183,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.748160833977361e-05,
+      "loss": 0.3849,
+      "step": 12125
+    },
+    {
+      "epoch": 2.920077034183919,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.742684386346751e-05,
+      "loss": 0.4225,
+      "step": 12130
+    },
+    {
+      "epoch": 2.921280693307655,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.737212493077439e-05,
+      "loss": 0.3444,
+      "step": 12135
+    },
+    {
+      "epoch": 2.9224843524313915,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.731745159080257e-05,
+      "loss": 0.4267,
+      "step": 12140
+    },
+    {
+      "epoch": 2.923688011555128,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.726282389261963e-05,
+      "loss": 0.3857,
+      "step": 12145
+    },
+    {
+      "epoch": 2.9248916706788637,
+      "grad_norm": 1.140625,
+      "learning_rate": 7.720824188525209e-05,
+      "loss": 0.4108,
+      "step": 12150
+    },
+    {
+      "epoch": 2.9260953298026,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.715370561768543e-05,
+      "loss": 0.3833,
+      "step": 12155
+    },
+    {
+      "epoch": 2.927298988926336,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.709921513886419e-05,
+      "loss": 0.3828,
+      "step": 12160
+    },
+    {
+      "epoch": 2.928502648050072,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.704477049769172e-05,
+      "loss": 0.3723,
+      "step": 12165
+    },
+    {
+      "epoch": 2.9297063071738085,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.699037174303027e-05,
+      "loss": 0.3847,
+      "step": 12170
+    },
+    {
+      "epoch": 2.9309099662975444,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.693601892370093e-05,
+      "loss": 0.3992,
+      "step": 12175
+    },
+    {
+      "epoch": 2.9321136254212807,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.68817120884835e-05,
+      "loss": 0.406,
+      "step": 12180
+    },
+    {
+      "epoch": 2.9333172845450166,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.682745128611655e-05,
+      "loss": 0.388,
+      "step": 12185
+    },
+    {
+      "epoch": 2.934520943668753,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.677323656529737e-05,
+      "loss": 0.4004,
+      "step": 12190
+    },
+    {
+      "epoch": 2.9357246027924893,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.671906797468184e-05,
+      "loss": 0.3684,
+      "step": 12195
+    },
+    {
+      "epoch": 2.9369282619162256,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.666494556288443e-05,
+      "loss": 0.3913,
+      "step": 12200
+    },
+    {
+      "epoch": 2.9381319210399615,
+      "grad_norm": 1.1328125,
+      "learning_rate": 7.661086937847827e-05,
+      "loss": 0.3862,
+      "step": 12205
+    },
+    {
+      "epoch": 2.939335580163698,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.655683946999486e-05,
+      "loss": 0.4004,
+      "step": 12210
+    },
+    {
+      "epoch": 2.9405392392874337,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.650285588592427e-05,
+      "loss": 0.3719,
+      "step": 12215
+    },
+    {
+      "epoch": 2.94174289841117,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.644891867471497e-05,
+      "loss": 0.3844,
+      "step": 12220
+    },
+    {
+      "epoch": 2.9429465575349063,
+      "grad_norm": 1.0,
+      "learning_rate": 7.639502788477377e-05,
+      "loss": 0.3966,
+      "step": 12225
+    },
+    {
+      "epoch": 2.944150216658642,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.634118356446585e-05,
+      "loss": 0.3759,
+      "step": 12230
+    },
+    {
+      "epoch": 2.9453538757823785,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.628738576211473e-05,
+      "loss": 0.4173,
+      "step": 12235
+    },
+    {
+      "epoch": 2.9465575349061144,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.623363452600211e-05,
+      "loss": 0.3775,
+      "step": 12240
+    },
+    {
+      "epoch": 2.9477611940298507,
+      "grad_norm": 1.0,
+      "learning_rate": 7.617992990436795e-05,
+      "loss": 0.3572,
+      "step": 12245
+    },
+    {
+      "epoch": 2.948964853153587,
+      "grad_norm": 0.95703125,
+      "learning_rate": 7.612627194541032e-05,
+      "loss": 0.3812,
+      "step": 12250
+    },
+    {
+      "epoch": 2.950168512277323,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.60726606972855e-05,
+      "loss": 0.3631,
+      "step": 12255
+    },
+    {
+      "epoch": 2.9513721714010592,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.60190962081077e-05,
+      "loss": 0.4028,
+      "step": 12260
+    },
+    {
+      "epoch": 2.952575830524795,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.596557852594935e-05,
+      "loss": 0.3728,
+      "step": 12265
+    },
+    {
+      "epoch": 2.9537794896485314,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.591210769884073e-05,
+      "loss": 0.3834,
+      "step": 12270
+    },
+    {
+      "epoch": 2.9549831487722678,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.585868377477015e-05,
+      "loss": 0.3753,
+      "step": 12275
+    },
+    {
+      "epoch": 2.956186807896004,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.580530680168376e-05,
+      "loss": 0.4026,
+      "step": 12280
+    },
+    {
+      "epoch": 2.95739046701974,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.575197682748565e-05,
+      "loss": 0.388,
+      "step": 12285
+    },
+    {
+      "epoch": 2.9585941261434763,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.569869390003767e-05,
+      "loss": 0.3591,
+      "step": 12290
+    },
+    {
+      "epoch": 2.959797785267212,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.564545806715948e-05,
+      "loss": 0.3786,
+      "step": 12295
+    },
+    {
+      "epoch": 2.9610014443909485,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.559226937662843e-05,
+      "loss": 0.3739,
+      "step": 12300
+    },
+    {
+      "epoch": 2.962205103514685,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.553912787617961e-05,
+      "loss": 0.3596,
+      "step": 12305
+    },
+    {
+      "epoch": 2.9634087626384207,
+      "grad_norm": 0.95703125,
+      "learning_rate": 7.548603361350571e-05,
+      "loss": 0.3795,
+      "step": 12310
+    },
+    {
+      "epoch": 2.964612421762157,
+      "grad_norm": 0.9609375,
+      "learning_rate": 7.543298663625714e-05,
+      "loss": 0.3797,
+      "step": 12315
+    },
+    {
+      "epoch": 2.965816080885893,
+      "grad_norm": 0.90234375,
+      "learning_rate": 7.53799869920417e-05,
+      "loss": 0.3962,
+      "step": 12320
+    },
+    {
+      "epoch": 2.967019740009629,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.532703472842484e-05,
+      "loss": 0.3624,
+      "step": 12325
+    },
+    {
+      "epoch": 2.9682233991333655,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.527412989292945e-05,
+      "loss": 0.3776,
+      "step": 12330
+    },
+    {
+      "epoch": 2.969427058257102,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.522127253303586e-05,
+      "loss": 0.394,
+      "step": 12335
+    },
+    {
+      "epoch": 2.9706307173808377,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.516846269618173e-05,
+      "loss": 0.4046,
+      "step": 12340
+    },
+    {
+      "epoch": 2.971834376504574,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.511570042976217e-05,
+      "loss": 0.3818,
+      "step": 12345
+    },
+    {
+      "epoch": 2.97303803562831,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.506298578112956e-05,
+      "loss": 0.3888,
+      "step": 12350
+    },
+    {
+      "epoch": 2.9742416947520463,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.501031879759347e-05,
+      "loss": 0.3774,
+      "step": 12355
+    },
+    {
+      "epoch": 2.9754453538757826,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.495769952642085e-05,
+      "loss": 0.3981,
+      "step": 12360
+    },
+    {
+      "epoch": 2.9766490129995185,
+      "grad_norm": 1.0,
+      "learning_rate": 7.490512801483567e-05,
+      "loss": 0.3815,
+      "step": 12365
+    },
+    {
+      "epoch": 2.9778526721232548,
+      "grad_norm": 0.92578125,
+      "learning_rate": 7.485260431001913e-05,
+      "loss": 0.3572,
+      "step": 12370
+    },
+    {
+      "epoch": 2.9790563312469907,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.48001284591095e-05,
+      "loss": 0.3738,
+      "step": 12375
+    },
+    {
+      "epoch": 2.980259990370727,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.474770050920209e-05,
+      "loss": 0.3787,
+      "step": 12380
+    },
+    {
+      "epoch": 2.9814636494944633,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.46953205073492e-05,
+      "loss": 0.3927,
+      "step": 12385
+    },
+    {
+      "epoch": 2.982667308618199,
+      "grad_norm": 0.87109375,
+      "learning_rate": 7.464298850056022e-05,
+      "loss": 0.3769,
+      "step": 12390
+    },
+    {
+      "epoch": 2.9838709677419355,
+      "grad_norm": 0.953125,
+      "learning_rate": 7.459070453580131e-05,
+      "loss": 0.406,
+      "step": 12395
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.453846865999561e-05,
+      "loss": 0.3943,
+      "step": 12400
+    },
+    {
+      "epoch": 2.9862782859894077,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.448628092002302e-05,
+      "loss": 0.3715,
+      "step": 12405
+    },
+    {
+      "epoch": 2.987481945113144,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.443414136272038e-05,
+      "loss": 0.3995,
+      "step": 12410
+    },
+    {
+      "epoch": 2.9886856042368803,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.438205003488113e-05,
+      "loss": 0.3824,
+      "step": 12415
+    },
+    {
+      "epoch": 2.9898892633606162,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.433000698325551e-05,
+      "loss": 0.3903,
+      "step": 12420
+    },
+    {
+      "epoch": 2.9910929224843525,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.427801225455045e-05,
+      "loss": 0.3945,
+      "step": 12425
+    },
+    {
+      "epoch": 2.9922965816080884,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.422606589542946e-05,
+      "loss": 0.3953,
+      "step": 12430
+    },
+    {
+      "epoch": 2.9935002407318247,
+      "grad_norm": 0.953125,
+      "learning_rate": 7.417416795251261e-05,
+      "loss": 0.3776,
+      "step": 12435
+    },
+    {
+      "epoch": 2.994703899855561,
+      "grad_norm": 1.0,
+      "learning_rate": 7.412231847237666e-05,
+      "loss": 0.4054,
+      "step": 12440
+    },
+    {
+      "epoch": 2.995907558979297,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.407051750155473e-05,
+      "loss": 0.3844,
+      "step": 12445
+    },
+    {
+      "epoch": 2.9971112181030333,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.401876508653645e-05,
+      "loss": 0.4058,
+      "step": 12450
+    },
+    {
+      "epoch": 2.998314877226769,
+      "grad_norm": 0.96875,
+      "learning_rate": 7.396706127376795e-05,
+      "loss": 0.3667,
+      "step": 12455
+    },
+    {
+      "epoch": 2.999277804525758,
+      "eval_loss": 0.3465866446495056,
+      "eval_runtime": 2.3382,
+      "eval_samples_per_second": 85.534,
+      "eval_steps_per_second": 85.534,
+      "step": 12459
+    },
+    {
+      "epoch": 2.9995185363505055,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.391540610965164e-05,
+      "loss": 0.3907,
+      "step": 12460
+    },
+    {
+      "epoch": 3.000722195474242,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.386379964054625e-05,
+      "loss": 0.3545,
+      "step": 12465
+    },
+    {
+      "epoch": 3.0019258545979777,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.381224191276692e-05,
+      "loss": 0.3321,
+      "step": 12470
+    },
+    {
+      "epoch": 3.003129513721714,
+      "grad_norm": 0.90234375,
+      "learning_rate": 7.376073297258498e-05,
+      "loss": 0.3542,
+      "step": 12475
+    },
+    {
+      "epoch": 3.0043331728454503,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.370927286622795e-05,
+      "loss": 0.3535,
+      "step": 12480
+    },
+    {
+      "epoch": 3.005536831969186,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.365786163987959e-05,
+      "loss": 0.3387,
+      "step": 12485
+    },
+    {
+      "epoch": 3.0067404910929225,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.360649933967977e-05,
+      "loss": 0.3413,
+      "step": 12490
+    },
+    {
+      "epoch": 3.007944150216659,
+      "grad_norm": 0.93359375,
+      "learning_rate": 7.35551860117244e-05,
+      "loss": 0.3215,
+      "step": 12495
+    },
+    {
+      "epoch": 3.0091478093403947,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.350392170206548e-05,
+      "loss": 0.3597,
+      "step": 12500
+    },
+    {
+      "epoch": 3.0091478093403947,
+      "eval_loss": 0.34771043062210083,
+      "eval_runtime": 2.3495,
+      "eval_samples_per_second": 85.125,
+      "eval_steps_per_second": 85.125,
+      "step": 12500
+    },
+    {
+      "epoch": 3.010351468464131,
+      "grad_norm": 0.9140625,
+      "learning_rate": 7.345270645671104e-05,
+      "loss": 0.3568,
+      "step": 12505
+    },
+    {
+      "epoch": 3.011555127587867,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.3401540321625e-05,
+      "loss": 0.3309,
+      "step": 12510
+    },
+    {
+      "epoch": 3.0127587867116032,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.335042334272732e-05,
+      "loss": 0.3244,
+      "step": 12515
+    },
+    {
+      "epoch": 3.0139624458353396,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.329935556589376e-05,
+      "loss": 0.3416,
+      "step": 12520
+    },
+    {
+      "epoch": 3.0151661049590754,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.324833703695593e-05,
+      "loss": 0.3388,
+      "step": 12525
+    },
+    {
+      "epoch": 3.0163697640828118,
+      "grad_norm": 0.86328125,
+      "learning_rate": 7.319736780170124e-05,
+      "loss": 0.3633,
+      "step": 12530
+    },
+    {
+      "epoch": 3.017573423206548,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.314644790587291e-05,
+      "loss": 0.3309,
+      "step": 12535
+    },
+    {
+      "epoch": 3.018777082330284,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.309557739516983e-05,
+      "loss": 0.3648,
+      "step": 12540
+    },
+    {
+      "epoch": 3.0199807414540203,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.304475631524658e-05,
+      "loss": 0.3419,
+      "step": 12545
+    },
+    {
+      "epoch": 3.021184400577756,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.29939847117134e-05,
+      "loss": 0.3701,
+      "step": 12550
+    },
+    {
+      "epoch": 3.0223880597014925,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.294326263013607e-05,
+      "loss": 0.3568,
+      "step": 12555
+    },
+    {
+      "epoch": 3.023591718825229,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.289259011603596e-05,
+      "loss": 0.3774,
+      "step": 12560
+    },
+    {
+      "epoch": 3.0247953779489647,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.284196721489001e-05,
+      "loss": 0.3156,
+      "step": 12565
+    },
+    {
+      "epoch": 3.025999037072701,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.279139397213056e-05,
+      "loss": 0.3582,
+      "step": 12570
+    },
+    {
+      "epoch": 3.0272026961964373,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.274087043314535e-05,
+      "loss": 0.3188,
+      "step": 12575
+    },
+    {
+      "epoch": 3.028406355320173,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.269039664327767e-05,
+      "loss": 0.3741,
+      "step": 12580
+    },
+    {
+      "epoch": 3.0296100144439095,
+      "grad_norm": 0.92578125,
+      "learning_rate": 7.263997264782599e-05,
+      "loss": 0.3596,
+      "step": 12585
+    },
+    {
+      "epoch": 3.030813673567646,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.258959849204415e-05,
+      "loss": 0.3508,
+      "step": 12590
+    },
+    {
+      "epoch": 3.0320173326913817,
+      "grad_norm": 0.89453125,
+      "learning_rate": 7.253927422114131e-05,
+      "loss": 0.3177,
+      "step": 12595
+    },
+    {
+      "epoch": 3.033220991815118,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.24889998802818e-05,
+      "loss": 0.3527,
+      "step": 12600
+    },
+    {
+      "epoch": 3.034424650938854,
+      "grad_norm": 0.86328125,
+      "learning_rate": 7.243877551458517e-05,
+      "loss": 0.34,
+      "step": 12605
+    },
+    {
+      "epoch": 3.0356283100625903,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.23886011691261e-05,
+      "loss": 0.3679,
+      "step": 12610
+    },
+    {
+      "epoch": 3.0368319691863266,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.233847688893441e-05,
+      "loss": 0.3341,
+      "step": 12615
+    },
+    {
+      "epoch": 3.0380356283100625,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.228840271899493e-05,
+      "loss": 0.3547,
+      "step": 12620
+    },
+    {
+      "epoch": 3.039239287433799,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.22383787042476e-05,
+      "loss": 0.3499,
+      "step": 12625
+    },
+    {
+      "epoch": 3.040442946557535,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.218840488958728e-05,
+      "loss": 0.3457,
+      "step": 12630
+    },
+    {
+      "epoch": 3.041646605681271,
+      "grad_norm": 1.0,
+      "learning_rate": 7.213848131986375e-05,
+      "loss": 0.3306,
+      "step": 12635
+    },
+    {
+      "epoch": 3.0428502648050073,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.208860803988181e-05,
+      "loss": 0.3348,
+      "step": 12640
+    },
+    {
+      "epoch": 3.044053923928743,
+      "grad_norm": 0.875,
+      "learning_rate": 7.203878509440106e-05,
+      "loss": 0.3394,
+      "step": 12645
+    },
+    {
+      "epoch": 3.0452575830524795,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.198901252813588e-05,
+      "loss": 0.3523,
+      "step": 12650
+    },
+    {
+      "epoch": 3.046461242176216,
+      "grad_norm": 0.98046875,
+      "learning_rate": 7.193929038575552e-05,
+      "loss": 0.3481,
+      "step": 12655
+    },
+    {
+      "epoch": 3.0476649012999517,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.188961871188397e-05,
+      "loss": 0.3477,
+      "step": 12660
+    },
+    {
+      "epoch": 3.048868560423688,
+      "grad_norm": 0.9609375,
+      "learning_rate": 7.183999755109982e-05,
+      "loss": 0.3483,
+      "step": 12665
+    },
+    {
+      "epoch": 3.0500722195474244,
+      "grad_norm": 1.0,
+      "learning_rate": 7.179042694793647e-05,
+      "loss": 0.3367,
+      "step": 12670
+    },
+    {
+      "epoch": 3.0512758786711602,
+      "grad_norm": 0.94140625,
+      "learning_rate": 7.174090694688183e-05,
+      "loss": 0.3189,
+      "step": 12675
+    },
+    {
+      "epoch": 3.0524795377948966,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.169143759237847e-05,
+      "loss": 0.3329,
+      "step": 12680
+    },
+    {
+      "epoch": 3.0536831969186324,
+      "grad_norm": 0.9296875,
+      "learning_rate": 7.164201892882349e-05,
+      "loss": 0.3347,
+      "step": 12685
+    },
+    {
+      "epoch": 3.0548868560423688,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.159265100056849e-05,
+      "loss": 0.3411,
+      "step": 12690
+    },
+    {
+      "epoch": 3.056090515166105,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.154333385191953e-05,
+      "loss": 0.3437,
+      "step": 12695
+    },
+    {
+      "epoch": 3.057294174289841,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.149406752713708e-05,
+      "loss": 0.3245,
+      "step": 12700
+    },
+    {
+      "epoch": 3.0584978334135773,
+      "grad_norm": 0.96875,
+      "learning_rate": 7.144485207043606e-05,
+      "loss": 0.3632,
+      "step": 12705
+    },
+    {
+      "epoch": 3.0597014925373136,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.139568752598565e-05,
+      "loss": 0.3394,
+      "step": 12710
+    },
+    {
+      "epoch": 3.0609051516610495,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.134657393790942e-05,
+      "loss": 0.3449,
+      "step": 12715
+    },
+    {
+      "epoch": 3.062108810784786,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.129751135028517e-05,
+      "loss": 0.3469,
+      "step": 12720
+    },
+    {
+      "epoch": 3.0633124699085217,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.124849980714494e-05,
+      "loss": 0.3697,
+      "step": 12725
+    },
+    {
+      "epoch": 3.064516129032258,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.11995393524749e-05,
+      "loss": 0.3212,
+      "step": 12730
+    },
+    {
+      "epoch": 3.0657197881559943,
+      "grad_norm": 0.96875,
+      "learning_rate": 7.115063003021548e-05,
+      "loss": 0.353,
+      "step": 12735
+    },
+    {
+      "epoch": 3.06692344727973,
+      "grad_norm": 1.0,
+      "learning_rate": 7.110177188426119e-05,
+      "loss": 0.3744,
+      "step": 12740
+    },
+    {
+      "epoch": 3.0681271064034665,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.105296495846046e-05,
+      "loss": 0.3667,
+      "step": 12745
+    },
+    {
+      "epoch": 3.069330765527203,
+      "grad_norm": 0.8828125,
+      "learning_rate": 7.1004209296616e-05,
+      "loss": 0.3405,
+      "step": 12750
+    },
+    {
+      "epoch": 3.0705344246509387,
+      "grad_norm": 0.91015625,
+      "learning_rate": 7.095550494248432e-05,
+      "loss": 0.3415,
+      "step": 12755
+    },
+    {
+      "epoch": 3.071738083774675,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.090685193977596e-05,
+      "loss": 0.3384,
+      "step": 12760
+    },
+    {
+      "epoch": 3.0729417428984114,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.085825033215539e-05,
+      "loss": 0.3497,
+      "step": 12765
+    },
+    {
+      "epoch": 3.0741454020221473,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.08097001632409e-05,
+      "loss": 0.3515,
+      "step": 12770
+    },
+    {
+      "epoch": 3.0753490611458836,
+      "grad_norm": 1.2421875,
+      "learning_rate": 7.076120147660461e-05,
+      "loss": 0.3463,
+      "step": 12775
+    },
+    {
+      "epoch": 3.0765527202696195,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.07127543157726e-05,
+      "loss": 0.3857,
+      "step": 12780
+    },
+    {
+      "epoch": 3.077756379393356,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.066435872422444e-05,
+      "loss": 0.3332,
+      "step": 12785
+    },
+    {
+      "epoch": 3.078960038517092,
+      "grad_norm": 0.9765625,
+      "learning_rate": 7.06160147453936e-05,
+      "loss": 0.3358,
+      "step": 12790
+    },
+    {
+      "epoch": 3.080163697640828,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.05677224226672e-05,
+      "loss": 0.3659,
+      "step": 12795
+    },
+    {
+      "epoch": 3.0813673567645643,
+      "grad_norm": 1.1015625,
+      "learning_rate": 7.051948179938598e-05,
+      "loss": 0.3594,
+      "step": 12800
+    },
+    {
+      "epoch": 3.0825710158883006,
+      "grad_norm": 0.95703125,
+      "learning_rate": 7.047129291884427e-05,
+      "loss": 0.3394,
+      "step": 12805
+    },
+    {
+      "epoch": 3.0837746750120365,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.042315582429e-05,
+      "loss": 0.3686,
+      "step": 12810
+    },
+    {
+      "epoch": 3.084978334135773,
+      "grad_norm": 1.1953125,
+      "learning_rate": 7.037507055892462e-05,
+      "loss": 0.3873,
+      "step": 12815
+    },
+    {
+      "epoch": 3.0861819932595087,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.0327037165903e-05,
+      "loss": 0.326,
+      "step": 12820
+    },
+    {
+      "epoch": 3.087385652383245,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.027905568833354e-05,
+      "loss": 0.3831,
+      "step": 12825
+    },
+    {
+      "epoch": 3.0885893115069813,
+      "grad_norm": 0.953125,
+      "learning_rate": 7.023112616927801e-05,
+      "loss": 0.3484,
+      "step": 12830
+    },
+    {
+      "epoch": 3.0897929706307172,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.018324865175151e-05,
+      "loss": 0.3124,
+      "step": 12835
+    },
+    {
+      "epoch": 3.0909966297544536,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.013542317872258e-05,
+      "loss": 0.3639,
+      "step": 12840
+    },
+    {
+      "epoch": 3.09220028887819,
+      "grad_norm": 1.0,
+      "learning_rate": 7.008764979311295e-05,
+      "loss": 0.3547,
+      "step": 12845
+    },
+    {
+      "epoch": 3.0934039480019258,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.003992853779762e-05,
+      "loss": 0.323,
+      "step": 12850
+    },
+    {
+      "epoch": 3.094607607125662,
+      "grad_norm": 0.95703125,
+      "learning_rate": 6.999225945560483e-05,
+      "loss": 0.3369,
+      "step": 12855
+    },
+    {
+      "epoch": 3.095811266249398,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.994464258931603e-05,
+      "loss": 0.3301,
+      "step": 12860
+    },
+    {
+      "epoch": 3.0970149253731343,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.989707798166569e-05,
+      "loss": 0.3101,
+      "step": 12865
+    },
+    {
+      "epoch": 3.0982185844968706,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.984956567534152e-05,
+      "loss": 0.354,
+      "step": 12870
+    },
+    {
+      "epoch": 3.0994222436206065,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.980210571298419e-05,
+      "loss": 0.3603,
+      "step": 12875
+    },
+    {
+      "epoch": 3.100625902744343,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.975469813718746e-05,
+      "loss": 0.3262,
+      "step": 12880
+    },
+    {
+      "epoch": 3.101829561868079,
+      "grad_norm": 1.0,
+      "learning_rate": 6.9707342990498e-05,
+      "loss": 0.345,
+      "step": 12885
+    },
+    {
+      "epoch": 3.103033220991815,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.966004031541552e-05,
+      "loss": 0.3723,
+      "step": 12890
+    },
+    {
+      "epoch": 3.1042368801155513,
+      "grad_norm": 1.0,
+      "learning_rate": 6.961279015439258e-05,
+      "loss": 0.3726,
+      "step": 12895
+    },
+    {
+      "epoch": 3.1054405392392876,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.95655925498346e-05,
+      "loss": 0.3356,
+      "step": 12900
+    },
+    {
+      "epoch": 3.1066441983630235,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.951844754409986e-05,
+      "loss": 0.3663,
+      "step": 12905
+    },
+    {
+      "epoch": 3.10784785748676,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.947135517949943e-05,
+      "loss": 0.3457,
+      "step": 12910
+    },
+    {
+      "epoch": 3.1090515166104957,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.942431549829713e-05,
+      "loss": 0.3421,
+      "step": 12915
+    },
+    {
+      "epoch": 3.110255175734232,
+      "grad_norm": 0.921875,
+      "learning_rate": 6.93773285427095e-05,
+      "loss": 0.3445,
+      "step": 12920
+    },
+    {
+      "epoch": 3.1114588348579684,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.933039435490577e-05,
+      "loss": 0.3413,
+      "step": 12925
+    },
+    {
+      "epoch": 3.1126624939817042,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.928351297700779e-05,
+      "loss": 0.3391,
+      "step": 12930
+    },
+    {
+      "epoch": 3.1138661531054406,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.923668445109006e-05,
+      "loss": 0.3242,
+      "step": 12935
+    },
+    {
+      "epoch": 3.115069812229177,
+      "grad_norm": 0.91015625,
+      "learning_rate": 6.918990881917961e-05,
+      "loss": 0.341,
+      "step": 12940
+    },
+    {
+      "epoch": 3.1162734713529128,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.914318612325595e-05,
+      "loss": 0.3464,
+      "step": 12945
+    },
+    {
+      "epoch": 3.117477130476649,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.909651640525119e-05,
+      "loss": 0.339,
+      "step": 12950
+    },
+    {
+      "epoch": 3.118680789600385,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.904989970704986e-05,
+      "loss": 0.3351,
+      "step": 12955
+    },
+    {
+      "epoch": 3.1198844487241213,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.900333607048884e-05,
+      "loss": 0.3555,
+      "step": 12960
+    },
+    {
+      "epoch": 3.1210881078478576,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.895682553735746e-05,
+      "loss": 0.335,
+      "step": 12965
+    },
+    {
+      "epoch": 3.1222917669715935,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.891036814939738e-05,
+      "loss": 0.3804,
+      "step": 12970
+    },
+    {
+      "epoch": 3.12349542609533,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.886396394830252e-05,
+      "loss": 0.363,
+      "step": 12975
+    },
+    {
+      "epoch": 3.124699085219066,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.881761297571912e-05,
+      "loss": 0.3495,
+      "step": 12980
+    },
+    {
+      "epoch": 3.125902744342802,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.877131527324568e-05,
+      "loss": 0.3524,
+      "step": 12985
+    },
+    {
+      "epoch": 3.1271064034665383,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.872507088243273e-05,
+      "loss": 0.3439,
+      "step": 12990
+    },
+    {
+      "epoch": 3.128310062590274,
+      "grad_norm": 0.91015625,
+      "learning_rate": 6.867887984478315e-05,
+      "loss": 0.3211,
+      "step": 12995
+    },
+    {
+      "epoch": 3.1295137217140105,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.863274220175183e-05,
+      "loss": 0.337,
+      "step": 13000
+    },
+    {
+      "epoch": 3.1295137217140105,
+      "eval_loss": 0.34540191292762756,
+      "eval_runtime": 2.36,
+      "eval_samples_per_second": 84.745,
+      "eval_steps_per_second": 84.745,
+      "step": 13000
+    },
+    {
+      "epoch": 3.130717380837747,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.858665799474575e-05,
+      "loss": 0.3674,
+      "step": 13005
+    },
+    {
+      "epoch": 3.1319210399614827,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.854062726512393e-05,
+      "loss": 0.3202,
+      "step": 13010
+    },
+    {
+      "epoch": 3.133124699085219,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.849465005419748e-05,
+      "loss": 0.34,
+      "step": 13015
+    },
+    {
+      "epoch": 3.1343283582089554,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.844872640322934e-05,
+      "loss": 0.355,
+      "step": 13020
+    },
+    {
+      "epoch": 3.1355320173326913,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.84028563534345e-05,
+      "loss": 0.3417,
+      "step": 13025
+    },
+    {
+      "epoch": 3.1367356764564276,
+      "grad_norm": 0.94921875,
+      "learning_rate": 6.835703994597976e-05,
+      "loss": 0.3517,
+      "step": 13030
+    },
+    {
+      "epoch": 3.137939335580164,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.831127722198384e-05,
+      "loss": 0.3334,
+      "step": 13035
+    },
+    {
+      "epoch": 3.1391429947039,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.826556822251721e-05,
+      "loss": 0.3414,
+      "step": 13040
+    },
+    {
+      "epoch": 3.140346653827636,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.821991298860223e-05,
+      "loss": 0.3483,
+      "step": 13045
+    },
+    {
+      "epoch": 3.141550312951372,
+      "grad_norm": 0.9296875,
+      "learning_rate": 6.817431156121292e-05,
+      "loss": 0.3474,
+      "step": 13050
+    },
+    {
+      "epoch": 3.1427539720751083,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.812876398127501e-05,
+      "loss": 0.3488,
+      "step": 13055
+    },
+    {
+      "epoch": 3.1439576311988446,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.808327028966598e-05,
+      "loss": 0.3364,
+      "step": 13060
+    },
+    {
+      "epoch": 3.1451612903225805,
+      "grad_norm": 0.8828125,
+      "learning_rate": 6.80378305272149e-05,
+      "loss": 0.3505,
+      "step": 13065
+    },
+    {
+      "epoch": 3.146364949446317,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.799244473470238e-05,
+      "loss": 0.3437,
+      "step": 13070
+    },
+    {
+      "epoch": 3.147568608570053,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.794711295286068e-05,
+      "loss": 0.3552,
+      "step": 13075
+    },
+    {
+      "epoch": 3.148772267693789,
+      "grad_norm": 0.9375,
+      "learning_rate": 6.790183522237358e-05,
+      "loss": 0.3362,
+      "step": 13080
+    },
+    {
+      "epoch": 3.1499759268175254,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.785661158387628e-05,
+      "loss": 0.355,
+      "step": 13085
+    },
+    {
+      "epoch": 3.1511795859412612,
+      "grad_norm": 0.94921875,
+      "learning_rate": 6.781144207795555e-05,
+      "loss": 0.371,
+      "step": 13090
+    },
+    {
+      "epoch": 3.1523832450649976,
+      "grad_norm": 0.89453125,
+      "learning_rate": 6.776632674514947e-05,
+      "loss": 0.3328,
+      "step": 13095
+    },
+    {
+      "epoch": 3.153586904188734,
+      "grad_norm": 0.94140625,
+      "learning_rate": 6.772126562594753e-05,
+      "loss": 0.329,
+      "step": 13100
+    },
+    {
+      "epoch": 3.1547905633124698,
+      "grad_norm": 0.921875,
+      "learning_rate": 6.767625876079065e-05,
+      "loss": 0.3319,
+      "step": 13105
+    },
+    {
+      "epoch": 3.155994222436206,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.763130619007091e-05,
+      "loss": 0.3534,
+      "step": 13110
+    },
+    {
+      "epoch": 3.1571978815599424,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.758640795413177e-05,
+      "loss": 0.3557,
+      "step": 13115
+    },
+    {
+      "epoch": 3.1584015406836783,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.754156409326787e-05,
+      "loss": 0.3297,
+      "step": 13120
+    },
+    {
+      "epoch": 3.1596051998074146,
+      "grad_norm": 1.0,
+      "learning_rate": 6.749677464772514e-05,
+      "loss": 0.3399,
+      "step": 13125
+    },
+    {
+      "epoch": 3.1608088589311505,
+      "grad_norm": 0.921875,
+      "learning_rate": 6.745203965770052e-05,
+      "loss": 0.367,
+      "step": 13130
+    },
+    {
+      "epoch": 3.162012518054887,
+      "grad_norm": 0.94140625,
+      "learning_rate": 6.740735916334226e-05,
+      "loss": 0.3482,
+      "step": 13135
+    },
+    {
+      "epoch": 3.163216177178623,
+      "grad_norm": 1.125,
+      "learning_rate": 6.736273320474956e-05,
+      "loss": 0.3678,
+      "step": 13140
+    },
+    {
+      "epoch": 3.164419836302359,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.731816182197273e-05,
+      "loss": 0.3366,
+      "step": 13145
+    },
+    {
+      "epoch": 3.1656234954260953,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.72736450550131e-05,
+      "loss": 0.321,
+      "step": 13150
+    },
+    {
+      "epoch": 3.1668271545498317,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.722918294382298e-05,
+      "loss": 0.3596,
+      "step": 13155
+    },
+    {
+      "epoch": 3.1680308136735675,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.718477552830562e-05,
+      "loss": 0.3759,
+      "step": 13160
+    },
+    {
+      "epoch": 3.169234472797304,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.71404228483152e-05,
+      "loss": 0.3124,
+      "step": 13165
+    },
+    {
+      "epoch": 3.17043813192104,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.709612494365679e-05,
+      "loss": 0.3314,
+      "step": 13170
+    },
+    {
+      "epoch": 3.171641791044776,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.705188185408626e-05,
+      "loss": 0.3554,
+      "step": 13175
+    },
+    {
+      "epoch": 3.1728454501685124,
+      "grad_norm": 1.1796875,
+      "learning_rate": 6.700769361931025e-05,
+      "loss": 0.3527,
+      "step": 13180
+    },
+    {
+      "epoch": 3.1740491092922483,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.696356027898635e-05,
+      "loss": 0.3492,
+      "step": 13185
+    },
+    {
+      "epoch": 3.1752527684159846,
+      "grad_norm": 0.9375,
+      "learning_rate": 6.691948187272265e-05,
+      "loss": 0.3435,
+      "step": 13190
+    },
+    {
+      "epoch": 3.176456427539721,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.687545844007805e-05,
+      "loss": 0.3392,
+      "step": 13195
+    },
+    {
+      "epoch": 3.177660086663457,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.683149002056217e-05,
+      "loss": 0.3423,
+      "step": 13200
+    },
+    {
+      "epoch": 3.178863745787193,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.678757665363517e-05,
+      "loss": 0.347,
+      "step": 13205
+    },
+    {
+      "epoch": 3.1800674049109294,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.67437183787078e-05,
+      "loss": 0.3524,
+      "step": 13210
+    },
+    {
+      "epoch": 3.1812710640346653,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.669991523514144e-05,
+      "loss": 0.3186,
+      "step": 13215
+    },
+    {
+      "epoch": 3.1824747231584016,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.665616726224792e-05,
+      "loss": 0.3259,
+      "step": 13220
+    },
+    {
+      "epoch": 3.1836783822821375,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.66124744992896e-05,
+      "loss": 0.3381,
+      "step": 13225
+    },
+    {
+      "epoch": 3.184882041405874,
+      "grad_norm": 1.0,
+      "learning_rate": 6.656883698547926e-05,
+      "loss": 0.3525,
+      "step": 13230
+    },
+    {
+      "epoch": 3.18608570052961,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.652525475998012e-05,
+      "loss": 0.3427,
+      "step": 13235
+    },
+    {
+      "epoch": 3.187289359653346,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.648172786190575e-05,
+      "loss": 0.3805,
+      "step": 13240
+    },
+    {
+      "epoch": 3.1884930187770824,
+      "grad_norm": 0.96484375,
+      "learning_rate": 6.643825633032012e-05,
+      "loss": 0.3594,
+      "step": 13245
+    },
+    {
+      "epoch": 3.1896966779008187,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.639484020423744e-05,
+      "loss": 0.3269,
+      "step": 13250
+    },
+    {
+      "epoch": 3.1909003370245546,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.635147952262225e-05,
+      "loss": 0.3463,
+      "step": 13255
+    },
+    {
+      "epoch": 3.192103996148291,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.630817432438933e-05,
+      "loss": 0.36,
+      "step": 13260
+    },
+    {
+      "epoch": 3.1933076552720268,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.626492464840365e-05,
+      "loss": 0.3656,
+      "step": 13265
+    },
+    {
+      "epoch": 3.194511314395763,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.62217305334803e-05,
+      "loss": 0.3441,
+      "step": 13270
+    },
+    {
+      "epoch": 3.1957149735194994,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.61785920183846e-05,
+      "loss": 0.3781,
+      "step": 13275
+    },
+    {
+      "epoch": 3.1969186326432353,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.613550914183189e-05,
+      "loss": 0.3317,
+      "step": 13280
+    },
+    {
+      "epoch": 3.1981222917669716,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.609248194248763e-05,
+      "loss": 0.3348,
+      "step": 13285
+    },
+    {
+      "epoch": 3.199325950890708,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.604951045896732e-05,
+      "loss": 0.3496,
+      "step": 13290
+    },
+    {
+      "epoch": 3.200529610014444,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.600659472983639e-05,
+      "loss": 0.3352,
+      "step": 13295
+    },
+    {
+      "epoch": 3.20173326913818,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.59637347936103e-05,
+      "loss": 0.3464,
+      "step": 13300
+    },
+    {
+      "epoch": 3.2029369282619164,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.592093068875438e-05,
+      "loss": 0.3564,
+      "step": 13305
+    },
+    {
+      "epoch": 3.2041405873856523,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.587818245368391e-05,
+      "loss": 0.33,
+      "step": 13310
+    },
+    {
+      "epoch": 3.2053442465093887,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.583549012676399e-05,
+      "loss": 0.3433,
+      "step": 13315
+    },
+    {
+      "epoch": 3.2065479056331245,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.579285374630955e-05,
+      "loss": 0.3354,
+      "step": 13320
+    },
+    {
+      "epoch": 3.207751564756861,
+      "grad_norm": 1.0,
+      "learning_rate": 6.575027335058534e-05,
+      "loss": 0.3514,
+      "step": 13325
+    },
+    {
+      "epoch": 3.208955223880597,
+      "grad_norm": 0.921875,
+      "learning_rate": 6.570774897780581e-05,
+      "loss": 0.3386,
+      "step": 13330
+    },
+    {
+      "epoch": 3.210158883004333,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.566528066613518e-05,
+      "loss": 0.3154,
+      "step": 13335
+    },
+    {
+      "epoch": 3.2113625421280694,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.562286845368736e-05,
+      "loss": 0.3296,
+      "step": 13340
+    },
+    {
+      "epoch": 3.2125662012518053,
+      "grad_norm": 0.8984375,
+      "learning_rate": 6.558051237852588e-05,
+      "loss": 0.316,
+      "step": 13345
+    },
+    {
+      "epoch": 3.2137698603755416,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.553821247866388e-05,
+      "loss": 0.3532,
+      "step": 13350
+    },
+    {
+      "epoch": 3.214973519499278,
+      "grad_norm": 0.93359375,
+      "learning_rate": 6.549596879206413e-05,
+      "loss": 0.3434,
+      "step": 13355
+    },
+    {
+      "epoch": 3.2161771786230138,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.545378135663893e-05,
+      "loss": 0.3361,
+      "step": 13360
+    },
+    {
+      "epoch": 3.21738083774675,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.541165021025005e-05,
+      "loss": 0.3409,
+      "step": 13365
+    },
+    {
+      "epoch": 3.2185844968704864,
+      "grad_norm": 1.0,
+      "learning_rate": 6.536957539070884e-05,
+      "loss": 0.362,
+      "step": 13370
+    },
+    {
+      "epoch": 3.2197881559942223,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.532755693577601e-05,
+      "loss": 0.3461,
+      "step": 13375
+    },
+    {
+      "epoch": 3.2209918151179586,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.528559488316172e-05,
+      "loss": 0.3417,
+      "step": 13380
+    },
+    {
+      "epoch": 3.222195474241695,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.524368927052557e-05,
+      "loss": 0.3542,
+      "step": 13385
+    },
+    {
+      "epoch": 3.223399133365431,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.520184013547637e-05,
+      "loss": 0.3373,
+      "step": 13390
+    },
+    {
+      "epoch": 3.224602792489167,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.516004751557233e-05,
+      "loss": 0.3423,
+      "step": 13395
+    },
+    {
+      "epoch": 3.225806451612903,
+      "grad_norm": 0.96484375,
+      "learning_rate": 6.511831144832095e-05,
+      "loss": 0.3369,
+      "step": 13400
+    },
+    {
+      "epoch": 3.2270101107366393,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.507663197117896e-05,
+      "loss": 0.338,
+      "step": 13405
+    },
+    {
+      "epoch": 3.2282137698603757,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.503500912155229e-05,
+      "loss": 0.3621,
+      "step": 13410
+    },
+    {
+      "epoch": 3.2294174289841115,
+      "grad_norm": 0.95703125,
+      "learning_rate": 6.499344293679604e-05,
+      "loss": 0.3393,
+      "step": 13415
+    },
+    {
+      "epoch": 3.230621088107848,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.495193345421451e-05,
+      "loss": 0.3412,
+      "step": 13420
+    },
+    {
+      "epoch": 3.231824747231584,
+      "grad_norm": 1.125,
+      "learning_rate": 6.491048071106106e-05,
+      "loss": 0.3617,
+      "step": 13425
+    },
+    {
+      "epoch": 3.23302840635532,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.48690847445381e-05,
+      "loss": 0.3357,
+      "step": 13430
+    },
+    {
+      "epoch": 3.2342320654790564,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.482774559179719e-05,
+      "loss": 0.3453,
+      "step": 13435
+    },
+    {
+      "epoch": 3.2354357246027927,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.478646328993876e-05,
+      "loss": 0.3393,
+      "step": 13440
+    },
+    {
+      "epoch": 3.2366393837265286,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.474523787601238e-05,
+      "loss": 0.3823,
+      "step": 13445
+    },
+    {
+      "epoch": 3.237843042850265,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.470406938701643e-05,
+      "loss": 0.3263,
+      "step": 13450
+    },
+    {
+      "epoch": 3.239046701974001,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.466295785989828e-05,
+      "loss": 0.3222,
+      "step": 13455
+    },
+    {
+      "epoch": 3.240250361097737,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.462190333155411e-05,
+      "loss": 0.3661,
+      "step": 13460
+    },
+    {
+      "epoch": 3.2414540202214734,
+      "grad_norm": 0.95703125,
+      "learning_rate": 6.458090583882904e-05,
+      "loss": 0.344,
+      "step": 13465
+    },
+    {
+      "epoch": 3.2426576793452093,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.453996541851691e-05,
+      "loss": 0.3596,
+      "step": 13470
+    },
+    {
+      "epoch": 3.2438613384689456,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.44990821073604e-05,
+      "loss": 0.3433,
+      "step": 13475
+    },
+    {
+      "epoch": 3.2450649975926815,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.44582559420509e-05,
+      "loss": 0.3786,
+      "step": 13480
+    },
+    {
+      "epoch": 3.246268656716418,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.441748695922856e-05,
+      "loss": 0.334,
+      "step": 13485
+    },
+    {
+      "epoch": 3.247472315840154,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.437677519548211e-05,
+      "loss": 0.3369,
+      "step": 13490
+    },
+    {
+      "epoch": 3.24867597496389,
+      "grad_norm": 0.9375,
+      "learning_rate": 6.433612068734909e-05,
+      "loss": 0.3204,
+      "step": 13495
+    },
+    {
+      "epoch": 3.2498796340876264,
+      "grad_norm": 3.4375,
+      "learning_rate": 6.42955234713155e-05,
+      "loss": 0.3363,
+      "step": 13500
+    },
+    {
+      "epoch": 3.2498796340876264,
+      "eval_loss": 0.3440934121608734,
+      "eval_runtime": 2.3774,
+      "eval_samples_per_second": 84.125,
+      "eval_steps_per_second": 84.125,
+      "step": 13500
+    },
+    {
+      "epoch": 3.2510832932113627,
+      "grad_norm": 0.8984375,
+      "learning_rate": 6.425498358381598e-05,
+      "loss": 0.3341,
+      "step": 13505
+    },
+    {
+      "epoch": 3.2522869523350986,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.421450106123377e-05,
+      "loss": 0.3631,
+      "step": 13510
+    },
+    {
+      "epoch": 3.253490611458835,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.417407593990055e-05,
+      "loss": 0.3716,
+      "step": 13515
+    },
+    {
+      "epoch": 3.254694270582571,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.41337082560965e-05,
+      "loss": 0.3568,
+      "step": 13520
+    },
+    {
+      "epoch": 3.255897929706307,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.40933980460503e-05,
+      "loss": 0.3578,
+      "step": 13525
+    },
+    {
+      "epoch": 3.2571015888300434,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.405314534593901e-05,
+      "loss": 0.3654,
+      "step": 13530
+    },
+    {
+      "epoch": 3.2583052479537793,
+      "grad_norm": 1.125,
+      "learning_rate": 6.401295019188805e-05,
+      "loss": 0.3528,
+      "step": 13535
+    },
+    {
+      "epoch": 3.2595089070775156,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.397281261997128e-05,
+      "loss": 0.3582,
+      "step": 13540
+    },
+    {
+      "epoch": 3.260712566201252,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.393273266621079e-05,
+      "loss": 0.3392,
+      "step": 13545
+    },
+    {
+      "epoch": 3.261916225324988,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.389271036657702e-05,
+      "loss": 0.3685,
+      "step": 13550
+    },
+    {
+      "epoch": 3.263119884448724,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.385274575698864e-05,
+      "loss": 0.3298,
+      "step": 13555
+    },
+    {
+      "epoch": 3.2643235435724605,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.381283887331255e-05,
+      "loss": 0.3407,
+      "step": 13560
+    },
+    {
+      "epoch": 3.2655272026961963,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.377298975136382e-05,
+      "loss": 0.3612,
+      "step": 13565
+    },
+    {
+      "epoch": 3.2667308618199327,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.373319842690575e-05,
+      "loss": 0.3512,
+      "step": 13570
+    },
+    {
+      "epoch": 3.267934520943669,
+      "grad_norm": 0.890625,
+      "learning_rate": 6.369346493564969e-05,
+      "loss": 0.3321,
+      "step": 13575
+    },
+    {
+      "epoch": 3.269138180067405,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.365378931325513e-05,
+      "loss": 0.3386,
+      "step": 13580
+    },
+    {
+      "epoch": 3.270341839191141,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.361417159532961e-05,
+      "loss": 0.3533,
+      "step": 13585
+    },
+    {
+      "epoch": 3.271545498314877,
+      "grad_norm": 0.95703125,
+      "learning_rate": 6.357461181742875e-05,
+      "loss": 0.343,
+      "step": 13590
+    },
+    {
+      "epoch": 3.2727491574386134,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.353511001505606e-05,
+      "loss": 0.3487,
+      "step": 13595
+    },
+    {
+      "epoch": 3.2739528165623497,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.349566622366313e-05,
+      "loss": 0.3406,
+      "step": 13600
+    },
+    {
+      "epoch": 3.2751564756860856,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.345628047864944e-05,
+      "loss": 0.3566,
+      "step": 13605
+    },
+    {
+      "epoch": 3.276360134809822,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.341695281536238e-05,
+      "loss": 0.3568,
+      "step": 13610
+    },
+    {
+      "epoch": 3.277563793933558,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.337768326909717e-05,
+      "loss": 0.3635,
+      "step": 13615
+    },
+    {
+      "epoch": 3.278767453057294,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.333847187509699e-05,
+      "loss": 0.3674,
+      "step": 13620
+    },
+    {
+      "epoch": 3.2799711121810304,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.32993186685527e-05,
+      "loss": 0.338,
+      "step": 13625
+    },
+    {
+      "epoch": 3.2811747713047663,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.326022368460304e-05,
+      "loss": 0.37,
+      "step": 13630
+    },
+    {
+      "epoch": 3.2823784304285026,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.32211869583344e-05,
+      "loss": 0.3665,
+      "step": 13635
+    },
+    {
+      "epoch": 3.283582089552239,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.318220852478097e-05,
+      "loss": 0.3533,
+      "step": 13640
+    },
+    {
+      "epoch": 3.284785748675975,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.314328841892456e-05,
+      "loss": 0.3204,
+      "step": 13645
+    },
+    {
+      "epoch": 3.285989407799711,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.31044266756947e-05,
+      "loss": 0.3261,
+      "step": 13650
+    },
+    {
+      "epoch": 3.2871930669234475,
+      "grad_norm": 0.98046875,
+      "learning_rate": 6.30656233299685e-05,
+      "loss": 0.3199,
+      "step": 13655
+    },
+    {
+      "epoch": 3.2883967260471834,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.302687841657063e-05,
+      "loss": 0.3339,
+      "step": 13660
+    },
+    {
+      "epoch": 3.2896003851709197,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.298819197027342e-05,
+      "loss": 0.3459,
+      "step": 13665
+    },
+    {
+      "epoch": 3.2908040442946556,
+      "grad_norm": 1.0,
+      "learning_rate": 6.294956402579663e-05,
+      "loss": 0.3342,
+      "step": 13670
+    },
+    {
+      "epoch": 3.292007703418392,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.291099461780751e-05,
+      "loss": 0.3461,
+      "step": 13675
+    },
+    {
+      "epoch": 3.293211362542128,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.287248378092086e-05,
+      "loss": 0.3533,
+      "step": 13680
+    },
+    {
+      "epoch": 3.294415021665864,
+      "grad_norm": 0.9375,
+      "learning_rate": 6.283403154969888e-05,
+      "loss": 0.336,
+      "step": 13685
+    },
+    {
+      "epoch": 3.2956186807896004,
+      "grad_norm": 1.125,
+      "learning_rate": 6.279563795865109e-05,
+      "loss": 0.3498,
+      "step": 13690
+    },
+    {
+      "epoch": 3.2968223399133367,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.275730304223454e-05,
+      "loss": 0.332,
+      "step": 13695
+    },
+    {
+      "epoch": 3.2980259990370726,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.271902683485349e-05,
+      "loss": 0.3471,
+      "step": 13700
+    },
+    {
+      "epoch": 3.299229658160809,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.268080937085956e-05,
+      "loss": 0.3766,
+      "step": 13705
+    },
+    {
+      "epoch": 3.3004333172845453,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.264265068455169e-05,
+      "loss": 0.3362,
+      "step": 13710
+    },
+    {
+      "epoch": 3.301636976408281,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.260455081017597e-05,
+      "loss": 0.3373,
+      "step": 13715
+    },
+    {
+      "epoch": 3.3028406355320175,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.256650978192575e-05,
+      "loss": 0.3598,
+      "step": 13720
+    },
+    {
+      "epoch": 3.3040442946557533,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.252852763394165e-05,
+      "loss": 0.3489,
+      "step": 13725
+    },
+    {
+      "epoch": 3.3052479537794897,
+      "grad_norm": 1.0,
+      "learning_rate": 6.249060440031133e-05,
+      "loss": 0.3449,
+      "step": 13730
+    },
+    {
+      "epoch": 3.306451612903226,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.245274011506963e-05,
+      "loss": 0.3432,
+      "step": 13735
+    },
+    {
+      "epoch": 3.307655272026962,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.241493481219849e-05,
+      "loss": 0.3522,
+      "step": 13740
+    },
+    {
+      "epoch": 3.308858931150698,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.237718852562692e-05,
+      "loss": 0.3388,
+      "step": 13745
+    },
+    {
+      "epoch": 3.310062590274434,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.233950128923092e-05,
+      "loss": 0.3411,
+      "step": 13750
+    },
+    {
+      "epoch": 3.3112662493981704,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.230187313683356e-05,
+      "loss": 0.3284,
+      "step": 13755
+    },
+    {
+      "epoch": 3.3124699085219067,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.226430410220482e-05,
+      "loss": 0.3417,
+      "step": 13760
+    },
+    {
+      "epoch": 3.3136735676456426,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.222679421906162e-05,
+      "loss": 0.3599,
+      "step": 13765
+    },
+    {
+      "epoch": 3.314877226769379,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.218934352106791e-05,
+      "loss": 0.3619,
+      "step": 13770
+    },
+    {
+      "epoch": 3.3160808858931152,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.21519520418344e-05,
+      "loss": 0.3273,
+      "step": 13775
+    },
+    {
+      "epoch": 3.317284545016851,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.211461981491866e-05,
+      "loss": 0.3437,
+      "step": 13780
+    },
+    {
+      "epoch": 3.3184882041405874,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.207734687382516e-05,
+      "loss": 0.3453,
+      "step": 13785
+    },
+    {
+      "epoch": 3.3196918632643238,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.204013325200512e-05,
+      "loss": 0.3571,
+      "step": 13790
+    },
+    {
+      "epoch": 3.3208955223880596,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.200297898285649e-05,
+      "loss": 0.3655,
+      "step": 13795
+    },
+    {
+      "epoch": 3.322099181511796,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.196588409972399e-05,
+      "loss": 0.3471,
+      "step": 13800
+    },
+    {
+      "epoch": 3.323302840635532,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.192884863589907e-05,
+      "loss": 0.344,
+      "step": 13805
+    },
+    {
+      "epoch": 3.324506499759268,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.189187262461979e-05,
+      "loss": 0.3415,
+      "step": 13810
+    },
+    {
+      "epoch": 3.3257101588830045,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.185495609907088e-05,
+      "loss": 0.3612,
+      "step": 13815
+    },
+    {
+      "epoch": 3.3269138180067404,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.181809909238371e-05,
+      "loss": 0.3635,
+      "step": 13820
+    },
+    {
+      "epoch": 3.3281174771304767,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.178130163763621e-05,
+      "loss": 0.3421,
+      "step": 13825
+    },
+    {
+      "epoch": 3.329321136254213,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.174456376785286e-05,
+      "loss": 0.3612,
+      "step": 13830
+    },
+    {
+      "epoch": 3.330524795377949,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.170788551600468e-05,
+      "loss": 0.3492,
+      "step": 13835
+    },
+    {
+      "epoch": 3.331728454501685,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.167126691500914e-05,
+      "loss": 0.3202,
+      "step": 13840
+    },
+    {
+      "epoch": 3.3329321136254215,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.163470799773023e-05,
+      "loss": 0.3267,
+      "step": 13845
+    },
+    {
+      "epoch": 3.3341357727491574,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.159820879697838e-05,
+      "loss": 0.3549,
+      "step": 13850
+    },
+    {
+      "epoch": 3.3353394318728937,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.156176934551037e-05,
+      "loss": 0.3464,
+      "step": 13855
+    },
+    {
+      "epoch": 3.3365430909966296,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.152538967602939e-05,
+      "loss": 0.3401,
+      "step": 13860
+    },
+    {
+      "epoch": 3.337746750120366,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.148906982118501e-05,
+      "loss": 0.3527,
+      "step": 13865
+    },
+    {
+      "epoch": 3.3389504092441022,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.145280981357309e-05,
+      "loss": 0.3363,
+      "step": 13870
+    },
+    {
+      "epoch": 3.340154068367838,
+      "grad_norm": 0.8828125,
+      "learning_rate": 6.141660968573572e-05,
+      "loss": 0.3187,
+      "step": 13875
+    },
+    {
+      "epoch": 3.3413577274915744,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.138046947016134e-05,
+      "loss": 0.3502,
+      "step": 13880
+    },
+    {
+      "epoch": 3.3425613866153103,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.134438919928458e-05,
+      "loss": 0.3357,
+      "step": 13885
+    },
+    {
+      "epoch": 3.3437650457390466,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.130836890548628e-05,
+      "loss": 0.3493,
+      "step": 13890
+    },
+    {
+      "epoch": 3.344968704862783,
+      "grad_norm": 0.92578125,
+      "learning_rate": 6.127240862109344e-05,
+      "loss": 0.3208,
+      "step": 13895
+    },
+    {
+      "epoch": 3.346172363986519,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.123650837837924e-05,
+      "loss": 0.3282,
+      "step": 13900
+    },
+    {
+      "epoch": 3.347376023110255,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.120066820956291e-05,
+      "loss": 0.3359,
+      "step": 13905
+    },
+    {
+      "epoch": 3.3485796822339915,
+      "grad_norm": 1.0,
+      "learning_rate": 6.116488814680984e-05,
+      "loss": 0.3584,
+      "step": 13910
+    },
+    {
+      "epoch": 3.3497833413577274,
+      "grad_norm": 1.0859375,
+      "learning_rate": 6.112916822223142e-05,
+      "loss": 0.3568,
+      "step": 13915
+    },
+    {
+      "epoch": 3.3509870004814637,
+      "grad_norm": 0.875,
+      "learning_rate": 6.109350846788507e-05,
+      "loss": 0.3071,
+      "step": 13920
+    },
+    {
+      "epoch": 3.3521906596052,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.105790891577428e-05,
+      "loss": 0.358,
+      "step": 13925
+    },
+    {
+      "epoch": 3.353394318728936,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.10223695978484e-05,
+      "loss": 0.3639,
+      "step": 13930
+    },
+    {
+      "epoch": 3.354597977852672,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.098689054600282e-05,
+      "loss": 0.357,
+      "step": 13935
+    },
+    {
+      "epoch": 3.355801636976408,
+      "grad_norm": 1.015625,
+      "learning_rate": 6.0951471792078786e-05,
+      "loss": 0.3392,
+      "step": 13940
+    },
+    {
+      "epoch": 3.3570052961001444,
+      "grad_norm": 0.953125,
+      "learning_rate": 6.0916113367863455e-05,
+      "loss": 0.3608,
+      "step": 13945
+    },
+    {
+      "epoch": 3.3582089552238807,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.0880815305089825e-05,
+      "loss": 0.3225,
+      "step": 13950
+    },
+    {
+      "epoch": 3.3594126143476166,
+      "grad_norm": 0.88671875,
+      "learning_rate": 6.084557763543672e-05,
+      "loss": 0.3569,
+      "step": 13955
+    },
+    {
+      "epoch": 3.360616273471353,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.0810400390528806e-05,
+      "loss": 0.3499,
+      "step": 13960
+    },
+    {
+      "epoch": 3.361819932595089,
+      "grad_norm": 0.859375,
+      "learning_rate": 6.077528360193644e-05,
+      "loss": 0.3558,
+      "step": 13965
+    },
+    {
+      "epoch": 3.363023591718825,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.074022730117577e-05,
+      "loss": 0.3178,
+      "step": 13970
+    },
+    {
+      "epoch": 3.3642272508425615,
+      "grad_norm": 0.96484375,
+      "learning_rate": 6.070523151970868e-05,
+      "loss": 0.3462,
+      "step": 13975
+    },
+    {
+      "epoch": 3.365430909966298,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.067029628894271e-05,
+      "loss": 0.3432,
+      "step": 13980
+    },
+    {
+      "epoch": 3.3666345690900337,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.063542164023102e-05,
+      "loss": 0.3437,
+      "step": 13985
+    },
+    {
+      "epoch": 3.36783822821377,
+      "grad_norm": 1.125,
+      "learning_rate": 6.060060760487253e-05,
+      "loss": 0.369,
+      "step": 13990
+    },
+    {
+      "epoch": 3.369041887337506,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.0565854214111596e-05,
+      "loss": 0.3419,
+      "step": 13995
+    },
+    {
+      "epoch": 3.370245546461242,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.053116149913823e-05,
+      "loss": 0.3235,
+      "step": 14000
+    },
+    {
+      "epoch": 3.370245546461242,
+      "eval_loss": 0.34172308444976807,
+      "eval_runtime": 2.3643,
+      "eval_samples_per_second": 84.593,
+      "eval_steps_per_second": 84.593,
+      "step": 14000
+    },
+    {
+      "epoch": 3.3714492055849785,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.049652949108801e-05,
+      "loss": 0.3573,
+      "step": 14005
+    },
+    {
+      "epoch": 3.3726528647087144,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.046195822104201e-05,
+      "loss": 0.3686,
+      "step": 14010
+    },
+    {
+      "epoch": 3.3738565238324507,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.042744772002675e-05,
+      "loss": 0.3464,
+      "step": 14015
+    },
+    {
+      "epoch": 3.3750601829561866,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.03929980190143e-05,
+      "loss": 0.3212,
+      "step": 14020
+    },
+    {
+      "epoch": 3.376263842079923,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.0358609148922093e-05,
+      "loss": 0.3435,
+      "step": 14025
+    },
+    {
+      "epoch": 3.3774675012036592,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.032428114061299e-05,
+      "loss": 0.3659,
+      "step": 14030
+    },
+    {
+      "epoch": 3.378671160327395,
+      "grad_norm": 0.91015625,
+      "learning_rate": 6.029001402489521e-05,
+      "loss": 0.3299,
+      "step": 14035
+    },
+    {
+      "epoch": 3.3798748194511314,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.02558078325224e-05,
+      "loss": 0.3321,
+      "step": 14040
+    },
+    {
+      "epoch": 3.3810784785748678,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.02216625941934e-05,
+      "loss": 0.3348,
+      "step": 14045
+    },
+    {
+      "epoch": 3.3822821376986036,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.018757834055248e-05,
+      "loss": 0.3504,
+      "step": 14050
+    },
+    {
+      "epoch": 3.38348579682234,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.015355510218911e-05,
+      "loss": 0.3268,
+      "step": 14055
+    },
+    {
+      "epoch": 3.3846894559460763,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.0119592909637983e-05,
+      "loss": 0.3582,
+      "step": 14060
+    },
+    {
+      "epoch": 3.385893115069812,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.0085691793379036e-05,
+      "loss": 0.36,
+      "step": 14065
+    },
+    {
+      "epoch": 3.3870967741935485,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.005185178383741e-05,
+      "loss": 0.3344,
+      "step": 14070
+    },
+    {
+      "epoch": 3.3883004333172844,
+      "grad_norm": 1.0546875,
+      "learning_rate": 6.0018072911383366e-05,
+      "loss": 0.3484,
+      "step": 14075
+    },
+    {
+      "epoch": 3.3895040924410207,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.9984355206332325e-05,
+      "loss": 0.3359,
+      "step": 14080
+    },
+    {
+      "epoch": 3.390707751564757,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.99506986989448e-05,
+      "loss": 0.3445,
+      "step": 14085
+    },
+    {
+      "epoch": 3.391911410688493,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.991710341942635e-05,
+      "loss": 0.3547,
+      "step": 14090
+    },
+    {
+      "epoch": 3.393115069812229,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.988356939792764e-05,
+      "loss": 0.3355,
+      "step": 14095
+    },
+    {
+      "epoch": 3.394318728935965,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.9850096664544355e-05,
+      "loss": 0.324,
+      "step": 14100
+    },
+    {
+      "epoch": 3.3955223880597014,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.981668524931712e-05,
+      "loss": 0.348,
+      "step": 14105
+    },
+    {
+      "epoch": 3.3967260471834377,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.978333518223156e-05,
+      "loss": 0.3541,
+      "step": 14110
+    },
+    {
+      "epoch": 3.397929706307174,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.9750046493218294e-05,
+      "loss": 0.3322,
+      "step": 14115
+    },
+    {
+      "epoch": 3.39913336543091,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.9716819212152753e-05,
+      "loss": 0.3251,
+      "step": 14120
+    },
+    {
+      "epoch": 3.4003370245546463,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.968365336885533e-05,
+      "loss": 0.3205,
+      "step": 14125
+    },
+    {
+      "epoch": 3.401540683678382,
+      "grad_norm": 1.171875,
+      "learning_rate": 5.965054899309127e-05,
+      "loss": 0.3562,
+      "step": 14130
+    },
+    {
+      "epoch": 3.4027443428021185,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.9617506114570645e-05,
+      "loss": 0.336,
+      "step": 14135
+    },
+    {
+      "epoch": 3.403948001925855,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.95845247629483e-05,
+      "loss": 0.3466,
+      "step": 14140
+    },
+    {
+      "epoch": 3.4051516610495907,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.9551604967823944e-05,
+      "loss": 0.3474,
+      "step": 14145
+    },
+    {
+      "epoch": 3.406355320173327,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.951874675874197e-05,
+      "loss": 0.3435,
+      "step": 14150
+    },
+    {
+      "epoch": 3.407558979297063,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.948595016519153e-05,
+      "loss": 0.3583,
+      "step": 14155
+    },
+    {
+      "epoch": 3.408762638420799,
+      "grad_norm": 1.0,
+      "learning_rate": 5.945321521660645e-05,
+      "loss": 0.3308,
+      "step": 14160
+    },
+    {
+      "epoch": 3.4099662975445355,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.942054194236528e-05,
+      "loss": 0.3807,
+      "step": 14165
+    },
+    {
+      "epoch": 3.4111699566682714,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.9387930371791156e-05,
+      "loss": 0.3566,
+      "step": 14170
+    },
+    {
+      "epoch": 3.4123736157920077,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5.9355380534151895e-05,
+      "loss": 0.3562,
+      "step": 14175
+    },
+    {
+      "epoch": 3.413577274915744,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.932289245865991e-05,
+      "loss": 0.3342,
+      "step": 14180
+    },
+    {
+      "epoch": 3.41478093403948,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.9290466174472115e-05,
+      "loss": 0.3464,
+      "step": 14185
+    },
+    {
+      "epoch": 3.4159845931632162,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.925810171069006e-05,
+      "loss": 0.3208,
+      "step": 14190
+    },
+    {
+      "epoch": 3.4171882522869526,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.9225799096359764e-05,
+      "loss": 0.3502,
+      "step": 14195
+    },
+    {
+      "epoch": 3.4183919114106884,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.919355836047169e-05,
+      "loss": 0.3542,
+      "step": 14200
+    },
+    {
+      "epoch": 3.4195955705344248,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.916137953196089e-05,
+      "loss": 0.3337,
+      "step": 14205
+    },
+    {
+      "epoch": 3.4207992296581606,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.912926263970675e-05,
+      "loss": 0.3264,
+      "step": 14210
+    },
+    {
+      "epoch": 3.422002888781897,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.9097207712533114e-05,
+      "loss": 0.3548,
+      "step": 14215
+    },
+    {
+      "epoch": 3.4232065479056333,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.9065214779208203e-05,
+      "loss": 0.3434,
+      "step": 14220
+    },
+    {
+      "epoch": 3.424410207029369,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.9033283868444635e-05,
+      "loss": 0.3539,
+      "step": 14225
+    },
+    {
+      "epoch": 3.4256138661531055,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.9001415008899296e-05,
+      "loss": 0.3385,
+      "step": 14230
+    },
+    {
+      "epoch": 3.4268175252768414,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.8969608229173445e-05,
+      "loss": 0.3483,
+      "step": 14235
+    },
+    {
+      "epoch": 3.4280211844005777,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.893786355781259e-05,
+      "loss": 0.3409,
+      "step": 14240
+    },
+    {
+      "epoch": 3.429224843524314,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.890618102330653e-05,
+      "loss": 0.3547,
+      "step": 14245
+    },
+    {
+      "epoch": 3.43042850264805,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.887456065408924e-05,
+      "loss": 0.3513,
+      "step": 14250
+    },
+    {
+      "epoch": 3.431632161771786,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.8843002478539e-05,
+      "loss": 0.3604,
+      "step": 14255
+    },
+    {
+      "epoch": 3.4328358208955225,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.88115065249782e-05,
+      "loss": 0.3776,
+      "step": 14260
+    },
+    {
+      "epoch": 3.4340394800192584,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.8780072821673364e-05,
+      "loss": 0.3429,
+      "step": 14265
+    },
+    {
+      "epoch": 3.4352431391429947,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5.874870139683524e-05,
+      "loss": 0.3288,
+      "step": 14270
+    },
+    {
+      "epoch": 3.436446798266731,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.871739227861864e-05,
+      "loss": 0.3335,
+      "step": 14275
+    },
+    {
+      "epoch": 3.437650457390467,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.8686145495122406e-05,
+      "loss": 0.3797,
+      "step": 14280
+    },
+    {
+      "epoch": 3.4388541165142033,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.86549610743895e-05,
+      "loss": 0.3687,
+      "step": 14285
+    },
+    {
+      "epoch": 3.440057775637939,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.8623839044406915e-05,
+      "loss": 0.3298,
+      "step": 14290
+    },
+    {
+      "epoch": 3.4412614347616755,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.8592779433105607e-05,
+      "loss": 0.3439,
+      "step": 14295
+    },
+    {
+      "epoch": 3.4424650938854118,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.856178226836056e-05,
+      "loss": 0.3586,
+      "step": 14300
+    },
+    {
+      "epoch": 3.4436687530091477,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.8530847577990686e-05,
+      "loss": 0.3458,
+      "step": 14305
+    },
+    {
+      "epoch": 3.444872412132884,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.8499975389758806e-05,
+      "loss": 0.3428,
+      "step": 14310
+    },
+    {
+      "epoch": 3.4460760712566203,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.846916573137175e-05,
+      "loss": 0.3766,
+      "step": 14315
+    },
+    {
+      "epoch": 3.447279730380356,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.843841863048009e-05,
+      "loss": 0.3195,
+      "step": 14320
+    },
+    {
+      "epoch": 3.4484833895040925,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.8407734114678327e-05,
+      "loss": 0.3432,
+      "step": 14325
+    },
+    {
+      "epoch": 3.449687048627829,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.837711221150482e-05,
+      "loss": 0.354,
+      "step": 14330
+    },
+    {
+      "epoch": 3.4508907077515647,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.8346552948441693e-05,
+      "loss": 0.3516,
+      "step": 14335
+    },
+    {
+      "epoch": 3.452094366875301,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.831605635291483e-05,
+      "loss": 0.3556,
+      "step": 14340
+    },
+    {
+      "epoch": 3.453298025999037,
+      "grad_norm": 1.0,
+      "learning_rate": 5.828562245229396e-05,
+      "loss": 0.3436,
+      "step": 14345
+    },
+    {
+      "epoch": 3.4545016851227732,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.8255251273892464e-05,
+      "loss": 0.3349,
+      "step": 14350
+    },
+    {
+      "epoch": 3.4557053442465095,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.822494284496746e-05,
+      "loss": 0.3444,
+      "step": 14355
+    },
+    {
+      "epoch": 3.4569090033702454,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.8194697192719754e-05,
+      "loss": 0.3472,
+      "step": 14360
+    },
+    {
+      "epoch": 3.4581126624939817,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.81645143442938e-05,
+      "loss": 0.3391,
+      "step": 14365
+    },
+    {
+      "epoch": 3.4593163216177176,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.813439432677768e-05,
+      "loss": 0.3546,
+      "step": 14370
+    },
+    {
+      "epoch": 3.460519980741454,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.810433716720315e-05,
+      "loss": 0.3275,
+      "step": 14375
+    },
+    {
+      "epoch": 3.4617236398651903,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.807434289254547e-05,
+      "loss": 0.3259,
+      "step": 14380
+    },
+    {
+      "epoch": 3.462927298988926,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.804441152972352e-05,
+      "loss": 0.3469,
+      "step": 14385
+    },
+    {
+      "epoch": 3.4641309581126625,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.801454310559967e-05,
+      "loss": 0.3466,
+      "step": 14390
+    },
+    {
+      "epoch": 3.465334617236399,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.798473764697987e-05,
+      "loss": 0.3149,
+      "step": 14395
+    },
+    {
+      "epoch": 3.4665382763601347,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.7954995180613475e-05,
+      "loss": 0.3607,
+      "step": 14400
+    },
+    {
+      "epoch": 3.467741935483871,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5.7925315733193386e-05,
+      "loss": 0.3495,
+      "step": 14405
+    },
+    {
+      "epoch": 3.4689455946076073,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.789569933135593e-05,
+      "loss": 0.3367,
+      "step": 14410
+    },
+    {
+      "epoch": 3.470149253731343,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.786614600168079e-05,
+      "loss": 0.3306,
+      "step": 14415
+    },
+    {
+      "epoch": 3.4713529128550795,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.7836655770691126e-05,
+      "loss": 0.3527,
+      "step": 14420
+    },
+    {
+      "epoch": 3.4725565719788154,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.780722866485344e-05,
+      "loss": 0.3611,
+      "step": 14425
+    },
+    {
+      "epoch": 3.4737602311025517,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.7777864710577555e-05,
+      "loss": 0.3543,
+      "step": 14430
+    },
+    {
+      "epoch": 3.474963890226288,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5.7748563934216656e-05,
+      "loss": 0.3584,
+      "step": 14435
+    },
+    {
+      "epoch": 3.476167549350024,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.7719326362067194e-05,
+      "loss": 0.3602,
+      "step": 14440
+    },
+    {
+      "epoch": 3.4773712084737602,
+      "grad_norm": 1.0,
+      "learning_rate": 5.7690152020368934e-05,
+      "loss": 0.3319,
+      "step": 14445
+    },
+    {
+      "epoch": 3.4785748675974966,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.7661040935304824e-05,
+      "loss": 0.3398,
+      "step": 14450
+    },
+    {
+      "epoch": 3.4797785267212324,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.763199313300115e-05,
+      "loss": 0.3396,
+      "step": 14455
+    },
+    {
+      "epoch": 3.4809821858449688,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.760300863952732e-05,
+      "loss": 0.366,
+      "step": 14460
+    },
+    {
+      "epoch": 3.482185844968705,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.757408748089592e-05,
+      "loss": 0.3411,
+      "step": 14465
+    },
+    {
+      "epoch": 3.483389504092441,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.754522968306278e-05,
+      "loss": 0.3489,
+      "step": 14470
+    },
+    {
+      "epoch": 3.4845931632161773,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.7516435271926795e-05,
+      "loss": 0.326,
+      "step": 14475
+    },
+    {
+      "epoch": 3.485796822339913,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.748770427332993e-05,
+      "loss": 0.334,
+      "step": 14480
+    },
+    {
+      "epoch": 3.4870004814636495,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.745903671305735e-05,
+      "loss": 0.3583,
+      "step": 14485
+    },
+    {
+      "epoch": 3.488204140587386,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.7430432616837215e-05,
+      "loss": 0.3238,
+      "step": 14490
+    },
+    {
+      "epoch": 3.4894077997111217,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5.740189201034075e-05,
+      "loss": 0.3388,
+      "step": 14495
+    },
+    {
+      "epoch": 3.490611458834858,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.73734149191822e-05,
+      "loss": 0.3436,
+      "step": 14500
+    },
+    {
+      "epoch": 3.490611458834858,
+      "eval_loss": 0.3381740152835846,
+      "eval_runtime": 2.367,
+      "eval_samples_per_second": 84.495,
+      "eval_steps_per_second": 84.495,
+      "step": 14500
+    },
+    {
+      "epoch": 3.491815117958594,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.734500136891878e-05,
+      "loss": 0.365,
+      "step": 14505
+    },
+    {
+      "epoch": 3.49301877708233,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.7316651385050734e-05,
+      "loss": 0.3437,
+      "step": 14510
+    },
+    {
+      "epoch": 3.4942224362060665,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.728836499302121e-05,
+      "loss": 0.3697,
+      "step": 14515
+    },
+    {
+      "epoch": 3.4954260953298024,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.72601422182163e-05,
+      "loss": 0.3529,
+      "step": 14520
+    },
+    {
+      "epoch": 3.4966297544535387,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.723198308596498e-05,
+      "loss": 0.3764,
+      "step": 14525
+    },
+    {
+      "epoch": 3.497833413577275,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.720388762153916e-05,
+      "loss": 0.3298,
+      "step": 14530
+    },
+    {
+      "epoch": 3.499037072701011,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.7175855850153565e-05,
+      "loss": 0.3494,
+      "step": 14535
+    },
+    {
+      "epoch": 3.5002407318247473,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.7147887796965785e-05,
+      "loss": 0.3273,
+      "step": 14540
+    },
+    {
+      "epoch": 3.5014443909484836,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.7119983487076195e-05,
+      "loss": 0.3568,
+      "step": 14545
+    },
+    {
+      "epoch": 3.5026480500722195,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.7092142945528e-05,
+      "loss": 0.342,
+      "step": 14550
+    },
+    {
+      "epoch": 3.503851709195956,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.706436619730715e-05,
+      "loss": 0.3436,
+      "step": 14555
+    },
+    {
+      "epoch": 3.5050553683196917,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.7036653267342335e-05,
+      "loss": 0.3431,
+      "step": 14560
+    },
+    {
+      "epoch": 3.506259027443428,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.7009004180505e-05,
+      "loss": 0.3597,
+      "step": 14565
+    },
+    {
+      "epoch": 3.5074626865671643,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.698141896160927e-05,
+      "loss": 0.3298,
+      "step": 14570
+    },
+    {
+      "epoch": 3.5086663456909,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.695389763541195e-05,
+      "loss": 0.3421,
+      "step": 14575
+    },
+    {
+      "epoch": 3.5098700048146365,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.692644022661251e-05,
+      "loss": 0.3536,
+      "step": 14580
+    },
+    {
+      "epoch": 3.5110736639383724,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.689904675985308e-05,
+      "loss": 0.3639,
+      "step": 14585
+    },
+    {
+      "epoch": 3.5122773230621087,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.687171725971834e-05,
+      "loss": 0.3403,
+      "step": 14590
+    },
+    {
+      "epoch": 3.513480982185845,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.684445175073566e-05,
+      "loss": 0.3236,
+      "step": 14595
+    },
+    {
+      "epoch": 3.5146846413095814,
+      "grad_norm": 1.0,
+      "learning_rate": 5.681725025737487e-05,
+      "loss": 0.3384,
+      "step": 14600
+    },
+    {
+      "epoch": 3.5158883004333172,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.67901128040484e-05,
+      "loss": 0.3495,
+      "step": 14605
+    },
+    {
+      "epoch": 3.5170919595570536,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.6763039415111256e-05,
+      "loss": 0.3099,
+      "step": 14610
+    },
+    {
+      "epoch": 3.5182956186807894,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.673603011486087e-05,
+      "loss": 0.3555,
+      "step": 14615
+    },
+    {
+      "epoch": 3.5194992778045258,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.6709084927537166e-05,
+      "loss": 0.3316,
+      "step": 14620
+    },
+    {
+      "epoch": 3.520702936928262,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.668220387732259e-05,
+      "loss": 0.3681,
+      "step": 14625
+    },
+    {
+      "epoch": 3.521906596051998,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.665538698834197e-05,
+      "loss": 0.3372,
+      "step": 14630
+    },
+    {
+      "epoch": 3.5231102551757343,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.6628634284662554e-05,
+      "loss": 0.3368,
+      "step": 14635
+    },
+    {
+      "epoch": 3.52431391429947,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.660194579029402e-05,
+      "loss": 0.356,
+      "step": 14640
+    },
+    {
+      "epoch": 3.5255175734232065,
+      "grad_norm": 1.0,
+      "learning_rate": 5.657532152918839e-05,
+      "loss": 0.3419,
+      "step": 14645
+    },
+    {
+      "epoch": 3.526721232546943,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.654876152524003e-05,
+      "loss": 0.3516,
+      "step": 14650
+    },
+    {
+      "epoch": 3.527924891670679,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.6522265802285684e-05,
+      "loss": 0.3724,
+      "step": 14655
+    },
+    {
+      "epoch": 3.529128550794415,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.6495834384104364e-05,
+      "loss": 0.3463,
+      "step": 14660
+    },
+    {
+      "epoch": 3.5303322099181513,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.64694672944174e-05,
+      "loss": 0.3494,
+      "step": 14665
+    },
+    {
+      "epoch": 3.531535869041887,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.6443164556888333e-05,
+      "loss": 0.369,
+      "step": 14670
+    },
+    {
+      "epoch": 3.5327395281656235,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.6416926195123055e-05,
+      "loss": 0.3541,
+      "step": 14675
+    },
+    {
+      "epoch": 3.53394318728936,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.639075223266955e-05,
+      "loss": 0.335,
+      "step": 14680
+    },
+    {
+      "epoch": 3.5351468464130957,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.636464269301814e-05,
+      "loss": 0.3527,
+      "step": 14685
+    },
+    {
+      "epoch": 3.536350505536832,
+      "grad_norm": 1.0,
+      "learning_rate": 5.6338597599601215e-05,
+      "loss": 0.3201,
+      "step": 14690
+    },
+    {
+      "epoch": 3.537554164660568,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.63126169757934e-05,
+      "loss": 0.3461,
+      "step": 14695
+    },
+    {
+      "epoch": 3.5387578237843043,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.62867008449114e-05,
+      "loss": 0.3517,
+      "step": 14700
+    },
+    {
+      "epoch": 3.5399614829080406,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.6260849230214125e-05,
+      "loss": 0.357,
+      "step": 14705
+    },
+    {
+      "epoch": 3.5411651420317765,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.623506215490252e-05,
+      "loss": 0.3412,
+      "step": 14710
+    },
+    {
+      "epoch": 3.542368801155513,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5.620933964211961e-05,
+      "loss": 0.3396,
+      "step": 14715
+    },
+    {
+      "epoch": 3.5435724602792487,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5.618368171495051e-05,
+      "loss": 0.3371,
+      "step": 14720
+    },
+    {
+      "epoch": 3.544776119402985,
+      "grad_norm": 1.125,
+      "learning_rate": 5.615808839642234e-05,
+      "loss": 0.3507,
+      "step": 14725
+    },
+    {
+      "epoch": 3.5459797785267213,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.6132559709504226e-05,
+      "loss": 0.3421,
+      "step": 14730
+    },
+    {
+      "epoch": 3.5471834376504576,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.6107095677107353e-05,
+      "loss": 0.3399,
+      "step": 14735
+    },
+    {
+      "epoch": 3.5483870967741935,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.608169632208482e-05,
+      "loss": 0.3217,
+      "step": 14740
+    },
+    {
+      "epoch": 3.54959075589793,
+      "grad_norm": 1.0,
+      "learning_rate": 5.6056361667231686e-05,
+      "loss": 0.3572,
+      "step": 14745
+    },
+    {
+      "epoch": 3.5507944150216657,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.603109173528498e-05,
+      "loss": 0.3253,
+      "step": 14750
+    },
+    {
+      "epoch": 3.551998074145402,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.600588654892361e-05,
+      "loss": 0.3153,
+      "step": 14755
+    },
+    {
+      "epoch": 3.5532017332691384,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.598074613076839e-05,
+      "loss": 0.3575,
+      "step": 14760
+    },
+    {
+      "epoch": 3.5544053923928742,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.5955670503382e-05,
+      "loss": 0.3395,
+      "step": 14765
+    },
+    {
+      "epoch": 3.5556090515166106,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.5930659689268986e-05,
+      "loss": 0.3422,
+      "step": 14770
+    },
+    {
+      "epoch": 3.5568127106403464,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.590571371087572e-05,
+      "loss": 0.3431,
+      "step": 14775
+    },
+    {
+      "epoch": 3.5580163697640828,
+      "grad_norm": 0.90625,
+      "learning_rate": 5.588083259059036e-05,
+      "loss": 0.3319,
+      "step": 14780
+    },
+    {
+      "epoch": 3.559220028887819,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.585601635074291e-05,
+      "loss": 0.3452,
+      "step": 14785
+    },
+    {
+      "epoch": 3.5604236880115554,
+      "grad_norm": 1.0,
+      "learning_rate": 5.583126501360509e-05,
+      "loss": 0.323,
+      "step": 14790
+    },
+    {
+      "epoch": 3.5616273471352913,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.5806578601390426e-05,
+      "loss": 0.329,
+      "step": 14795
+    },
+    {
+      "epoch": 3.5628310062590276,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.5781957136254145e-05,
+      "loss": 0.3565,
+      "step": 14800
+    },
+    {
+      "epoch": 3.5640346653827635,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.575740064029314e-05,
+      "loss": 0.3354,
+      "step": 14805
+    },
+    {
+      "epoch": 3.5652383245065,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.57329091355461e-05,
+      "loss": 0.3594,
+      "step": 14810
+    },
+    {
+      "epoch": 3.566441983630236,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.5708482643993317e-05,
+      "loss": 0.3818,
+      "step": 14815
+    },
+    {
+      "epoch": 3.567645642753972,
+      "grad_norm": 1.125,
+      "learning_rate": 5.568412118755674e-05,
+      "loss": 0.3591,
+      "step": 14820
+    },
+    {
+      "epoch": 3.5688493018777083,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5.565982478809997e-05,
+      "loss": 0.3148,
+      "step": 14825
+    },
+    {
+      "epoch": 3.570052961001444,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.563559346742823e-05,
+      "loss": 0.356,
+      "step": 14830
+    },
+    {
+      "epoch": 3.5712566201251805,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.561142724728828e-05,
+      "loss": 0.3711,
+      "step": 14835
+    },
+    {
+      "epoch": 3.572460279248917,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.5587326149368534e-05,
+      "loss": 0.3343,
+      "step": 14840
+    },
+    {
+      "epoch": 3.5736639383726527,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.55632901952989e-05,
+      "loss": 0.3275,
+      "step": 14845
+    },
+    {
+      "epoch": 3.574867597496389,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.553931940665085e-05,
+      "loss": 0.3304,
+      "step": 14850
+    },
+    {
+      "epoch": 3.576071256620125,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.551541380493736e-05,
+      "loss": 0.3415,
+      "step": 14855
+    },
+    {
+      "epoch": 3.5772749157438612,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.549157341161291e-05,
+      "loss": 0.3509,
+      "step": 14860
+    },
+    {
+      "epoch": 3.5784785748675976,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.546779824807347e-05,
+      "loss": 0.3565,
+      "step": 14865
+    },
+    {
+      "epoch": 3.579682233991334,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.544408833565643e-05,
+      "loss": 0.3311,
+      "step": 14870
+    },
+    {
+      "epoch": 3.5808858931150698,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.5420443695640675e-05,
+      "loss": 0.337,
+      "step": 14875
+    },
+    {
+      "epoch": 3.582089552238806,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.539686434924647e-05,
+      "loss": 0.3516,
+      "step": 14880
+    },
+    {
+      "epoch": 3.583293211362542,
+      "grad_norm": 1.0,
+      "learning_rate": 5.537335031763545e-05,
+      "loss": 0.3529,
+      "step": 14885
+    },
+    {
+      "epoch": 3.5844968704862783,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.534990162191073e-05,
+      "loss": 0.343,
+      "step": 14890
+    },
+    {
+      "epoch": 3.5857005296100146,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.53265182831167e-05,
+      "loss": 0.3246,
+      "step": 14895
+    },
+    {
+      "epoch": 3.5869041887337505,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.530320032223911e-05,
+      "loss": 0.3618,
+      "step": 14900
+    },
+    {
+      "epoch": 3.588107847857487,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.52799477602051e-05,
+      "loss": 0.3423,
+      "step": 14905
+    },
+    {
+      "epoch": 3.5893115069812227,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.525676061788302e-05,
+      "loss": 0.3092,
+      "step": 14910
+    },
+    {
+      "epoch": 3.590515166104959,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.523363891608258e-05,
+      "loss": 0.373,
+      "step": 14915
+    },
+    {
+      "epoch": 3.5917188252286953,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.521058267555473e-05,
+      "loss": 0.3353,
+      "step": 14920
+    },
+    {
+      "epoch": 3.5929224843524317,
+      "grad_norm": 1.0,
+      "learning_rate": 5.518759191699167e-05,
+      "loss": 0.3508,
+      "step": 14925
+    },
+    {
+      "epoch": 3.5941261434761675,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.5164666661026834e-05,
+      "loss": 0.3361,
+      "step": 14930
+    },
+    {
+      "epoch": 3.595329802599904,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.514180692823489e-05,
+      "loss": 0.3531,
+      "step": 14935
+    },
+    {
+      "epoch": 3.5965334617236397,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.5119012739131675e-05,
+      "loss": 0.3201,
+      "step": 14940
+    },
+    {
+      "epoch": 3.597737120847376,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.50962841141742e-05,
+      "loss": 0.3592,
+      "step": 14945
+    },
+    {
+      "epoch": 3.5989407799711124,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.507362107376066e-05,
+      "loss": 0.3405,
+      "step": 14950
+    },
+    {
+      "epoch": 3.6001444390948483,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.505102363823037e-05,
+      "loss": 0.3519,
+      "step": 14955
+    },
+    {
+      "epoch": 3.6013480982185846,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.502849182786378e-05,
+      "loss": 0.3538,
+      "step": 14960
+    },
+    {
+      "epoch": 3.6025517573423205,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.500602566288243e-05,
+      "loss": 0.3666,
+      "step": 14965
+    },
+    {
+      "epoch": 3.603755416466057,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.498362516344897e-05,
+      "loss": 0.3551,
+      "step": 14970
+    },
+    {
+      "epoch": 3.604959075589793,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.496129034966706e-05,
+      "loss": 0.3422,
+      "step": 14975
+    },
+    {
+      "epoch": 3.606162734713529,
+      "grad_norm": 1.1640625,
+      "learning_rate": 5.4939021241581475e-05,
+      "loss": 0.3526,
+      "step": 14980
+    },
+    {
+      "epoch": 3.6073663938372653,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.491681785917801e-05,
+      "loss": 0.3574,
+      "step": 14985
+    },
+    {
+      "epoch": 3.608570052961001,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5.489468022238343e-05,
+      "loss": 0.3294,
+      "step": 14990
+    },
+    {
+      "epoch": 3.6097737120847375,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.487260835106553e-05,
+      "loss": 0.3211,
+      "step": 14995
+    },
+    {
+      "epoch": 3.610977371208474,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.485060226503311e-05,
+      "loss": 0.3342,
+      "step": 15000
+    },
+    {
+      "epoch": 3.610977371208474,
+      "eval_loss": 0.3394707143306732,
+      "eval_runtime": 2.3647,
+      "eval_samples_per_second": 84.578,
+      "eval_steps_per_second": 84.578,
+      "step": 15000
+    },
+    {
+      "epoch": 3.61218103033221,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.482866198403584e-05,
+      "loss": 0.3518,
+      "step": 15005
+    },
+    {
+      "epoch": 3.613384689455946,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.480678752776443e-05,
+      "loss": 0.3401,
+      "step": 15010
+    },
+    {
+      "epoch": 3.6145883485796824,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.4784978915850455e-05,
+      "loss": 0.3555,
+      "step": 15015
+    },
+    {
+      "epoch": 3.6157920077034182,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.4763236167866436e-05,
+      "loss": 0.3341,
+      "step": 15020
+    },
+    {
+      "epoch": 3.6169956668271546,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.474155930332572e-05,
+      "loss": 0.343,
+      "step": 15025
+    },
+    {
+      "epoch": 3.618199325950891,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.471994834168263e-05,
+      "loss": 0.352,
+      "step": 15030
+    },
+    {
+      "epoch": 3.6194029850746268,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.4698403302332235e-05,
+      "loss": 0.3534,
+      "step": 15035
+    },
+    {
+      "epoch": 3.620606644198363,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.4676924204610496e-05,
+      "loss": 0.3236,
+      "step": 15040
+    },
+    {
+      "epoch": 3.621810303322099,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.46555110677942e-05,
+      "loss": 0.3651,
+      "step": 15045
+    },
+    {
+      "epoch": 3.6230139624458353,
+      "grad_norm": 0.90625,
+      "learning_rate": 5.4634163911100905e-05,
+      "loss": 0.3284,
+      "step": 15050
+    },
+    {
+      "epoch": 3.6242176215695716,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.461288275368897e-05,
+      "loss": 0.3413,
+      "step": 15055
+    },
+    {
+      "epoch": 3.6254212806933075,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.4591667614657547e-05,
+      "loss": 0.3349,
+      "step": 15060
+    },
+    {
+      "epoch": 3.626624939817044,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.4570518513046485e-05,
+      "loss": 0.3589,
+      "step": 15065
+    },
+    {
+      "epoch": 3.62782859894078,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.45494354678364e-05,
+      "loss": 0.3301,
+      "step": 15070
+    },
+    {
+      "epoch": 3.629032258064516,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.452841849794864e-05,
+      "loss": 0.3509,
+      "step": 15075
+    },
+    {
+      "epoch": 3.6302359171882523,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.450746762224524e-05,
+      "loss": 0.3293,
+      "step": 15080
+    },
+    {
+      "epoch": 3.6314395763119887,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.448658285952887e-05,
+      "loss": 0.3473,
+      "step": 15085
+    },
+    {
+      "epoch": 3.6326432354357245,
+      "grad_norm": 1.125,
+      "learning_rate": 5.446576422854293e-05,
+      "loss": 0.3568,
+      "step": 15090
+    },
+    {
+      "epoch": 3.633846894559461,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.444501174797144e-05,
+      "loss": 0.3332,
+      "step": 15095
+    },
+    {
+      "epoch": 3.6350505536831967,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.442432543643907e-05,
+      "loss": 0.3503,
+      "step": 15100
+    },
+    {
+      "epoch": 3.636254212806933,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.440370531251108e-05,
+      "loss": 0.3264,
+      "step": 15105
+    },
+    {
+      "epoch": 3.6374578719306694,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.438315139469334e-05,
+      "loss": 0.3426,
+      "step": 15110
+    },
+    {
+      "epoch": 3.6386615310544053,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.436266370143234e-05,
+      "loss": 0.3338,
+      "step": 15115
+    },
+    {
+      "epoch": 3.6398651901781416,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.4342242251115044e-05,
+      "loss": 0.3471,
+      "step": 15120
+    },
+    {
+      "epoch": 3.6410688493018775,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.432188706206905e-05,
+      "loss": 0.3636,
+      "step": 15125
+    },
+    {
+      "epoch": 3.642272508425614,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.430159815256247e-05,
+      "loss": 0.3257,
+      "step": 15130
+    },
+    {
+      "epoch": 3.64347616754935,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.4281375540803896e-05,
+      "loss": 0.3411,
+      "step": 15135
+    },
+    {
+      "epoch": 3.6446798266730864,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.4261219244942474e-05,
+      "loss": 0.3496,
+      "step": 15140
+    },
+    {
+      "epoch": 3.6458834857968223,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.424112928306778e-05,
+      "loss": 0.329,
+      "step": 15145
+    },
+    {
+      "epoch": 3.6470871449205586,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.422110567320989e-05,
+      "loss": 0.3532,
+      "step": 15150
+    },
+    {
+      "epoch": 3.6482908040442945,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.420114843333934e-05,
+      "loss": 0.337,
+      "step": 15155
+    },
+    {
+      "epoch": 3.649494463168031,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.418125758136708e-05,
+      "loss": 0.3806,
+      "step": 15160
+    },
+    {
+      "epoch": 3.650698122291767,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.416143313514446e-05,
+      "loss": 0.3449,
+      "step": 15165
+    },
+    {
+      "epoch": 3.651901781415503,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.414167511246329e-05,
+      "loss": 0.3387,
+      "step": 15170
+    },
+    {
+      "epoch": 3.6531054405392394,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.412198353105573e-05,
+      "loss": 0.3675,
+      "step": 15175
+    },
+    {
+      "epoch": 3.6543090996629752,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.410235840859431e-05,
+      "loss": 0.3574,
+      "step": 15180
+    },
+    {
+      "epoch": 3.6555127587867116,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.408279976269192e-05,
+      "loss": 0.3633,
+      "step": 15185
+    },
+    {
+      "epoch": 3.656716417910448,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.4063307610901785e-05,
+      "loss": 0.3423,
+      "step": 15190
+    },
+    {
+      "epoch": 3.6579200770341838,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.404388197071745e-05,
+      "loss": 0.3425,
+      "step": 15195
+    },
+    {
+      "epoch": 3.65912373615792,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.402452285957283e-05,
+      "loss": 0.3431,
+      "step": 15200
+    },
+    {
+      "epoch": 3.660327395281656,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.400523029484202e-05,
+      "loss": 0.3677,
+      "step": 15205
+    },
+    {
+      "epoch": 3.6615310544053923,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.398600429383948e-05,
+      "loss": 0.3514,
+      "step": 15210
+    },
+    {
+      "epoch": 3.6627347135291286,
+      "grad_norm": 1.0,
+      "learning_rate": 5.396684487381991e-05,
+      "loss": 0.3557,
+      "step": 15215
+    },
+    {
+      "epoch": 3.663938372652865,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.394775205197822e-05,
+      "loss": 0.3363,
+      "step": 15220
+    },
+    {
+      "epoch": 3.665142031776601,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.3928725845449605e-05,
+      "loss": 0.3386,
+      "step": 15225
+    },
+    {
+      "epoch": 3.666345690900337,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5.390976627130945e-05,
+      "loss": 0.338,
+      "step": 15230
+    },
+    {
+      "epoch": 3.667549350024073,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.389087334657332e-05,
+      "loss": 0.343,
+      "step": 15235
+    },
+    {
+      "epoch": 3.6687530091478093,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.3872047088197005e-05,
+      "loss": 0.3591,
+      "step": 15240
+    },
+    {
+      "epoch": 3.6699566682715457,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.385328751307644e-05,
+      "loss": 0.3547,
+      "step": 15245
+    },
+    {
+      "epoch": 3.6711603273952815,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.38345946380477e-05,
+      "loss": 0.3399,
+      "step": 15250
+    },
+    {
+      "epoch": 3.672363986519018,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.381596847988701e-05,
+      "loss": 0.3389,
+      "step": 15255
+    },
+    {
+      "epoch": 3.6735676456427537,
+      "grad_norm": 1.0,
+      "learning_rate": 5.379740905531075e-05,
+      "loss": 0.3352,
+      "step": 15260
+    },
+    {
+      "epoch": 3.67477130476649,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.377891638097538e-05,
+      "loss": 0.3545,
+      "step": 15265
+    },
+    {
+      "epoch": 3.6759749638902264,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.376049047347745e-05,
+      "loss": 0.3365,
+      "step": 15270
+    },
+    {
+      "epoch": 3.6771786230139627,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.3742131349353575e-05,
+      "loss": 0.3588,
+      "step": 15275
+    },
+    {
+      "epoch": 3.6783822821376986,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.372383902508049e-05,
+      "loss": 0.3671,
+      "step": 15280
+    },
+    {
+      "epoch": 3.679585941261435,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.370561351707494e-05,
+      "loss": 0.3786,
+      "step": 15285
+    },
+    {
+      "epoch": 3.680789600385171,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.368745484169369e-05,
+      "loss": 0.3296,
+      "step": 15290
+    },
+    {
+      "epoch": 3.681993259508907,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.366936301523356e-05,
+      "loss": 0.3619,
+      "step": 15295
+    },
+    {
+      "epoch": 3.6831969186326434,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.365133805393134e-05,
+      "loss": 0.3617,
+      "step": 15300
+    },
+    {
+      "epoch": 3.6844005777563793,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.3633379973963846e-05,
+      "loss": 0.34,
+      "step": 15305
+    },
+    {
+      "epoch": 3.6856042368801156,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.361548879144786e-05,
+      "loss": 0.3475,
+      "step": 15310
+    },
+    {
+      "epoch": 3.6868078960038515,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.3597664522440115e-05,
+      "loss": 0.3509,
+      "step": 15315
+    },
+    {
+      "epoch": 3.688011555127588,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.357990718293728e-05,
+      "loss": 0.3417,
+      "step": 15320
+    },
+    {
+      "epoch": 3.689215214251324,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.3562216788876e-05,
+      "loss": 0.3435,
+      "step": 15325
+    },
+    {
+      "epoch": 3.69041887337506,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.3544593356132804e-05,
+      "loss": 0.3141,
+      "step": 15330
+    },
+    {
+      "epoch": 3.6916225324987963,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.352703690052413e-05,
+      "loss": 0.349,
+      "step": 15335
+    },
+    {
+      "epoch": 3.6928261916225322,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.3509547437806294e-05,
+      "loss": 0.3408,
+      "step": 15340
+    },
+    {
+      "epoch": 3.6940298507462686,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.349212498367553e-05,
+      "loss": 0.3223,
+      "step": 15345
+    },
+    {
+      "epoch": 3.695233509870005,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.34747695537679e-05,
+      "loss": 0.3268,
+      "step": 15350
+    },
+    {
+      "epoch": 3.696437168993741,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.3457481163659324e-05,
+      "loss": 0.3554,
+      "step": 15355
+    },
+    {
+      "epoch": 3.697640828117477,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.344025982886554e-05,
+      "loss": 0.3465,
+      "step": 15360
+    },
+    {
+      "epoch": 3.6988444872412134,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.342310556484213e-05,
+      "loss": 0.356,
+      "step": 15365
+    },
+    {
+      "epoch": 3.7000481463649493,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.34060183869845e-05,
+      "loss": 0.3274,
+      "step": 15370
+    },
+    {
+      "epoch": 3.7012518054886856,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.338899831062777e-05,
+      "loss": 0.3492,
+      "step": 15375
+    },
+    {
+      "epoch": 3.702455464612422,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.337204535104692e-05,
+      "loss": 0.3457,
+      "step": 15380
+    },
+    {
+      "epoch": 3.703659123736158,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.335515952345666e-05,
+      "loss": 0.3204,
+      "step": 15385
+    },
+    {
+      "epoch": 3.704862782859894,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.333834084301146e-05,
+      "loss": 0.3522,
+      "step": 15390
+    },
+    {
+      "epoch": 3.70606644198363,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.33215893248055e-05,
+      "loss": 0.3335,
+      "step": 15395
+    },
+    {
+      "epoch": 3.7072701011073663,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.3304904983872756e-05,
+      "loss": 0.3494,
+      "step": 15400
+    },
+    {
+      "epoch": 3.7084737602311026,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.3288287835186826e-05,
+      "loss": 0.3287,
+      "step": 15405
+    },
+    {
+      "epoch": 3.709677419354839,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.327173789366104e-05,
+      "loss": 0.3796,
+      "step": 15410
+    },
+    {
+      "epoch": 3.710881078478575,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.325525517414845e-05,
+      "loss": 0.3433,
+      "step": 15415
+    },
+    {
+      "epoch": 3.712084737602311,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.323883969144172e-05,
+      "loss": 0.3313,
+      "step": 15420
+    },
+    {
+      "epoch": 3.713288396726047,
+      "grad_norm": 1.0,
+      "learning_rate": 5.322249146027321e-05,
+      "loss": 0.3512,
+      "step": 15425
+    },
+    {
+      "epoch": 3.7144920558497834,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.320621049531491e-05,
+      "loss": 0.3135,
+      "step": 15430
+    },
+    {
+      "epoch": 3.7156957149735197,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.318999681117844e-05,
+      "loss": 0.3512,
+      "step": 15435
+    },
+    {
+      "epoch": 3.7168993740972556,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.3173850422415045e-05,
+      "loss": 0.3635,
+      "step": 15440
+    },
+    {
+      "epoch": 3.718103033220992,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.3157771343515555e-05,
+      "loss": 0.3522,
+      "step": 15445
+    },
+    {
+      "epoch": 3.7193066923447278,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.314175958891041e-05,
+      "loss": 0.3445,
+      "step": 15450
+    },
+    {
+      "epoch": 3.720510351468464,
+      "grad_norm": 1.203125,
+      "learning_rate": 5.312581517296963e-05,
+      "loss": 0.3667,
+      "step": 15455
+    },
+    {
+      "epoch": 3.7217140105922004,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.3109938110002795e-05,
+      "loss": 0.3653,
+      "step": 15460
+    },
+    {
+      "epoch": 3.7229176697159363,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.309412841425904e-05,
+      "loss": 0.3213,
+      "step": 15465
+    },
+    {
+      "epoch": 3.7241213288396726,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.3078386099927036e-05,
+      "loss": 0.3379,
+      "step": 15470
+    },
+    {
+      "epoch": 3.7253249879634085,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.306271118113498e-05,
+      "loss": 0.3527,
+      "step": 15475
+    },
+    {
+      "epoch": 3.726528647087145,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.30471036719506e-05,
+      "loss": 0.3339,
+      "step": 15480
+    },
+    {
+      "epoch": 3.727732306210881,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.30315635863811e-05,
+      "loss": 0.3222,
+      "step": 15485
+    },
+    {
+      "epoch": 3.7289359653346175,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.301609093837318e-05,
+      "loss": 0.3596,
+      "step": 15490
+    },
+    {
+      "epoch": 3.7301396244583533,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.300068574181305e-05,
+      "loss": 0.3306,
+      "step": 15495
+    },
+    {
+      "epoch": 3.7313432835820897,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.298534801052634e-05,
+      "loss": 0.3437,
+      "step": 15500
+    },
+    {
+      "epoch": 3.7313432835820897,
+      "eval_loss": 0.3347737789154053,
+      "eval_runtime": 2.3604,
+      "eval_samples_per_second": 84.732,
+      "eval_steps_per_second": 84.732,
+      "step": 15500
+    },
+    {
+      "epoch": 3.7325469427058255,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.2970077758278165e-05,
+      "loss": 0.3379,
+      "step": 15505
+    },
+    {
+      "epoch": 3.733750601829562,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.295487499877306e-05,
+      "loss": 0.35,
+      "step": 15510
+    },
+    {
+      "epoch": 3.734954260953298,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.293973974565499e-05,
+      "loss": 0.3394,
+      "step": 15515
+    },
+    {
+      "epoch": 3.736157920077034,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.292467201250733e-05,
+      "loss": 0.3363,
+      "step": 15520
+    },
+    {
+      "epoch": 3.7373615792007704,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.290967181285289e-05,
+      "loss": 0.3265,
+      "step": 15525
+    },
+    {
+      "epoch": 3.7385652383245063,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.289473916015383e-05,
+      "loss": 0.3244,
+      "step": 15530
+    },
+    {
+      "epoch": 3.7397688974482426,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.28798740678117e-05,
+      "loss": 0.3384,
+      "step": 15535
+    },
+    {
+      "epoch": 3.740972556571979,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.286507654916745e-05,
+      "loss": 0.3313,
+      "step": 15540
+    },
+    {
+      "epoch": 3.7421762156957152,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.285034661750135e-05,
+      "loss": 0.3394,
+      "step": 15545
+    },
+    {
+      "epoch": 3.743379874819451,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.2835684286033e-05,
+      "loss": 0.3314,
+      "step": 15550
+    },
+    {
+      "epoch": 3.7445835339431874,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.2821089567921366e-05,
+      "loss": 0.3309,
+      "step": 15555
+    },
+    {
+      "epoch": 3.7457871930669233,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.280656247626473e-05,
+      "loss": 0.3338,
+      "step": 15560
+    },
+    {
+      "epoch": 3.7469908521906596,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.279210302410067e-05,
+      "loss": 0.339,
+      "step": 15565
+    },
+    {
+      "epoch": 3.748194511314396,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5.277771122440604e-05,
+      "loss": 0.3269,
+      "step": 15570
+    },
+    {
+      "epoch": 3.749398170438132,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.276338709009701e-05,
+      "loss": 0.3614,
+      "step": 15575
+    },
+    {
+      "epoch": 3.750601829561868,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.274913063402902e-05,
+      "loss": 0.3604,
+      "step": 15580
+    },
+    {
+      "epoch": 3.751805488685604,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.273494186899675e-05,
+      "loss": 0.3384,
+      "step": 15585
+    },
+    {
+      "epoch": 3.7530091478093404,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.272082080773417e-05,
+      "loss": 0.3522,
+      "step": 15590
+    },
+    {
+      "epoch": 3.7542128069330767,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.270676746291442e-05,
+      "loss": 0.3366,
+      "step": 15595
+    },
+    {
+      "epoch": 3.7554164660568126,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.2692781847149925e-05,
+      "loss": 0.3689,
+      "step": 15600
+    },
+    {
+      "epoch": 3.756620125180549,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.267886397299233e-05,
+      "loss": 0.3324,
+      "step": 15605
+    },
+    {
+      "epoch": 3.7578237843042848,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.2665013852932424e-05,
+      "loss": 0.368,
+      "step": 15610
+    },
+    {
+      "epoch": 3.759027443428021,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.265123149940025e-05,
+      "loss": 0.329,
+      "step": 15615
+    },
+    {
+      "epoch": 3.7602311025517574,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.263751692476501e-05,
+      "loss": 0.3431,
+      "step": 15620
+    },
+    {
+      "epoch": 3.7614347616754937,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.262387014133507e-05,
+      "loss": 0.3233,
+      "step": 15625
+    },
+    {
+      "epoch": 3.7626384207992296,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.2610291161357956e-05,
+      "loss": 0.3568,
+      "step": 15630
+    },
+    {
+      "epoch": 3.763842079922966,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.259677999702035e-05,
+      "loss": 0.346,
+      "step": 15635
+    },
+    {
+      "epoch": 3.765045739046702,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.258333666044809e-05,
+      "loss": 0.344,
+      "step": 15640
+    },
+    {
+      "epoch": 3.766249398170438,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.25699611637061e-05,
+      "loss": 0.3517,
+      "step": 15645
+    },
+    {
+      "epoch": 3.7674530572941745,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.255665351879845e-05,
+      "loss": 0.3419,
+      "step": 15650
+    },
+    {
+      "epoch": 3.7686567164179103,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.254341373766829e-05,
+      "loss": 0.319,
+      "step": 15655
+    },
+    {
+      "epoch": 3.7698603755416467,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.253024183219791e-05,
+      "loss": 0.3361,
+      "step": 15660
+    },
+    {
+      "epoch": 3.7710640346653825,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.251713781420864e-05,
+      "loss": 0.3364,
+      "step": 15665
+    },
+    {
+      "epoch": 3.772267693789119,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.2504101695460894e-05,
+      "loss": 0.3446,
+      "step": 15670
+    },
+    {
+      "epoch": 3.773471352912855,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.249113348765416e-05,
+      "loss": 0.3298,
+      "step": 15675
+    },
+    {
+      "epoch": 3.7746750120365915,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.247823320242697e-05,
+      "loss": 0.3571,
+      "step": 15680
+    },
+    {
+      "epoch": 3.7758786711603274,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.24654008513569e-05,
+      "loss": 0.3535,
+      "step": 15685
+    },
+    {
+      "epoch": 3.7770823302840637,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.245263644596055e-05,
+      "loss": 0.3546,
+      "step": 15690
+    },
+    {
+      "epoch": 3.7782859894077996,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.2439939997693545e-05,
+      "loss": 0.3297,
+      "step": 15695
+    },
+    {
+      "epoch": 3.779489648531536,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.242731151795054e-05,
+      "loss": 0.3396,
+      "step": 15700
+    },
+    {
+      "epoch": 3.7806933076552722,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.2414751018065144e-05,
+      "loss": 0.3238,
+      "step": 15705
+    },
+    {
+      "epoch": 3.781896966779008,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.240225850931003e-05,
+      "loss": 0.3333,
+      "step": 15710
+    },
+    {
+      "epoch": 3.7831006259027444,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.238983400289676e-05,
+      "loss": 0.3391,
+      "step": 15715
+    },
+    {
+      "epoch": 3.7843042850264803,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.237747750997594e-05,
+      "loss": 0.3692,
+      "step": 15720
+    },
+    {
+      "epoch": 3.7855079441502166,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.23651890416371e-05,
+      "loss": 0.3529,
+      "step": 15725
+    },
+    {
+      "epoch": 3.786711603273953,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.235296860890874e-05,
+      "loss": 0.3221,
+      "step": 15730
+    },
+    {
+      "epoch": 3.787915262397689,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.2340816222758284e-05,
+      "loss": 0.3291,
+      "step": 15735
+    },
+    {
+      "epoch": 3.789118921521425,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5.232873189409209e-05,
+      "loss": 0.3229,
+      "step": 15740
+    },
+    {
+      "epoch": 3.790322580645161,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.231671563375543e-05,
+      "loss": 0.3369,
+      "step": 15745
+    },
+    {
+      "epoch": 3.7915262397688974,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.230476745253252e-05,
+      "loss": 0.3619,
+      "step": 15750
+    },
+    {
+      "epoch": 3.7927298988926337,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.229288736114643e-05,
+      "loss": 0.3542,
+      "step": 15755
+    },
+    {
+      "epoch": 3.79393355801637,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.228107537025916e-05,
+      "loss": 0.3703,
+      "step": 15760
+    },
+    {
+      "epoch": 3.795137217140106,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.226933149047157e-05,
+      "loss": 0.3661,
+      "step": 15765
+    },
+    {
+      "epoch": 3.796340876263842,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.2257655732323405e-05,
+      "loss": 0.3275,
+      "step": 15770
+    },
+    {
+      "epoch": 3.797544535387578,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.224604810629327e-05,
+      "loss": 0.3559,
+      "step": 15775
+    },
+    {
+      "epoch": 3.7987481945113144,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.2234508622798604e-05,
+      "loss": 0.3496,
+      "step": 15780
+    },
+    {
+      "epoch": 3.7999518536350507,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.222303729219571e-05,
+      "loss": 0.3414,
+      "step": 15785
+    },
+    {
+      "epoch": 3.8011555127587866,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.221163412477975e-05,
+      "loss": 0.3521,
+      "step": 15790
+    },
+    {
+      "epoch": 3.802359171882523,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.2200299130784665e-05,
+      "loss": 0.3314,
+      "step": 15795
+    },
+    {
+      "epoch": 3.803562831006259,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.218903232038323e-05,
+      "loss": 0.3445,
+      "step": 15800
+    },
+    {
+      "epoch": 3.804766490129995,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.217783370368705e-05,
+      "loss": 0.3408,
+      "step": 15805
+    },
+    {
+      "epoch": 3.8059701492537314,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.2166703290746497e-05,
+      "loss": 0.3299,
+      "step": 15810
+    },
+    {
+      "epoch": 3.8071738083774678,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.215564109155072e-05,
+      "loss": 0.341,
+      "step": 15815
+    },
+    {
+      "epoch": 3.8083774675012037,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.214464711602772e-05,
+      "loss": 0.3629,
+      "step": 15820
+    },
+    {
+      "epoch": 3.80958112662494,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.213372137404418e-05,
+      "loss": 0.3291,
+      "step": 15825
+    },
+    {
+      "epoch": 3.810784785748676,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.212286387540563e-05,
+      "loss": 0.3362,
+      "step": 15830
+    },
+    {
+      "epoch": 3.811988444872412,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.211207462985629e-05,
+      "loss": 0.3777,
+      "step": 15835
+    },
+    {
+      "epoch": 3.8131921039961485,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.2101353647079125e-05,
+      "loss": 0.3549,
+      "step": 15840
+    },
+    {
+      "epoch": 3.8143957631198844,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.2090700936695904e-05,
+      "loss": 0.3499,
+      "step": 15845
+    },
+    {
+      "epoch": 3.8155994222436207,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.2080116508267046e-05,
+      "loss": 0.3195,
+      "step": 15850
+    },
+    {
+      "epoch": 3.8168030813673566,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.2069600371291736e-05,
+      "loss": 0.3662,
+      "step": 15855
+    },
+    {
+      "epoch": 3.818006740491093,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.2059152535207844e-05,
+      "loss": 0.3775,
+      "step": 15860
+    },
+    {
+      "epoch": 3.819210399614829,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.204877300939196e-05,
+      "loss": 0.3277,
+      "step": 15865
+    },
+    {
+      "epoch": 3.820414058738565,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.203846180315938e-05,
+      "loss": 0.342,
+      "step": 15870
+    },
+    {
+      "epoch": 3.8216177178623014,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.202821892576403e-05,
+      "loss": 0.3453,
+      "step": 15875
+    },
+    {
+      "epoch": 3.8228213769860373,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.2018044386398564e-05,
+      "loss": 0.3344,
+      "step": 15880
+    },
+    {
+      "epoch": 3.8240250361097736,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.2007938194194315e-05,
+      "loss": 0.3328,
+      "step": 15885
+    },
+    {
+      "epoch": 3.82522869523351,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.199790035822122e-05,
+      "loss": 0.3652,
+      "step": 15890
+    },
+    {
+      "epoch": 3.8264323543572463,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.1987930887487925e-05,
+      "loss": 0.3357,
+      "step": 15895
+    },
+    {
+      "epoch": 3.827636013480982,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.1978029790941675e-05,
+      "loss": 0.3472,
+      "step": 15900
+    },
+    {
+      "epoch": 3.8288396726047185,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.196819707746839e-05,
+      "loss": 0.3413,
+      "step": 15905
+    },
+    {
+      "epoch": 3.8300433317284543,
+      "grad_norm": 1.0,
+      "learning_rate": 5.195843275589259e-05,
+      "loss": 0.3479,
+      "step": 15910
+    },
+    {
+      "epoch": 3.8312469908521907,
+      "grad_norm": 1.0,
+      "learning_rate": 5.1948736834977444e-05,
+      "loss": 0.3568,
+      "step": 15915
+    },
+    {
+      "epoch": 3.832450649975927,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.193910932342471e-05,
+      "loss": 0.3484,
+      "step": 15920
+    },
+    {
+      "epoch": 3.833654309099663,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.1929550229874755e-05,
+      "loss": 0.3567,
+      "step": 15925
+    },
+    {
+      "epoch": 3.834857968223399,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.192005956290654e-05,
+      "loss": 0.3341,
+      "step": 15930
+    },
+    {
+      "epoch": 3.836061627347135,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.1910637331037625e-05,
+      "loss": 0.3504,
+      "step": 15935
+    },
+    {
+      "epoch": 3.8372652864708714,
+      "grad_norm": 0.94921875,
+      "learning_rate": 5.190128354272415e-05,
+      "loss": 0.3345,
+      "step": 15940
+    },
+    {
+      "epoch": 3.8384689455946077,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.189199820636083e-05,
+      "loss": 0.3462,
+      "step": 15945
+    },
+    {
+      "epoch": 3.839672604718344,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.188278133028095e-05,
+      "loss": 0.3447,
+      "step": 15950
+    },
+    {
+      "epoch": 3.84087626384208,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.187363292275632e-05,
+      "loss": 0.3598,
+      "step": 15955
+    },
+    {
+      "epoch": 3.8420799229658162,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5.186455299199736e-05,
+      "loss": 0.3583,
+      "step": 15960
+    },
+    {
+      "epoch": 3.843283582089552,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.1855541546153e-05,
+      "loss": 0.3579,
+      "step": 15965
+    },
+    {
+      "epoch": 3.8444872412132884,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.18465985933107e-05,
+      "loss": 0.3427,
+      "step": 15970
+    },
+    {
+      "epoch": 3.8456909003370248,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.183772414149647e-05,
+      "loss": 0.3364,
+      "step": 15975
+    },
+    {
+      "epoch": 3.8468945594607606,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.182891819867483e-05,
+      "loss": 0.3399,
+      "step": 15980
+    },
+    {
+      "epoch": 3.848098218584497,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.1820180772748816e-05,
+      "loss": 0.3418,
+      "step": 15985
+    },
+    {
+      "epoch": 3.849301877708233,
+      "grad_norm": 1.0,
+      "learning_rate": 5.181151187155997e-05,
+      "loss": 0.3435,
+      "step": 15990
+    },
+    {
+      "epoch": 3.850505536831969,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.1802911502888375e-05,
+      "loss": 0.3267,
+      "step": 15995
+    },
+    {
+      "epoch": 3.8517091959557055,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.179437967445253e-05,
+      "loss": 0.371,
+      "step": 16000
+    },
+    {
+      "epoch": 3.8517091959557055,
+      "eval_loss": 0.3354093134403229,
+      "eval_runtime": 2.3419,
+      "eval_samples_per_second": 85.399,
+      "eval_steps_per_second": 85.399,
+      "step": 16000
+    },
+    {
+      "epoch": 3.8529128550794414,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.17859163939095e-05,
+      "loss": 0.3552,
+      "step": 16005
+    },
+    {
+      "epoch": 3.8541165142031777,
+      "grad_norm": 1.125,
+      "learning_rate": 5.177752166885478e-05,
+      "loss": 0.345,
+      "step": 16010
+    },
+    {
+      "epoch": 3.8553201733269136,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.1769195506822346e-05,
+      "loss": 0.3512,
+      "step": 16015
+    },
+    {
+      "epoch": 3.85652383245065,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.176093791528467e-05,
+      "loss": 0.3225,
+      "step": 16020
+    },
+    {
+      "epoch": 3.857727491574386,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.175274890165267e-05,
+      "loss": 0.3409,
+      "step": 16025
+    },
+    {
+      "epoch": 3.8589311506981225,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.174462847327567e-05,
+      "loss": 0.3573,
+      "step": 16030
+    },
+    {
+      "epoch": 3.8601348098218584,
+      "grad_norm": 0.96875,
+      "learning_rate": 5.1736576637441525e-05,
+      "loss": 0.3035,
+      "step": 16035
+    },
+    {
+      "epoch": 3.8613384689455947,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.172859340137648e-05,
+      "loss": 0.3489,
+      "step": 16040
+    },
+    {
+      "epoch": 3.8625421280693306,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.172067877224522e-05,
+      "loss": 0.3604,
+      "step": 16045
+    },
+    {
+      "epoch": 3.863745787193067,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.1712832757150844e-05,
+      "loss": 0.3288,
+      "step": 16050
+    },
+    {
+      "epoch": 3.8649494463168033,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.1705055363134914e-05,
+      "loss": 0.3458,
+      "step": 16055
+    },
+    {
+      "epoch": 3.866153105440539,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.169734659717736e-05,
+      "loss": 0.3483,
+      "step": 16060
+    },
+    {
+      "epoch": 3.8673567645642755,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.168970646619655e-05,
+      "loss": 0.3537,
+      "step": 16065
+    },
+    {
+      "epoch": 3.8685604236880113,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.168213497704924e-05,
+      "loss": 0.336,
+      "step": 16070
+    },
+    {
+      "epoch": 3.8697640828117477,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.167463213653062e-05,
+      "loss": 0.3587,
+      "step": 16075
+    },
+    {
+      "epoch": 3.870967741935484,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.16671979513742e-05,
+      "loss": 0.3255,
+      "step": 16080
+    },
+    {
+      "epoch": 3.87217140105922,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.1659832428251913e-05,
+      "loss": 0.3408,
+      "step": 16085
+    },
+    {
+      "epoch": 3.873375060182956,
+      "grad_norm": 1.0,
+      "learning_rate": 5.1652535573774104e-05,
+      "loss": 0.3414,
+      "step": 16090
+    },
+    {
+      "epoch": 3.8745787193066925,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.1645307394489424e-05,
+      "loss": 0.3479,
+      "step": 16095
+    },
+    {
+      "epoch": 3.8757823784304284,
+      "grad_norm": 1.0,
+      "learning_rate": 5.1638147896884934e-05,
+      "loss": 0.3461,
+      "step": 16100
+    },
+    {
+      "epoch": 3.8769860375541647,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.1631057087386055e-05,
+      "loss": 0.3442,
+      "step": 16105
+    },
+    {
+      "epoch": 3.878189696677901,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.162403497235653e-05,
+      "loss": 0.3725,
+      "step": 16110
+    },
+    {
+      "epoch": 3.879393355801637,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.161708155809848e-05,
+      "loss": 0.3272,
+      "step": 16115
+    },
+    {
+      "epoch": 3.8805970149253732,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.1610196850852366e-05,
+      "loss": 0.3591,
+      "step": 16120
+    },
+    {
+      "epoch": 3.881800674049109,
+      "grad_norm": 1.0,
+      "learning_rate": 5.1603380856796994e-05,
+      "loss": 0.3516,
+      "step": 16125
+    },
+    {
+      "epoch": 3.8830043331728454,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.159663358204947e-05,
+      "loss": 0.3509,
+      "step": 16130
+    },
+    {
+      "epoch": 3.8842079922965818,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.1589955032665246e-05,
+      "loss": 0.3404,
+      "step": 16135
+    },
+    {
+      "epoch": 3.8854116514203176,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5.158334521463811e-05,
+      "loss": 0.3288,
+      "step": 16140
+    },
+    {
+      "epoch": 3.886615310544054,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.1576804133900156e-05,
+      "loss": 0.3379,
+      "step": 16145
+    },
+    {
+      "epoch": 3.88781896966779,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.157033179632178e-05,
+      "loss": 0.3431,
+      "step": 16150
+    },
+    {
+      "epoch": 3.889022628791526,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.156392820771166e-05,
+      "loss": 0.3554,
+      "step": 16155
+    },
+    {
+      "epoch": 3.8902262879152625,
+      "grad_norm": 0.9609375,
+      "learning_rate": 5.155759337381686e-05,
+      "loss": 0.3218,
+      "step": 16160
+    },
+    {
+      "epoch": 3.891429947038999,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.155132730032264e-05,
+      "loss": 0.3497,
+      "step": 16165
+    },
+    {
+      "epoch": 3.8926336061627347,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.1545129992852605e-05,
+      "loss": 0.3446,
+      "step": 16170
+    },
+    {
+      "epoch": 3.893837265286471,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.153900145696864e-05,
+      "loss": 0.3709,
+      "step": 16175
+    },
+    {
+      "epoch": 3.895040924410207,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.153294169817088e-05,
+      "loss": 0.3415,
+      "step": 16180
+    },
+    {
+      "epoch": 3.896244583533943,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.1526950721897754e-05,
+      "loss": 0.3492,
+      "step": 16185
+    },
+    {
+      "epoch": 3.8974482426576795,
+      "grad_norm": 1.1796875,
+      "learning_rate": 5.1521028533525996e-05,
+      "loss": 0.3502,
+      "step": 16190
+    },
+    {
+      "epoch": 3.8986519017814154,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.151517513837053e-05,
+      "loss": 0.3349,
+      "step": 16195
+    },
+    {
+      "epoch": 3.8998555609051517,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.1509390541684635e-05,
+      "loss": 0.3307,
+      "step": 16200
+    },
+    {
+      "epoch": 3.9010592200288876,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.150367474865973e-05,
+      "loss": 0.3512,
+      "step": 16205
+    },
+    {
+      "epoch": 3.902262879152624,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.14980277644256e-05,
+      "loss": 0.3342,
+      "step": 16210
+    },
+    {
+      "epoch": 3.9034665382763603,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.14924495940502e-05,
+      "loss": 0.3709,
+      "step": 16215
+    },
+    {
+      "epoch": 3.904670197400096,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.148694024253976e-05,
+      "loss": 0.3364,
+      "step": 16220
+    },
+    {
+      "epoch": 3.9058738565238325,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.148149971483873e-05,
+      "loss": 0.3621,
+      "step": 16225
+    },
+    {
+      "epoch": 3.9070775156475683,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.14761280158298e-05,
+      "loss": 0.3408,
+      "step": 16230
+    },
+    {
+      "epoch": 3.9082811747713047,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.147082515033391e-05,
+      "loss": 0.3505,
+      "step": 16235
+    },
+    {
+      "epoch": 3.909484833895041,
+      "grad_norm": 0.96484375,
+      "learning_rate": 5.146559112311019e-05,
+      "loss": 0.3368,
+      "step": 16240
+    },
+    {
+      "epoch": 3.9106884930187773,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.1460425938855996e-05,
+      "loss": 0.3625,
+      "step": 16245
+    },
+    {
+      "epoch": 3.911892152142513,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.145532960220693e-05,
+      "loss": 0.344,
+      "step": 16250
+    },
+    {
+      "epoch": 3.9130958112662495,
+      "grad_norm": 1.0,
+      "learning_rate": 5.145030211773677e-05,
+      "loss": 0.3412,
+      "step": 16255
+    },
+    {
+      "epoch": 3.9142994703899854,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.144534348995751e-05,
+      "loss": 0.3345,
+      "step": 16260
+    },
+    {
+      "epoch": 3.9155031295137217,
+      "grad_norm": 1.0,
+      "learning_rate": 5.144045372331936e-05,
+      "loss": 0.3764,
+      "step": 16265
+    },
+    {
+      "epoch": 3.916706788637458,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.143563282221074e-05,
+      "loss": 0.363,
+      "step": 16270
+    },
+    {
+      "epoch": 3.917910447761194,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.143088079095822e-05,
+      "loss": 0.3371,
+      "step": 16275
+    },
+    {
+      "epoch": 3.9191141068849302,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.1426197633826586e-05,
+      "loss": 0.3737,
+      "step": 16280
+    },
+    {
+      "epoch": 3.920317766008666,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5.1421583355018846e-05,
+      "loss": 0.3295,
+      "step": 16285
+    },
+    {
+      "epoch": 3.9215214251324024,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.141703795867613e-05,
+      "loss": 0.3657,
+      "step": 16290
+    },
+    {
+      "epoch": 3.9227250842561388,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.14125614488778e-05,
+      "loss": 0.3511,
+      "step": 16295
+    },
+    {
+      "epoch": 3.923928743379875,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.140815382964137e-05,
+      "loss": 0.3348,
+      "step": 16300
+    },
+    {
+      "epoch": 3.925132402503611,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.1403815104922525e-05,
+      "loss": 0.3702,
+      "step": 16305
+    },
+    {
+      "epoch": 3.9263360616273473,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.139954527861513e-05,
+      "loss": 0.3417,
+      "step": 16310
+    },
+    {
+      "epoch": 3.927539720751083,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.139534435455119e-05,
+      "loss": 0.3503,
+      "step": 16315
+    },
+    {
+      "epoch": 3.9287433798748195,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.139121233650094e-05,
+      "loss": 0.3542,
+      "step": 16320
+    },
+    {
+      "epoch": 3.929947038998556,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.1387149228172694e-05,
+      "loss": 0.3381,
+      "step": 16325
+    },
+    {
+      "epoch": 3.9311506981222917,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.138315503321295e-05,
+      "loss": 0.329,
+      "step": 16330
+    },
+    {
+      "epoch": 3.932354357246028,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.1379229755206394e-05,
+      "loss": 0.357,
+      "step": 16335
+    },
+    {
+      "epoch": 3.933558016369764,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.1375373397675805e-05,
+      "loss": 0.3399,
+      "step": 16340
+    },
+    {
+      "epoch": 3.9347616754935,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.137158596408214e-05,
+      "loss": 0.3636,
+      "step": 16345
+    },
+    {
+      "epoch": 3.9359653346172365,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.13678674578245e-05,
+      "loss": 0.3488,
+      "step": 16350
+    },
+    {
+      "epoch": 3.9371689937409724,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.136421788224013e-05,
+      "loss": 0.3518,
+      "step": 16355
+    },
+    {
+      "epoch": 3.9383726528647087,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.1360637240604397e-05,
+      "loss": 0.3338,
+      "step": 16360
+    },
+    {
+      "epoch": 3.9395763119884446,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.135712553613078e-05,
+      "loss": 0.3321,
+      "step": 16365
+    },
+    {
+      "epoch": 3.940779971112181,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.135368277197096e-05,
+      "loss": 0.331,
+      "step": 16370
+    },
+    {
+      "epoch": 3.9419836302359172,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.135030895121467e-05,
+      "loss": 0.3358,
+      "step": 16375
+    },
+    {
+      "epoch": 3.9431872893596536,
+      "grad_norm": 0.95703125,
+      "learning_rate": 5.1347004076889815e-05,
+      "loss": 0.3442,
+      "step": 16380
+    },
+    {
+      "epoch": 3.9443909484833894,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.13437681519624e-05,
+      "loss": 0.3449,
+      "step": 16385
+    },
+    {
+      "epoch": 3.9455946076071258,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.134060117933657e-05,
+      "loss": 0.3439,
+      "step": 16390
+    },
+    {
+      "epoch": 3.9467982667308616,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.133750316185458e-05,
+      "loss": 0.3429,
+      "step": 16395
+    },
+    {
+      "epoch": 3.948001925854598,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.133447410229679e-05,
+      "loss": 0.3405,
+      "step": 16400
+    },
+    {
+      "epoch": 3.9492055849783343,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.1331514003381674e-05,
+      "loss": 0.3476,
+      "step": 16405
+    },
+    {
+      "epoch": 3.95040924410207,
+      "grad_norm": 1.0546875,
+      "learning_rate": 5.132862286776583e-05,
+      "loss": 0.3408,
+      "step": 16410
+    },
+    {
+      "epoch": 3.9516129032258065,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.132580069804395e-05,
+      "loss": 0.3411,
+      "step": 16415
+    },
+    {
+      "epoch": 3.9528165623495424,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.132304749674884e-05,
+      "loss": 0.3416,
+      "step": 16420
+    },
+    {
+      "epoch": 3.9540202214732787,
+      "grad_norm": 1.0703125,
+      "learning_rate": 5.132036326635141e-05,
+      "loss": 0.3396,
+      "step": 16425
+    },
+    {
+      "epoch": 3.955223880597015,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5.1317748009260655e-05,
+      "loss": 0.3345,
+      "step": 16430
+    },
+    {
+      "epoch": 3.9564275397207513,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.1315201727823694e-05,
+      "loss": 0.3565,
+      "step": 16435
+    },
+    {
+      "epoch": 3.957631198844487,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.131272442432574e-05,
+      "loss": 0.3553,
+      "step": 16440
+    },
+    {
+      "epoch": 3.9588348579682235,
+      "grad_norm": 1.109375,
+      "learning_rate": 5.1310316100990064e-05,
+      "loss": 0.3636,
+      "step": 16445
+    },
+    {
+      "epoch": 3.9600385170919594,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.130797675997806e-05,
+      "loss": 0.3387,
+      "step": 16450
+    },
+    {
+      "epoch": 3.9612421762156957,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.130570640338921e-05,
+      "loss": 0.3547,
+      "step": 16455
+    },
+    {
+      "epoch": 3.962445835339432,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5.1303505033261115e-05,
+      "loss": 0.335,
+      "step": 16460
+    },
+    {
+      "epoch": 3.963649494463168,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.1301372651569374e-05,
+      "loss": 0.3568,
+      "step": 16465
+    },
+    {
+      "epoch": 3.9648531535869043,
+      "grad_norm": 0.98828125,
+      "learning_rate": 5.1299309260227776e-05,
+      "loss": 0.3424,
+      "step": 16470
+    },
+    {
+      "epoch": 3.96605681271064,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.129731486108812e-05,
+      "loss": 0.361,
+      "step": 16475
+    },
+    {
+      "epoch": 3.9672604718343765,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.129538945594032e-05,
+      "loss": 0.3611,
+      "step": 16480
+    },
+    {
+      "epoch": 3.968464130958113,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.129353304651238e-05,
+      "loss": 0.3501,
+      "step": 16485
+    },
+    {
+      "epoch": 3.9696677900818487,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.129174563447034e-05,
+      "loss": 0.3535,
+      "step": 16490
+    },
+    {
+      "epoch": 3.970871449205585,
+      "grad_norm": 1.0859375,
+      "learning_rate": 5.129002722141834e-05,
+      "loss": 0.363,
+      "step": 16495
+    },
+    {
+      "epoch": 3.972075108329321,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.128837780889862e-05,
+      "loss": 0.3431,
+      "step": 16500
+    },
+    {
+      "epoch": 3.972075108329321,
+      "eval_loss": 0.3325212895870209,
+      "eval_runtime": 2.3455,
+      "eval_samples_per_second": 85.271,
+      "eval_steps_per_second": 85.271,
+      "step": 16500
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 16616,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.13867765661696e+17,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}