diff --git "a/checkpoints/checkpoint-11000/trainer_state.json" "b/checkpoints/checkpoint-11000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoints/checkpoint-11000/trainer_state.json"
@@ -0,0 +1,7721 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7982004208693129,
+  "eval_steps": 500,
+  "global_step": 11000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0007256367462448298,
+      "grad_norm": 13.1875,
+      "learning_rate": 5e-06,
+      "loss": 3.553,
+      "step": 10
+    },
+    {
+      "epoch": 0.0014512734924896596,
+      "grad_norm": 9.375,
+      "learning_rate": 1e-05,
+      "loss": 3.5658,
+      "step": 20
+    },
+    {
+      "epoch": 0.0021769102387344894,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.5e-05,
+      "loss": 3.4669,
+      "step": 30
+    },
+    {
+      "epoch": 0.0029025469849793192,
+      "grad_norm": 13.25,
+      "learning_rate": 2e-05,
+      "loss": 3.5007,
+      "step": 40
+    },
+    {
+      "epoch": 0.003628183731224149,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.5e-05,
+      "loss": 3.5062,
+      "step": 50
+    },
+    {
+      "epoch": 0.004353820477468979,
+      "grad_norm": 9.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.4178,
+      "step": 60
+    },
+    {
+      "epoch": 0.005079457223713809,
+      "grad_norm": 17.125,
+      "learning_rate": 3.5e-05,
+      "loss": 3.4136,
+      "step": 70
+    },
+    {
+      "epoch": 0.0058050939699586385,
+      "grad_norm": 17.5,
+      "learning_rate": 4e-05,
+      "loss": 3.3792,
+      "step": 80
+    },
+    {
+      "epoch": 0.006530730716203468,
+      "grad_norm": 17.0,
+      "learning_rate": 4.5e-05,
+      "loss": 3.3684,
+      "step": 90
+    },
+    {
+      "epoch": 0.007256367462448298,
+      "grad_norm": 21.875,
+      "learning_rate": 5e-05,
+      "loss": 3.2698,
+      "step": 100
+    },
+    {
+      "epoch": 0.007982004208693128,
+      "grad_norm": 32.5,
+      "learning_rate": 4.9999934086574596e-05,
+      "loss": 3.1523,
+      "step": 110
+    },
+    {
+      "epoch": 0.008707640954937958,
+      "grad_norm": 73.5,
+      "learning_rate": 4.9999736346645943e-05,
+      "loss": 3.0884,
+      "step": 120
+    },
+    {
+      "epoch": 0.009433277701182788,
+      "grad_norm": 34.75,
+      "learning_rate": 4.999940678125673e-05,
+      "loss": 2.9507,
+      "step": 130
+    },
+    {
+      "epoch": 0.010158914447427617,
+      "grad_norm": 47.5,
+      "learning_rate": 4.9998945392144796e-05,
+      "loss": 2.8659,
+      "step": 140
+    },
+    {
+      "epoch": 0.010884551193672447,
+      "grad_norm": 35.25,
+      "learning_rate": 4.999835218174307e-05,
+      "loss": 2.9846,
+      "step": 150
+    },
+    {
+      "epoch": 0.011610187939917277,
+      "grad_norm": 27.5,
+      "learning_rate": 4.99976271531796e-05,
+      "loss": 2.8913,
+      "step": 160
+    },
+    {
+      "epoch": 0.012335824686162107,
+      "grad_norm": 38.5,
+      "learning_rate": 4.9996770310277506e-05,
+      "loss": 2.8982,
+      "step": 170
+    },
+    {
+      "epoch": 0.013061461432406937,
+      "grad_norm": 52.0,
+      "learning_rate": 4.9995781657555e-05,
+      "loss": 2.7165,
+      "step": 180
+    },
+    {
+      "epoch": 0.013787098178651766,
+      "grad_norm": 34.75,
+      "learning_rate": 4.99946612002253e-05,
+      "loss": 2.4166,
+      "step": 190
+    },
+    {
+      "epoch": 0.014512734924896596,
+      "grad_norm": 40.25,
+      "learning_rate": 4.9993408944196676e-05,
+      "loss": 2.2161,
+      "step": 200
+    },
+    {
+      "epoch": 0.015238371671141426,
+      "grad_norm": 39.25,
+      "learning_rate": 4.9992024896072364e-05,
+      "loss": 1.7825,
+      "step": 210
+    },
+    {
+      "epoch": 0.015964008417386256,
+      "grad_norm": 37.25,
+      "learning_rate": 4.999050906315055e-05,
+      "loss": 1.46,
+      "step": 220
+    },
+    {
+      "epoch": 0.016689645163631087,
+      "grad_norm": 13.625,
+      "learning_rate": 4.998886145342434e-05,
+      "loss": 1.3157,
+      "step": 230
+    },
+    {
+      "epoch": 0.017415281909875915,
+      "grad_norm": 10.4375,
+      "learning_rate": 4.9987082075581684e-05,
+      "loss": 1.1408,
+      "step": 240
+    },
+    {
+      "epoch": 0.018140918656120747,
+      "grad_norm": 14.0,
+      "learning_rate": 4.9985170939005386e-05,
+      "loss": 0.8782,
+      "step": 250
+    },
+    {
+      "epoch": 0.018866555402365575,
+      "grad_norm": 22.0,
+      "learning_rate": 4.998312805377302e-05,
+      "loss": 0.9109,
+      "step": 260
+    },
+    {
+      "epoch": 0.019592192148610407,
+      "grad_norm": 12.625,
+      "learning_rate": 4.998095343065685e-05,
+      "loss": 0.7414,
+      "step": 270
+    },
+    {
+      "epoch": 0.020317828894855235,
+      "grad_norm": 15.8125,
+      "learning_rate": 4.997864708112384e-05,
+      "loss": 0.7926,
+      "step": 280
+    },
+    {
+      "epoch": 0.021043465641100066,
+      "grad_norm": 16.25,
+      "learning_rate": 4.997620901733554e-05,
+      "loss": 0.7382,
+      "step": 290
+    },
+    {
+      "epoch": 0.021769102387344894,
+      "grad_norm": 22.0,
+      "learning_rate": 4.997363925214803e-05,
+      "loss": 0.7158,
+      "step": 300
+    },
+    {
+      "epoch": 0.022494739133589726,
+      "grad_norm": 13.375,
+      "learning_rate": 4.9970937799111896e-05,
+      "loss": 0.526,
+      "step": 310
+    },
+    {
+      "epoch": 0.023220375879834554,
+      "grad_norm": 12.9375,
+      "learning_rate": 4.996810467247207e-05,
+      "loss": 0.6137,
+      "step": 320
+    },
+    {
+      "epoch": 0.023946012626079385,
+      "grad_norm": 6.78125,
+      "learning_rate": 4.9965139887167856e-05,
+      "loss": 0.4864,
+      "step": 330
+    },
+    {
+      "epoch": 0.024671649372324213,
+      "grad_norm": 10.25,
+      "learning_rate": 4.996204345883278e-05,
+      "loss": 0.6921,
+      "step": 340
+    },
+    {
+      "epoch": 0.025397286118569045,
+      "grad_norm": 6.3125,
+      "learning_rate": 4.9958815403794546e-05,
+      "loss": 0.5826,
+      "step": 350
+    },
+    {
+      "epoch": 0.026122922864813873,
+      "grad_norm": 10.625,
+      "learning_rate": 4.995545573907492e-05,
+      "loss": 0.5838,
+      "step": 360
+    },
+    {
+      "epoch": 0.026848559611058705,
+      "grad_norm": 17.25,
+      "learning_rate": 4.995196448238966e-05,
+      "loss": 0.5941,
+      "step": 370
+    },
+    {
+      "epoch": 0.027574196357303533,
+      "grad_norm": 9.875,
+      "learning_rate": 4.9948341652148436e-05,
+      "loss": 0.5055,
+      "step": 380
+    },
+    {
+      "epoch": 0.028299833103548364,
+      "grad_norm": 21.25,
+      "learning_rate": 4.994458726745468e-05,
+      "loss": 0.6114,
+      "step": 390
+    },
+    {
+      "epoch": 0.029025469849793192,
+      "grad_norm": 3.375,
+      "learning_rate": 4.9940701348105554e-05,
+      "loss": 0.5372,
+      "step": 400
+    },
+    {
+      "epoch": 0.029751106596038024,
+      "grad_norm": 12.625,
+      "learning_rate": 4.99366839145918e-05,
+      "loss": 0.4815,
+      "step": 410
+    },
+    {
+      "epoch": 0.030476743342282852,
+      "grad_norm": 10.8125,
+      "learning_rate": 4.993253498809762e-05,
+      "loss": 0.4903,
+      "step": 420
+    },
+    {
+      "epoch": 0.031202380088527683,
+      "grad_norm": 9.0,
+      "learning_rate": 4.9928254590500646e-05,
+      "loss": 0.5927,
+      "step": 430
+    },
+    {
+      "epoch": 0.03192801683477251,
+      "grad_norm": 6.71875,
+      "learning_rate": 4.9923842744371707e-05,
+      "loss": 0.4424,
+      "step": 440
+    },
+    {
+      "epoch": 0.03265365358101734,
+      "grad_norm": 5.0,
+      "learning_rate": 4.99192994729748e-05,
+      "loss": 0.362,
+      "step": 450
+    },
+    {
+      "epoch": 0.033379290327262175,
+      "grad_norm": 8.625,
+      "learning_rate": 4.991462480026693e-05,
+      "loss": 0.593,
+      "step": 460
+    },
+    {
+      "epoch": 0.034104927073507,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.9909818750898e-05,
+      "loss": 0.4769,
+      "step": 470
+    },
+    {
+      "epoch": 0.03483056381975183,
+      "grad_norm": 11.125,
+      "learning_rate": 4.990488135021065e-05,
+      "loss": 0.4455,
+      "step": 480
+    },
+    {
+      "epoch": 0.03555620056599666,
+      "grad_norm": 8.375,
+      "learning_rate": 4.989981262424017e-05,
+      "loss": 0.4713,
+      "step": 490
+    },
+    {
+      "epoch": 0.036281837312241494,
+      "grad_norm": 7.75,
+      "learning_rate": 4.989461259971432e-05,
+      "loss": 0.2778,
+      "step": 500
+    },
+    {
+      "epoch": 0.03700747405848632,
+      "grad_norm": 5.65625,
+      "learning_rate": 4.988928130405323e-05,
+      "loss": 0.3604,
+      "step": 510
+    },
+    {
+      "epoch": 0.03773311080473115,
+      "grad_norm": 7.5625,
+      "learning_rate": 4.9883818765369194e-05,
+      "loss": 0.4234,
+      "step": 520
+    },
+    {
+      "epoch": 0.03845874755097598,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.98782250124666e-05,
+      "loss": 0.3531,
+      "step": 530
+    },
+    {
+      "epoch": 0.03918438429722081,
+      "grad_norm": 6.5,
+      "learning_rate": 4.987250007484172e-05,
+      "loss": 0.3281,
+      "step": 540
+    },
+    {
+      "epoch": 0.03991002104346564,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.986664398268256e-05,
+      "loss": 0.3477,
+      "step": 550
+    },
+    {
+      "epoch": 0.04063565778971047,
+      "grad_norm": 6.4375,
+      "learning_rate": 4.986065676686874e-05,
+      "loss": 0.369,
+      "step": 560
+    },
+    {
+      "epoch": 0.0413612945359553,
+      "grad_norm": 8.0625,
+      "learning_rate": 4.98545384589713e-05,
+      "loss": 0.3068,
+      "step": 570
+    },
+    {
+      "epoch": 0.04208693128220013,
+      "grad_norm": 6.28125,
+      "learning_rate": 4.984828909125251e-05,
+      "loss": 0.2926,
+      "step": 580
+    },
+    {
+      "epoch": 0.042812568028444964,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.9841908696665764e-05,
+      "loss": 0.2771,
+      "step": 590
+    },
+    {
+      "epoch": 0.04353820477468979,
+      "grad_norm": 4.15625,
+      "learning_rate": 4.9835397308855344e-05,
+      "loss": 0.3676,
+      "step": 600
+    },
+    {
+      "epoch": 0.04426384152093462,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.9828754962156286e-05,
+      "loss": 0.3307,
+      "step": 610
+    },
+    {
+      "epoch": 0.04498947826717945,
+      "grad_norm": 13.75,
+      "learning_rate": 4.9821981691594175e-05,
+      "loss": 0.3601,
+      "step": 620
+    },
+    {
+      "epoch": 0.04571511501342428,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.981507753288497e-05,
+      "loss": 0.2939,
+      "step": 630
+    },
+    {
+      "epoch": 0.04644075175966911,
+      "grad_norm": 6.1875,
+      "learning_rate": 4.9808042522434814e-05,
+      "loss": 0.293,
+      "step": 640
+    },
+    {
+      "epoch": 0.04716638850591394,
+      "grad_norm": 4.65625,
+      "learning_rate": 4.9800876697339824e-05,
+      "loss": 0.3213,
+      "step": 650
+    },
+    {
+      "epoch": 0.04789202525215877,
+      "grad_norm": 7.65625,
+      "learning_rate": 4.979358009538594e-05,
+      "loss": 0.3903,
+      "step": 660
+    },
+    {
+      "epoch": 0.0486176619984036,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.978615275504869e-05,
+      "loss": 0.2936,
+      "step": 670
+    },
+    {
+      "epoch": 0.04934329874464843,
+      "grad_norm": 8.4375,
+      "learning_rate": 4.977859471549297e-05,
+      "loss": 0.3681,
+      "step": 680
+    },
+    {
+      "epoch": 0.05006893549089326,
+      "grad_norm": 6.28125,
+      "learning_rate": 4.977090601657289e-05,
+      "loss": 0.2591,
+      "step": 690
+    },
+    {
+      "epoch": 0.05079457223713809,
+      "grad_norm": 10.8125,
+      "learning_rate": 4.976308669883153e-05,
+      "loss": 0.3351,
+      "step": 700
+    },
+    {
+      "epoch": 0.05152020898338292,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.975513680350073e-05,
+      "loss": 0.3042,
+      "step": 710
+    },
+    {
+      "epoch": 0.052245845729627746,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.974705637250089e-05,
+      "loss": 0.2396,
+      "step": 720
+    },
+    {
+      "epoch": 0.05297148247587258,
+      "grad_norm": 6.1875,
+      "learning_rate": 4.97388454484407e-05,
+      "loss": 0.2892,
+      "step": 730
+    },
+    {
+      "epoch": 0.05369711922211741,
+      "grad_norm": 3.78125,
+      "learning_rate": 4.973050407461698e-05,
+      "loss": 0.3032,
+      "step": 740
+    },
+    {
+      "epoch": 0.05442275596836224,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.972203229501441e-05,
+      "loss": 0.2522,
+      "step": 750
+    },
+    {
+      "epoch": 0.055148392714607065,
+      "grad_norm": 7.84375,
+      "learning_rate": 4.971343015430532e-05,
+      "loss": 0.2502,
+      "step": 760
+    },
+    {
+      "epoch": 0.0558740294608519,
+      "grad_norm": 10.6875,
+      "learning_rate": 4.970469769784941e-05,
+      "loss": 0.2664,
+      "step": 770
+    },
+    {
+      "epoch": 0.05659966620709673,
+      "grad_norm": 7.90625,
+      "learning_rate": 4.96958349716936e-05,
+      "loss": 0.332,
+      "step": 780
+    },
+    {
+      "epoch": 0.05732530295334156,
+      "grad_norm": 10.75,
+      "learning_rate": 4.968684202257169e-05,
+      "loss": 0.3737,
+      "step": 790
+    },
+    {
+      "epoch": 0.058050939699586385,
+      "grad_norm": 6.5,
+      "learning_rate": 4.967771889790416e-05,
+      "loss": 0.2794,
+      "step": 800
+    },
+    {
+      "epoch": 0.058776576445831216,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.966846564579792e-05,
+      "loss": 0.2375,
+      "step": 810
+    },
+    {
+      "epoch": 0.05950221319207605,
+      "grad_norm": 6.09375,
+      "learning_rate": 4.965908231504607e-05,
+      "loss": 0.365,
+      "step": 820
+    },
+    {
+      "epoch": 0.06022784993832088,
+      "grad_norm": 9.0625,
+      "learning_rate": 4.964956895512759e-05,
+      "loss": 0.3469,
+      "step": 830
+    },
+    {
+      "epoch": 0.060953486684565704,
+      "grad_norm": 5.25,
+      "learning_rate": 4.963992561620714e-05,
+      "loss": 0.2904,
+      "step": 840
+    },
+    {
+      "epoch": 0.061679123430810535,
+      "grad_norm": 5.5,
+      "learning_rate": 4.963015234913475e-05,
+      "loss": 0.2779,
+      "step": 850
+    },
+    {
+      "epoch": 0.06240476017705537,
+      "grad_norm": 5.6875,
+      "learning_rate": 4.96202492054456e-05,
+      "loss": 0.2658,
+      "step": 860
+    },
+    {
+      "epoch": 0.06313039692330019,
+      "grad_norm": 8.8125,
+      "learning_rate": 4.9610216237359684e-05,
+      "loss": 0.3973,
+      "step": 870
+    },
+    {
+      "epoch": 0.06385603366954502,
+      "grad_norm": 10.0,
+      "learning_rate": 4.960005349778159e-05,
+      "loss": 0.3169,
+      "step": 880
+    },
+    {
+      "epoch": 0.06458167041578985,
+      "grad_norm": 11.9375,
+      "learning_rate": 4.95897610403002e-05,
+      "loss": 0.3172,
+      "step": 890
+    },
+    {
+      "epoch": 0.06530730716203469,
+      "grad_norm": 3.0,
+      "learning_rate": 4.95793389191884e-05,
+      "loss": 0.2664,
+      "step": 900
+    },
+    {
+      "epoch": 0.06603294390827952,
+      "grad_norm": 6.90625,
+      "learning_rate": 4.95687871894028e-05,
+      "loss": 0.2507,
+      "step": 910
+    },
+    {
+      "epoch": 0.06675858065452435,
+      "grad_norm": 3.796875,
+      "learning_rate": 4.9558105906583466e-05,
+      "loss": 0.2383,
+      "step": 920
+    },
+    {
+      "epoch": 0.06748421740076918,
+      "grad_norm": 6.28125,
+      "learning_rate": 4.9547295127053586e-05,
+      "loss": 0.3942,
+      "step": 930
+    },
+    {
+      "epoch": 0.068209854147014,
+      "grad_norm": 9.0,
+      "learning_rate": 4.95363549078192e-05,
+      "loss": 0.2427,
+      "step": 940
+    },
+    {
+      "epoch": 0.06893549089325883,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.952528530656889e-05,
+      "loss": 0.2351,
+      "step": 950
+    },
+    {
+      "epoch": 0.06966112763950366,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.9514086381673496e-05,
+      "loss": 0.282,
+      "step": 960
+    },
+    {
+      "epoch": 0.0703867643857485,
+      "grad_norm": 10.0625,
+      "learning_rate": 4.9502758192185774e-05,
+      "loss": 0.2687,
+      "step": 970
+    },
+    {
+      "epoch": 0.07111240113199332,
+      "grad_norm": 9.0625,
+      "learning_rate": 4.949130079784009e-05,
+      "loss": 0.3135,
+      "step": 980
+    },
+    {
+      "epoch": 0.07183803787823816,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.9479714259052143e-05,
+      "loss": 0.3207,
+      "step": 990
+    },
+    {
+      "epoch": 0.07256367462448299,
+      "grad_norm": 8.3125,
+      "learning_rate": 4.946799863691862e-05,
+      "loss": 0.2572,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07328931137072782,
+      "grad_norm": 10.75,
+      "learning_rate": 4.9456153993216855e-05,
+      "loss": 0.3406,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07401494811697264,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.944418039040453e-05,
+      "loss": 0.288,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07474058486321747,
+      "grad_norm": 5.375,
+      "learning_rate": 4.943207789161934e-05,
+      "loss": 0.2024,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0754662216094623,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.9419846560678657e-05,
+      "loss": 0.2623,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07619185835570713,
+      "grad_norm": 8.5625,
+      "learning_rate": 4.940748646207919e-05,
+      "loss": 0.3,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07691749510195196,
+      "grad_norm": 8.8125,
+      "learning_rate": 4.9394997660996664e-05,
+      "loss": 0.2651,
+      "step": 1060
+    },
+    {
+      "epoch": 0.0776431318481968,
+      "grad_norm": 4.0,
+      "learning_rate": 4.938238022328545e-05,
+      "loss": 0.2841,
+      "step": 1070
+    },
+    {
+      "epoch": 0.07836876859444163,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.9369634215478225e-05,
+      "loss": 0.256,
+      "step": 1080
+    },
+    {
+      "epoch": 0.07909440534068646,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.935675970478563e-05,
+      "loss": 0.1948,
+      "step": 1090
+    },
+    {
+      "epoch": 0.07982004208693128,
+      "grad_norm": 6.46875,
+      "learning_rate": 4.9343756759095924e-05,
+      "loss": 0.2309,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08054567883317611,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.933062544697459e-05,
+      "loss": 0.2647,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08127131557942094,
+      "grad_norm": 4.71875,
+      "learning_rate": 4.931736583766402e-05,
+      "loss": 0.3253,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08199695232566577,
+      "grad_norm": 6.125,
+      "learning_rate": 4.9303978001083105e-05,
+      "loss": 0.2177,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0827225890719106,
+      "grad_norm": 6.78125,
+      "learning_rate": 4.929046200782691e-05,
+      "loss": 0.2809,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08344822581815543,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.927681792916626e-05,
+      "loss": 0.3722,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08417386256440026,
+      "grad_norm": 5.71875,
+      "learning_rate": 4.926304583704739e-05,
+      "loss": 0.2601,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0848994993106451,
+      "grad_norm": 5.21875,
+      "learning_rate": 4.924914580409157e-05,
+      "loss": 0.3125,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08562513605688993,
+      "grad_norm": 4.125,
+      "learning_rate": 4.9235117903594696e-05,
+      "loss": 0.2442,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08635077280313475,
+      "grad_norm": 6.6875,
+      "learning_rate": 4.922096220952692e-05,
+      "loss": 0.2837,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08707640954937958,
+      "grad_norm": 7.5625,
+      "learning_rate": 4.9206678796532285e-05,
+      "loss": 0.2564,
+      "step": 1200
+    },
+    {
+      "epoch": 0.08780204629562441,
+      "grad_norm": 12.0625,
+      "learning_rate": 4.919226773992826e-05,
+      "loss": 0.2853,
+      "step": 1210
+    },
+    {
+      "epoch": 0.08852768304186924,
+      "grad_norm": 8.0,
+      "learning_rate": 4.917772911570544e-05,
+      "loss": 0.2109,
+      "step": 1220
+    },
+    {
+      "epoch": 0.08925331978811407,
+      "grad_norm": 4.90625,
+      "learning_rate": 4.916306300052704e-05,
+      "loss": 0.281,
+      "step": 1230
+    },
+    {
+      "epoch": 0.0899789565343589,
+      "grad_norm": 3.9375,
+      "learning_rate": 4.91482694717286e-05,
+      "loss": 0.3385,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09070459328060373,
+      "grad_norm": 2.5625,
+      "learning_rate": 4.9133348607317464e-05,
+      "loss": 0.2559,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09143023002684857,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.911830048597247e-05,
+      "loss": 0.2376,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09215586677309338,
+      "grad_norm": 6.25,
+      "learning_rate": 4.910312518704348e-05,
+      "loss": 0.2155,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09288150351933822,
+      "grad_norm": 6.96875,
+      "learning_rate": 4.908782279055095e-05,
+      "loss": 0.2329,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09360714026558305,
+      "grad_norm": 7.59375,
+      "learning_rate": 4.907239337718557e-05,
+      "loss": 0.2468,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09433277701182788,
+      "grad_norm": 5.875,
+      "learning_rate": 4.9056837028307755e-05,
+      "loss": 0.2792,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09505841375807271,
+      "grad_norm": 3.78125,
+      "learning_rate": 4.9041153825947314e-05,
+      "loss": 0.1919,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09578405050431754,
+      "grad_norm": 5.65625,
+      "learning_rate": 4.902534385280292e-05,
+      "loss": 0.2775,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09650968725056237,
+      "grad_norm": 8.625,
+      "learning_rate": 4.900940719224173e-05,
+      "loss": 0.3544,
+      "step": 1330
+    },
+    {
+      "epoch": 0.0972353239968072,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.899334392829893e-05,
+      "loss": 0.3571,
+      "step": 1340
+    },
+    {
+      "epoch": 0.09796096074305202,
+      "grad_norm": 10.25,
+      "learning_rate": 4.897715414567732e-05,
+      "loss": 0.309,
+      "step": 1350
+    },
+    {
+      "epoch": 0.09868659748929685,
+      "grad_norm": 9.6875,
+      "learning_rate": 4.8960837929746794e-05,
+      "loss": 0.2712,
+      "step": 1360
+    },
+    {
+      "epoch": 0.09941223423554169,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.894439536654399e-05,
+      "loss": 0.2685,
+      "step": 1370
+    },
+    {
+      "epoch": 0.10013787098178652,
+      "grad_norm": 6.53125,
+      "learning_rate": 4.892782654277175e-05,
+      "loss": 0.266,
+      "step": 1380
+    },
+    {
+      "epoch": 0.10086350772803135,
+      "grad_norm": 4.25,
+      "learning_rate": 4.8911131545798715e-05,
+      "loss": 0.2936,
+      "step": 1390
+    },
+    {
+      "epoch": 0.10158914447427618,
+      "grad_norm": 10.9375,
+      "learning_rate": 4.889431046365883e-05,
+      "loss": 0.1927,
+      "step": 1400
+    },
+    {
+      "epoch": 0.10231478122052101,
+      "grad_norm": 5.875,
+      "learning_rate": 4.887736338505092e-05,
+      "loss": 0.2864,
+      "step": 1410
+    },
+    {
+      "epoch": 0.10304041796676584,
+      "grad_norm": 3.875,
+      "learning_rate": 4.886029039933817e-05,
+      "loss": 0.2369,
+      "step": 1420
+    },
+    {
+      "epoch": 0.10376605471301066,
+      "grad_norm": 0.91015625,
+      "learning_rate": 4.884309159654771e-05,
+      "loss": 0.2904,
+      "step": 1430
+    },
+    {
+      "epoch": 0.10449169145925549,
+      "grad_norm": 9.0,
+      "learning_rate": 4.8825767067370086e-05,
+      "loss": 0.2294,
+      "step": 1440
+    },
+    {
+      "epoch": 0.10521732820550032,
+      "grad_norm": 6.84375,
+      "learning_rate": 4.880831690315884e-05,
+      "loss": 0.2302,
+      "step": 1450
+    },
+    {
+      "epoch": 0.10594296495174516,
+      "grad_norm": 8.25,
+      "learning_rate": 4.879074119592998e-05,
+      "loss": 0.3332,
+      "step": 1460
+    },
+    {
+      "epoch": 0.10666860169798999,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.877304003836149e-05,
+      "loss": 0.2783,
+      "step": 1470
+    },
+    {
+      "epoch": 0.10739423844423482,
+      "grad_norm": 3.84375,
+      "learning_rate": 4.87552135237929e-05,
+      "loss": 0.2884,
+      "step": 1480
+    },
+    {
+      "epoch": 0.10811987519047965,
+      "grad_norm": 6.4375,
+      "learning_rate": 4.873726174622475e-05,
+      "loss": 0.3063,
+      "step": 1490
+    },
+    {
+      "epoch": 0.10884551193672448,
+      "grad_norm": 7.78125,
+      "learning_rate": 4.8719184800318066e-05,
+      "loss": 0.2767,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1095711486829693,
+      "grad_norm": 6.59375,
+      "learning_rate": 4.870098278139394e-05,
+      "loss": 0.2703,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11029678542921413,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.8682655785432975e-05,
+      "loss": 0.2799,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11102242217545896,
+      "grad_norm": 3.984375,
+      "learning_rate": 4.866420390907476e-05,
+      "loss": 0.2277,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1117480589217038,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.8645627249617406e-05,
+      "loss": 0.2159,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11247369566794863,
+      "grad_norm": 7.0625,
+      "learning_rate": 4.8626925905017017e-05,
+      "loss": 0.1614,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11319933241419346,
+      "grad_norm": 13.6875,
+      "learning_rate": 4.8608099973887167e-05,
+      "loss": 0.2458,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11392496916043829,
+      "grad_norm": 6.1875,
+      "learning_rate": 4.858914955549839e-05,
+      "loss": 0.2246,
+      "step": 1570
+    },
+    {
+      "epoch": 0.11465060590668312,
+      "grad_norm": 8.0,
+      "learning_rate": 4.857007474977764e-05,
+      "loss": 0.3887,
+      "step": 1580
+    },
+    {
+      "epoch": 0.11537624265292794,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.855087565730778e-05,
+      "loss": 0.2589,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11610187939917277,
+      "grad_norm": 6.09375,
+      "learning_rate": 4.853155237932705e-05,
+      "loss": 0.276,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1168275161454176,
+      "grad_norm": 2.71875,
+      "learning_rate": 4.8512105017728514e-05,
+      "loss": 0.2841,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11755315289166243,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.8492533675059564e-05,
+      "loss": 0.2426,
+      "step": 1620
+    },
+    {
+      "epoch": 0.11827878963790726,
+      "grad_norm": 10.375,
+      "learning_rate": 4.847283845452133e-05,
+      "loss": 0.2559,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1190044263841521,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.8453019459968174e-05,
+      "loss": 0.2093,
+      "step": 1640
+    },
+    {
+      "epoch": 0.11973006313039693,
+      "grad_norm": 9.75,
+      "learning_rate": 4.843307679590711e-05,
+      "loss": 0.3073,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12045569987664176,
+      "grad_norm": 6.84375,
+      "learning_rate": 4.8413010567497295e-05,
+      "loss": 0.2148,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12118133662288658,
+      "grad_norm": 3.765625,
+      "learning_rate": 4.8392820880549434e-05,
+      "loss": 0.1865,
+      "step": 1670
+    },
+    {
+      "epoch": 0.12190697336913141,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.837250784152524e-05,
+      "loss": 0.3362,
+      "step": 1680
+    },
+    {
+      "epoch": 0.12263261011537624,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.835207155753686e-05,
+      "loss": 0.1977,
+      "step": 1690
+    },
+    {
+      "epoch": 0.12335824686162107,
+      "grad_norm": 5.71875,
+      "learning_rate": 4.833151213634635e-05,
+      "loss": 0.2128,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1240838836078659,
+      "grad_norm": 4.375,
+      "learning_rate": 4.831082968636505e-05,
+      "loss": 0.1508,
+      "step": 1710
+    },
+    {
+      "epoch": 0.12480952035411073,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.8290024316653055e-05,
+      "loss": 0.1477,
+      "step": 1720
+    },
+    {
+      "epoch": 0.12553515710035557,
+      "grad_norm": 4.34375,
+      "learning_rate": 4.826909613691861e-05,
+      "loss": 0.164,
+      "step": 1730
+    },
+    {
+      "epoch": 0.12626079384660038,
+      "grad_norm": 5.25,
+      "learning_rate": 4.824804525751757e-05,
+      "loss": 0.2322,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12698643059284523,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.822687178945278e-05,
+      "loss": 0.2295,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12771206733909005,
+      "grad_norm": 6.6875,
+      "learning_rate": 4.820557584437349e-05,
+      "loss": 0.2771,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1284377040853349,
+      "grad_norm": 8.5,
+      "learning_rate": 4.8184157534574806e-05,
+      "loss": 0.3078,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1291633408315797,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.816261697299706e-05,
+      "loss": 0.2051,
+      "step": 1780
+    },
+    {
+      "epoch": 0.12988897757782455,
+      "grad_norm": 13.1875,
+      "learning_rate": 4.814095427322522e-05,
+      "loss": 0.2311,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13061461432406937,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.811916954948832e-05,
+      "loss": 0.2827,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1313402510703142,
+      "grad_norm": 7.90625,
+      "learning_rate": 4.809726291665881e-05,
+      "loss": 0.2158,
+      "step": 1810
+    },
+    {
+      "epoch": 0.13206588781655904,
+      "grad_norm": 6.875,
+      "learning_rate": 4.8075234490251994e-05,
+      "loss": 0.2128,
+      "step": 1820
+    },
+    {
+      "epoch": 0.13279152456280385,
+      "grad_norm": 12.5625,
+      "learning_rate": 4.805308438642539e-05,
+      "loss": 0.2479,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1335171613090487,
+      "grad_norm": 6.46875,
+      "learning_rate": 4.8030812721978135e-05,
+      "loss": 0.2439,
+      "step": 1840
+    },
+    {
+      "epoch": 0.13424279805529352,
+      "grad_norm": 10.125,
+      "learning_rate": 4.800841961435037e-05,
+      "loss": 0.2566,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13496843480153836,
+      "grad_norm": 9.6875,
+      "learning_rate": 4.7985905181622605e-05,
+      "loss": 0.2342,
+      "step": 1860
+    },
+    {
+      "epoch": 0.13569407154778318,
+      "grad_norm": 2.25,
+      "learning_rate": 4.7963269542515106e-05,
+      "loss": 0.1505,
+      "step": 1870
+    },
+    {
+      "epoch": 0.136419708294028,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.794051281638729e-05,
+      "loss": 0.2658,
+      "step": 1880
+    },
+    {
+      "epoch": 0.13714534504027284,
+      "grad_norm": 7.21875,
+      "learning_rate": 4.7917635123237025e-05,
+      "loss": 0.3074,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13787098178651766,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.789463658370013e-05,
+      "loss": 0.2701,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1385966185327625,
+      "grad_norm": 5.84375,
+      "learning_rate": 4.7871517319049566e-05,
+      "loss": 0.2645,
+      "step": 1910
+    },
+    {
+      "epoch": 0.13932225527900732,
+      "grad_norm": 3.6875,
+      "learning_rate": 4.784827745119493e-05,
+      "loss": 0.2724,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14004789202525217,
+      "grad_norm": 0.9375,
+      "learning_rate": 4.7824917102681796e-05,
+      "loss": 0.21,
+      "step": 1930
+    },
+    {
+      "epoch": 0.140773528771497,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.780143639669098e-05,
+      "loss": 0.1734,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14149916551774183,
+      "grad_norm": 6.84375,
+      "learning_rate": 4.7777835457038e-05,
+      "loss": 0.2173,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14222480226398665,
+      "grad_norm": 4.5,
+      "learning_rate": 4.775411440817236e-05,
+      "loss": 0.2905,
+      "step": 1960
+    },
+    {
+      "epoch": 0.14295043901023147,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.7730273375176885e-05,
+      "loss": 0.2289,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1436760757564763,
+      "grad_norm": 12.4375,
+      "learning_rate": 4.770631248376713e-05,
+      "loss": 0.2817,
+      "step": 1980
+    },
+    {
+      "epoch": 0.14440171250272113,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.768223186029064e-05,
+      "loss": 0.3234,
+      "step": 1990
+    },
+    {
+      "epoch": 0.14512734924896598,
+      "grad_norm": 11.9375,
+      "learning_rate": 4.765803163172632e-05,
+      "loss": 0.3516,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1458529859952108,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.7633711925683786e-05,
+      "loss": 0.1774,
+      "step": 2010
+    },
+    {
+      "epoch": 0.14657862274145564,
+      "grad_norm": 4.875,
+      "learning_rate": 4.760927287040262e-05,
+      "loss": 0.2695,
+      "step": 2020
+    },
+    {
+      "epoch": 0.14730425948770046,
+      "grad_norm": 7.5,
+      "learning_rate": 4.75847145947518e-05,
+      "loss": 0.2207,
+      "step": 2030
+    },
+    {
+      "epoch": 0.14802989623394527,
+      "grad_norm": 6.03125,
+      "learning_rate": 4.756003722822891e-05,
+      "loss": 0.2679,
+      "step": 2040
+    },
+    {
+      "epoch": 0.14875553298019012,
+      "grad_norm": 5.8125,
+      "learning_rate": 4.753524090095953e-05,
+      "loss": 0.2325,
+      "step": 2050
+    },
+    {
+      "epoch": 0.14948116972643494,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.751032574369655e-05,
+      "loss": 0.2521,
+      "step": 2060
+    },
+    {
+      "epoch": 0.15020680647267978,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.748529188781942e-05,
+      "loss": 0.2213,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1509324432189246,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.746013946533352e-05,
+      "loss": 0.2019,
+      "step": 2080
+    },
+    {
+      "epoch": 0.15165807996516945,
+      "grad_norm": 6.75,
+      "learning_rate": 4.7434868608869435e-05,
+      "loss": 0.2303,
+      "step": 2090
+    },
+    {
+      "epoch": 0.15238371671141426,
+      "grad_norm": 2.34375,
+      "learning_rate": 4.740947945168227e-05,
+      "loss": 0.2411,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1531093534576591,
+      "grad_norm": 5.28125,
+      "learning_rate": 4.7383972127650924e-05,
+      "loss": 0.2548,
+      "step": 2110
+    },
+    {
+      "epoch": 0.15383499020390393,
+      "grad_norm": 2.640625,
+      "learning_rate": 4.73583467712774e-05,
+      "loss": 0.1987,
+      "step": 2120
+    },
+    {
+      "epoch": 0.15456062695014874,
+      "grad_norm": 10.3125,
+      "learning_rate": 4.733260351768611e-05,
+      "loss": 0.294,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1552862636963936,
+      "grad_norm": 6.15625,
+      "learning_rate": 4.7306742502623124e-05,
+      "loss": 0.178,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1560119004426384,
+      "grad_norm": 5.5,
+      "learning_rate": 4.72807638624555e-05,
+      "loss": 0.2031,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15673753718888325,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.7254667734170526e-05,
+      "loss": 0.1921,
+      "step": 2160
+    },
+    {
+      "epoch": 0.15746317393512807,
+      "grad_norm": 2.4375,
+      "learning_rate": 4.7228454255375024e-05,
+      "loss": 0.2581,
+      "step": 2170
+    },
+    {
+      "epoch": 0.15818881068137292,
+      "grad_norm": 7.125,
+      "learning_rate": 4.72021235642946e-05,
+      "loss": 0.2175,
+      "step": 2180
+    },
+    {
+      "epoch": 0.15891444742761773,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.7175675799772935e-05,
+      "loss": 0.212,
+      "step": 2190
+    },
+    {
+      "epoch": 0.15964008417386255,
+      "grad_norm": 8.0625,
+      "learning_rate": 4.714911110127106e-05,
+      "loss": 0.1766,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1603657209201074,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.712242960886659e-05,
+      "loss": 0.2605,
+      "step": 2210
+    },
+    {
+      "epoch": 0.16109135766635221,
+      "grad_norm": 0.87890625,
+      "learning_rate": 4.709563146325301e-05,
+      "loss": 0.211,
+      "step": 2220
+    },
+    {
+      "epoch": 0.16181699441259706,
+      "grad_norm": 6.625,
+      "learning_rate": 4.7068716805738935e-05,
+      "loss": 0.2391,
+      "step": 2230
+    },
+    {
+      "epoch": 0.16254263115884188,
+      "grad_norm": 5.15625,
+      "learning_rate": 4.704168577824733e-05,
+      "loss": 0.259,
+      "step": 2240
+    },
+    {
+      "epoch": 0.16326826790508672,
+      "grad_norm": 8.5,
+      "learning_rate": 4.701453852331482e-05,
+      "loss": 0.2316,
+      "step": 2250
+    },
+    {
+      "epoch": 0.16399390465133154,
+      "grad_norm": 5.125,
+      "learning_rate": 4.6987275184090876e-05,
+      "loss": 0.2271,
+      "step": 2260
+    },
+    {
+      "epoch": 0.16471954139757639,
+      "grad_norm": 7.15625,
+      "learning_rate": 4.6959895904337115e-05,
+      "loss": 0.2099,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1654451781438212,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.6932400828426505e-05,
+      "loss": 0.1646,
+      "step": 2280
+    },
+    {
+      "epoch": 0.16617081489006602,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.690479010134261e-05,
+      "loss": 0.2043,
+      "step": 2290
+    },
+    {
+      "epoch": 0.16689645163631087,
+      "grad_norm": 4.625,
+      "learning_rate": 4.687706386867885e-05,
+      "loss": 0.2699,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16762208838255568,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.6849222276637686e-05,
+      "loss": 0.2132,
+      "step": 2310
+    },
+    {
+      "epoch": 0.16834772512880053,
+      "grad_norm": 12.0625,
+      "learning_rate": 4.682126547202991e-05,
+      "loss": 0.2614,
+      "step": 2320
+    },
+    {
+      "epoch": 0.16907336187504535,
+      "grad_norm": 2.484375,
+      "learning_rate": 4.679319360227381e-05,
+      "loss": 0.1741,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1697989986212902,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.676500681539444e-05,
+      "loss": 0.2116,
+      "step": 2340
+    },
+    {
+      "epoch": 0.170524635367535,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.673670526002281e-05,
+      "loss": 0.2688,
+      "step": 2350
+    },
+    {
+      "epoch": 0.17125027211377986,
+      "grad_norm": 2.15625,
+      "learning_rate": 4.6708289085395115e-05,
+      "loss": 0.2458,
+      "step": 2360
+    },
+    {
+      "epoch": 0.17197590886002467,
+      "grad_norm": 8.625,
+      "learning_rate": 4.6679758441351964e-05,
+      "loss": 0.1767,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1727015456062695,
+      "grad_norm": 7.5,
+      "learning_rate": 4.665111347833754e-05,
+      "loss": 0.2735,
+      "step": 2380
+    },
+    {
+      "epoch": 0.17342718235251434,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.662235434739885e-05,
+      "loss": 0.1716,
+      "step": 2390
+    },
+    {
+      "epoch": 0.17415281909875915,
+      "grad_norm": 8.625,
+      "learning_rate": 4.659348120018493e-05,
+      "loss": 0.2536,
+      "step": 2400
+    },
+    {
+      "epoch": 0.174878455845004,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.6564494188946015e-05,
+      "loss": 0.2293,
+      "step": 2410
+    },
+    {
+      "epoch": 0.17560409259124882,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.653539346653278e-05,
+      "loss": 0.2266,
+      "step": 2420
+    },
+    {
+      "epoch": 0.17632972933749366,
+      "grad_norm": 5.875,
+      "learning_rate": 4.650617918639546e-05,
+      "loss": 0.2485,
+      "step": 2430
+    },
+    {
+      "epoch": 0.17705536608373848,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.647685150258314e-05,
+      "loss": 0.2105,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1777810028299833,
+      "grad_norm": 5.34375,
+      "learning_rate": 4.644741056974287e-05,
+      "loss": 0.2731,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17850663957622814,
+      "grad_norm": 8.1875,
+      "learning_rate": 4.641785654311884e-05,
+      "loss": 0.2572,
+      "step": 2460
+    },
+    {
+      "epoch": 0.17923227632247296,
+      "grad_norm": 7.71875,
+      "learning_rate": 4.6388189578551646e-05,
+      "loss": 0.2705,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1799579130687178,
+      "grad_norm": 1.234375,
+      "learning_rate": 4.635840983247738e-05,
+      "loss": 0.2216,
+      "step": 2480
+    },
+    {
+      "epoch": 0.18068354981496262,
+      "grad_norm": 3.796875,
+      "learning_rate": 4.632851746192685e-05,
+      "loss": 0.2416,
+      "step": 2490
+    },
+    {
+      "epoch": 0.18140918656120747,
+      "grad_norm": 8.9375,
+      "learning_rate": 4.6298512624524734e-05,
+      "loss": 0.0987,
+      "step": 2500
+    },
+    {
+      "epoch": 0.1821348233074523,
+      "grad_norm": 7.09375,
+      "learning_rate": 4.626839547848877e-05,
+      "loss": 0.1668,
+      "step": 2510
+    },
+    {
+      "epoch": 0.18286046005369713,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.62381661826289e-05,
+      "loss": 0.1733,
+      "step": 2520
+    },
+    {
+      "epoch": 0.18358609679994195,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.620782489634642e-05,
+      "loss": 0.2451,
+      "step": 2530
+    },
+    {
+      "epoch": 0.18431173354618677,
+      "grad_norm": 8.3125,
+      "learning_rate": 4.617737177963319e-05,
+      "loss": 0.1714,
+      "step": 2540
+    },
+    {
+      "epoch": 0.1850373702924316,
+      "grad_norm": 3.796875,
+      "learning_rate": 4.614680699307076e-05,
+      "loss": 0.173,
+      "step": 2550
+    },
+    {
+      "epoch": 0.18576300703867643,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.6116130697829495e-05,
+      "loss": 0.2125,
+      "step": 2560
+    },
+    {
+      "epoch": 0.18648864378492128,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.6085343055667786e-05,
+      "loss": 0.2366,
+      "step": 2570
+    },
+    {
+      "epoch": 0.1872142805311661,
+      "grad_norm": 1.1875,
+      "learning_rate": 4.6054444228931136e-05,
+      "loss": 0.2829,
+      "step": 2580
+    },
+    {
+      "epoch": 0.18793991727741094,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.6023434380551354e-05,
+      "loss": 0.2183,
+      "step": 2590
+    },
+    {
+      "epoch": 0.18866555402365576,
+      "grad_norm": 6.46875,
+      "learning_rate": 4.5992313674045664e-05,
+      "loss": 0.2144,
+      "step": 2600
+    },
+    {
+      "epoch": 0.18939119076990057,
+      "grad_norm": 5.53125,
+      "learning_rate": 4.5961082273515856e-05,
+      "loss": 0.1141,
+      "step": 2610
+    },
+    {
+      "epoch": 0.19011682751614542,
+      "grad_norm": 4.125,
+      "learning_rate": 4.592974034364741e-05,
+      "loss": 0.3297,
+      "step": 2620
+    },
+    {
+      "epoch": 0.19084246426239024,
+      "grad_norm": 3.734375,
+      "learning_rate": 4.589828804970865e-05,
+      "loss": 0.2677,
+      "step": 2630
+    },
+    {
+      "epoch": 0.19156810100863508,
+      "grad_norm": 6.125,
+      "learning_rate": 4.5866725557549864e-05,
+      "loss": 0.1603,
+      "step": 2640
+    },
+    {
+      "epoch": 0.1922937377548799,
+      "grad_norm": 3.25,
+      "learning_rate": 4.583505303360238e-05,
+      "loss": 0.2041,
+      "step": 2650
+    },
+    {
+      "epoch": 0.19301937450112475,
+      "grad_norm": 4.71875,
+      "learning_rate": 4.580327064487778e-05,
+      "loss": 0.2155,
+      "step": 2660
+    },
+    {
+      "epoch": 0.19374501124736956,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.577137855896695e-05,
+      "loss": 0.1948,
+      "step": 2670
+    },
+    {
+      "epoch": 0.1944706479936144,
+      "grad_norm": 10.375,
+      "learning_rate": 4.5739376944039225e-05,
+      "loss": 0.286,
+      "step": 2680
+    },
+    {
+      "epoch": 0.19519628473985923,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.5707265968841475e-05,
+      "loss": 0.1766,
+      "step": 2690
+    },
+    {
+      "epoch": 0.19592192148610404,
+      "grad_norm": 8.625,
+      "learning_rate": 4.5675045802697266e-05,
+      "loss": 0.1973,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1966475582323489,
+      "grad_norm": 7.5625,
+      "learning_rate": 4.564271661550591e-05,
+      "loss": 0.2644,
+      "step": 2710
+    },
+    {
+      "epoch": 0.1973731949785937,
+      "grad_norm": 6.09375,
+      "learning_rate": 4.56102785777416e-05,
+      "loss": 0.2265,
+      "step": 2720
+    },
+    {
+      "epoch": 0.19809883172483855,
+      "grad_norm": 3.890625,
+      "learning_rate": 4.5577731860452525e-05,
+      "loss": 0.2042,
+      "step": 2730
+    },
+    {
+      "epoch": 0.19882446847108337,
+      "grad_norm": 5.625,
+      "learning_rate": 4.5545076635259924e-05,
+      "loss": 0.1945,
+      "step": 2740
+    },
+    {
+      "epoch": 0.19955010521732822,
+      "grad_norm": 7.625,
+      "learning_rate": 4.5512313074357225e-05,
+      "loss": 0.2212,
+      "step": 2750
+    },
+    {
+      "epoch": 0.20027574196357303,
+      "grad_norm": 7.6875,
+      "learning_rate": 4.54794413505091e-05,
+      "loss": 0.1836,
+      "step": 2760
+    },
+    {
+      "epoch": 0.20100137870981785,
+      "grad_norm": 5.53125,
+      "learning_rate": 4.544646163705059e-05,
+      "loss": 0.2268,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2017270154560627,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.541337410788616e-05,
+      "loss": 0.1336,
+      "step": 2780
+    },
+    {
+      "epoch": 0.20245265220230751,
+      "grad_norm": 9.4375,
+      "learning_rate": 4.538017893748881e-05,
+      "loss": 0.161,
+      "step": 2790
+    },
+    {
+      "epoch": 0.20317828894855236,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.534687630089912e-05,
+      "loss": 0.2052,
+      "step": 2800
+    },
+    {
+      "epoch": 0.20390392569479718,
+      "grad_norm": 9.25,
+      "learning_rate": 4.5313466373724366e-05,
+      "loss": 0.288,
+      "step": 2810
+    },
+    {
+      "epoch": 0.20462956244104202,
+      "grad_norm": 0.87109375,
+      "learning_rate": 4.5279949332137564e-05,
+      "loss": 0.1063,
+      "step": 2820
+    },
+    {
+      "epoch": 0.20535519918728684,
+      "grad_norm": 9.625,
+      "learning_rate": 4.524632535287655e-05,
+      "loss": 0.2642,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2060808359335317,
+      "grad_norm": 7.21875,
+      "learning_rate": 4.521259461324308e-05,
+      "loss": 0.1667,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2068064726797765,
+      "grad_norm": 5.78125,
+      "learning_rate": 4.51787572911018e-05,
+      "loss": 0.1666,
+      "step": 2850
+    },
+    {
+      "epoch": 0.20753210942602132,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.514481356487945e-05,
+      "loss": 0.1352,
+      "step": 2860
+    },
+    {
+      "epoch": 0.20825774617226617,
+      "grad_norm": 4.53125,
+      "learning_rate": 4.51107636135638e-05,
+      "loss": 0.1691,
+      "step": 2870
+    },
+    {
+      "epoch": 0.20898338291851098,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.507660761670277e-05,
+      "loss": 0.1735,
+      "step": 2880
+    },
+    {
+      "epoch": 0.20970901966475583,
+      "grad_norm": 6.8125,
+      "learning_rate": 4.504234575440345e-05,
+      "loss": 0.2212,
+      "step": 2890
+    },
+    {
+      "epoch": 0.21043465641100065,
+      "grad_norm": 8.4375,
+      "learning_rate": 4.500797820733117e-05,
+      "loss": 0.1927,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2111602931572455,
+      "grad_norm": 5.0,
+      "learning_rate": 4.4973505156708576e-05,
+      "loss": 0.2425,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2118859299034903,
+      "grad_norm": 4.375,
+      "learning_rate": 4.4938926784314585e-05,
+      "loss": 0.2642,
+      "step": 2920
+    },
+    {
+      "epoch": 0.21261156664973516,
+      "grad_norm": 6.4375,
+      "learning_rate": 4.490424327248354e-05,
+      "loss": 0.1719,
+      "step": 2930
+    },
+    {
+      "epoch": 0.21333720339597997,
+      "grad_norm": 3.390625,
+      "learning_rate": 4.486945480410415e-05,
+      "loss": 0.1807,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2140628401422248,
+      "grad_norm": 11.0,
+      "learning_rate": 4.4834561562618595e-05,
+      "loss": 0.2051,
+      "step": 2950
+    },
+    {
+      "epoch": 0.21478847688846964,
+      "grad_norm": 5.78125,
+      "learning_rate": 4.479956373202151e-05,
+      "loss": 0.1103,
+      "step": 2960
+    },
+    {
+      "epoch": 0.21551411363471445,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.476446149685906e-05,
+      "loss": 0.2068,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2162397503809593,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.472925504222792e-05,
+      "loss": 0.3126,
+      "step": 2980
+    },
+    {
+      "epoch": 0.21696538712720412,
+      "grad_norm": 24.875,
+      "learning_rate": 4.4693944553774335e-05,
+      "loss": 0.2166,
+      "step": 2990
+    },
+    {
+      "epoch": 0.21769102387344896,
+      "grad_norm": 7.15625,
+      "learning_rate": 4.465853021769313e-05,
+      "loss": 0.277,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21841666061969378,
+      "grad_norm": 5.9375,
+      "learning_rate": 4.4623012220726705e-05,
+      "loss": 0.2262,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2191422973659386,
+      "grad_norm": 6.90625,
+      "learning_rate": 4.45873907501641e-05,
+      "loss": 0.1627,
+      "step": 3020
+    },
+    {
+      "epoch": 0.21986793411218344,
+      "grad_norm": 8.3125,
+      "learning_rate": 4.455166599383997e-05,
+      "loss": 0.1624,
+      "step": 3030
+    },
+    {
+      "epoch": 0.22059357085842826,
+      "grad_norm": 3.875,
+      "learning_rate": 4.4515838140133594e-05,
+      "loss": 0.2401,
+      "step": 3040
+    },
+    {
+      "epoch": 0.2213192076046731,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.447990737796789e-05,
+      "loss": 0.1749,
+      "step": 3050
+    },
+    {
+      "epoch": 0.22204484435091792,
+      "grad_norm": 2.46875,
+      "learning_rate": 4.4443873896808434e-05,
+      "loss": 0.2449,
+      "step": 3060
+    },
+    {
+      "epoch": 0.22277048109716277,
+      "grad_norm": 9.125,
+      "learning_rate": 4.440773788666244e-05,
+      "loss": 0.2794,
+      "step": 3070
+    },
+    {
+      "epoch": 0.2234961178434076,
+      "grad_norm": 8.5625,
+      "learning_rate": 4.437149953807777e-05,
+      "loss": 0.2724,
+      "step": 3080
+    },
+    {
+      "epoch": 0.22422175458965243,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.433515904214191e-05,
+      "loss": 0.1973,
+      "step": 3090
+    },
+    {
+      "epoch": 0.22494739133589725,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.429871659048099e-05,
+      "loss": 0.3198,
+      "step": 3100
+    },
+    {
+      "epoch": 0.22567302808214207,
+      "grad_norm": 7.96875,
+      "learning_rate": 4.426217237525875e-05,
+      "loss": 0.275,
+      "step": 3110
+    },
+    {
+      "epoch": 0.22639866482838691,
+      "grad_norm": 7.09375,
+      "learning_rate": 4.422552658917556e-05,
+      "loss": 0.1821,
+      "step": 3120
+    },
+    {
+      "epoch": 0.22712430157463173,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.418877942546733e-05,
+      "loss": 0.2716,
+      "step": 3130
+    },
+    {
+      "epoch": 0.22784993832087658,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.4151931077904606e-05,
+      "loss": 0.1725,
+      "step": 3140
+    },
+    {
+      "epoch": 0.2285755750671214,
+      "grad_norm": 2.375,
+      "learning_rate": 4.4114981740791436e-05,
+      "loss": 0.1706,
+      "step": 3150
+    },
+    {
+      "epoch": 0.22930121181336624,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.407793160896442e-05,
+      "loss": 0.1741,
+      "step": 3160
+    },
+    {
+      "epoch": 0.23002684855961106,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.404078087779164e-05,
+      "loss": 0.1467,
+      "step": 3170
+    },
+    {
+      "epoch": 0.23075248530585588,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.400352974317166e-05,
+      "loss": 0.2124,
+      "step": 3180
+    },
+    {
+      "epoch": 0.23147812205210072,
+      "grad_norm": 5.15625,
+      "learning_rate": 4.396617840153245e-05,
+      "loss": 0.197,
+      "step": 3190
+    },
+    {
+      "epoch": 0.23220375879834554,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.392872704983042e-05,
+      "loss": 0.2052,
+      "step": 3200
+    },
+    {
+      "epoch": 0.23292939554459038,
+      "grad_norm": 6.5,
+      "learning_rate": 4.389117588554932e-05,
+      "loss": 0.2435,
+      "step": 3210
+    },
+    {
+      "epoch": 0.2336550322908352,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.38535251066992e-05,
+      "loss": 0.1877,
+      "step": 3220
+    },
+    {
+      "epoch": 0.23438066903708005,
+      "grad_norm": 7.625,
+      "learning_rate": 4.381577491181542e-05,
+      "loss": 0.3165,
+      "step": 3230
+    },
+    {
+      "epoch": 0.23510630578332486,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.377792549995755e-05,
+      "loss": 0.1709,
+      "step": 3240
+    },
+    {
+      "epoch": 0.2358319425295697,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.373997707070835e-05,
+      "loss": 0.161,
+      "step": 3250
+    },
+    {
+      "epoch": 0.23655757927581453,
+      "grad_norm": 4.8125,
+      "learning_rate": 4.370192982417268e-05,
+      "loss": 0.233,
+      "step": 3260
+    },
+    {
+      "epoch": 0.23728321602205935,
+      "grad_norm": 7.46875,
+      "learning_rate": 4.366378396097649e-05,
+      "loss": 0.1879,
+      "step": 3270
+    },
+    {
+      "epoch": 0.2380088527683042,
+      "grad_norm": 2.25,
+      "learning_rate": 4.362553968226575e-05,
+      "loss": 0.2066,
+      "step": 3280
+    },
+    {
+      "epoch": 0.238734489514549,
+      "grad_norm": 7.53125,
+      "learning_rate": 4.3587197189705375e-05,
+      "loss": 0.1907,
+      "step": 3290
+    },
+    {
+      "epoch": 0.23946012626079385,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.354875668547815e-05,
+      "loss": 0.138,
+      "step": 3300
+    },
+    {
+      "epoch": 0.24018576300703867,
+      "grad_norm": 8.375,
+      "learning_rate": 4.351021837228372e-05,
+      "loss": 0.2027,
+      "step": 3310
+    },
+    {
+      "epoch": 0.24091139975328352,
+      "grad_norm": 9.375,
+      "learning_rate": 4.347158245333744e-05,
+      "loss": 0.1607,
+      "step": 3320
+    },
+    {
+      "epoch": 0.24163703649952833,
+      "grad_norm": 3.890625,
+      "learning_rate": 4.343284913236939e-05,
+      "loss": 0.1537,
+      "step": 3330
+    },
+    {
+      "epoch": 0.24236267324577315,
+      "grad_norm": 10.0625,
+      "learning_rate": 4.3394018613623235e-05,
+      "loss": 0.2282,
+      "step": 3340
+    },
+    {
+      "epoch": 0.243088309992018,
+      "grad_norm": 6.53125,
+      "learning_rate": 4.335509110185517e-05,
+      "loss": 0.1904,
+      "step": 3350
+    },
+    {
+      "epoch": 0.24381394673826282,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.3316066802332854e-05,
+      "loss": 0.1727,
+      "step": 3360
+    },
+    {
+      "epoch": 0.24453958348450766,
+      "grad_norm": 7.59375,
+      "learning_rate": 4.327694592083431e-05,
+      "loss": 0.2824,
+      "step": 3370
+    },
+    {
+      "epoch": 0.24526522023075248,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.323772866364683e-05,
+      "loss": 0.295,
+      "step": 3380
+    },
+    {
+      "epoch": 0.24599085697699732,
+      "grad_norm": 3.625,
+      "learning_rate": 4.319841523756591e-05,
+      "loss": 0.232,
+      "step": 3390
+    },
+    {
+      "epoch": 0.24671649372324214,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.3159005849894175e-05,
+      "loss": 0.2444,
+      "step": 3400
+    },
+    {
+      "epoch": 0.247442130469487,
+      "grad_norm": 4.4375,
+      "learning_rate": 4.3119500708440244e-05,
+      "loss": 0.1459,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2481677672157318,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.3079900021517626e-05,
+      "loss": 0.2661,
+      "step": 3420
+    },
+    {
+      "epoch": 0.24889340396197662,
+      "grad_norm": 1.0859375,
+      "learning_rate": 4.304020399794371e-05,
+      "loss": 0.3032,
+      "step": 3430
+    },
+    {
+      "epoch": 0.24961904070822147,
+      "grad_norm": 7.40625,
+      "learning_rate": 4.300041284703853e-05,
+      "loss": 0.2178,
+      "step": 3440
+    },
+    {
+      "epoch": 0.2503446774544663,
+      "grad_norm": 5.90625,
+      "learning_rate": 4.29605267786238e-05,
+      "loss": 0.2001,
+      "step": 3450
+    },
+    {
+      "epoch": 0.25107031420071113,
+      "grad_norm": 8.0625,
+      "learning_rate": 4.292054600302171e-05,
+      "loss": 0.1923,
+      "step": 3460
+    },
+    {
+      "epoch": 0.25179595094695595,
+      "grad_norm": 4.125,
+      "learning_rate": 4.288047073105384e-05,
+      "loss": 0.1961,
+      "step": 3470
+    },
+    {
+      "epoch": 0.25252158769320077,
+      "grad_norm": 8.25,
+      "learning_rate": 4.2840301174040056e-05,
+      "loss": 0.2374,
+      "step": 3480
+    },
+    {
+      "epoch": 0.25324722443944564,
+      "grad_norm": 6.125,
+      "learning_rate": 4.280003754379742e-05,
+      "loss": 0.2464,
+      "step": 3490
+    },
+    {
+      "epoch": 0.25397286118569046,
+      "grad_norm": 9.25,
+      "learning_rate": 4.275968005263904e-05,
+      "loss": 0.2261,
+      "step": 3500
+    },
+    {
+      "epoch": 0.2546984979319353,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.2719228913372946e-05,
+      "loss": 0.1982,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2554241346781801,
+      "grad_norm": 1.9921875,
+      "learning_rate": 4.2678684339300987e-05,
+      "loss": 0.1902,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2561497714244249,
+      "grad_norm": 4.46875,
+      "learning_rate": 4.263804654421771e-05,
+      "loss": 0.239,
+      "step": 3530
+    },
+    {
+      "epoch": 0.2568754081706698,
+      "grad_norm": 8.375,
+      "learning_rate": 4.2597315742409214e-05,
+      "loss": 0.2097,
+      "step": 3540
+    },
+    {
+      "epoch": 0.2576010449169146,
+      "grad_norm": 6.3125,
+      "learning_rate": 4.2556492148652036e-05,
+      "loss": 0.234,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2583266816631594,
+      "grad_norm": 4.875,
+      "learning_rate": 4.2515575978212006e-05,
+      "loss": 0.3032,
+      "step": 3560
+    },
+    {
+      "epoch": 0.25905231840940424,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.2474567446843115e-05,
+      "loss": 0.285,
+      "step": 3570
+    },
+    {
+      "epoch": 0.2597779551556491,
+      "grad_norm": 3.625,
+      "learning_rate": 4.243346677078639e-05,
+      "loss": 0.1276,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2605035919018939,
+      "grad_norm": 5.40625,
+      "learning_rate": 4.239227416676874e-05,
+      "loss": 0.1852,
+      "step": 3590
+    },
+    {
+      "epoch": 0.26122922864813874,
+      "grad_norm": 6.6875,
+      "learning_rate": 4.2350989852001824e-05,
+      "loss": 0.2314,
+      "step": 3600
+    },
+    {
+      "epoch": 0.26195486539438356,
+      "grad_norm": 3.8125,
+      "learning_rate": 4.2309614044180856e-05,
+      "loss": 0.1953,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2626805021406284,
+      "grad_norm": 6.90625,
+      "learning_rate": 4.2268146961483573e-05,
+      "loss": 0.2408,
+      "step": 3620
+    },
+    {
+      "epoch": 0.26340613888687325,
+      "grad_norm": 4.40625,
+      "learning_rate": 4.2226588822568956e-05,
+      "loss": 0.184,
+      "step": 3630
+    },
+    {
+      "epoch": 0.26413177563311807,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.218493984657614e-05,
+      "loss": 0.1992,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2648574123793629,
+      "grad_norm": 5.71875,
+      "learning_rate": 4.214320025312328e-05,
+      "loss": 0.1468,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2655830491256077,
+      "grad_norm": 5.65625,
+      "learning_rate": 4.2101370262306325e-05,
+      "loss": 0.233,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2663086858718525,
+      "grad_norm": 6.84375,
+      "learning_rate": 4.2059450094697913e-05,
+      "loss": 0.1435,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2670343226180974,
+      "grad_norm": 5.90625,
+      "learning_rate": 4.2017439971346195e-05,
+      "loss": 0.2085,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2677599593643422,
+      "grad_norm": 3.125,
+      "learning_rate": 4.197534011377367e-05,
+      "loss": 0.1725,
+      "step": 3690
+    },
+    {
+      "epoch": 0.26848559611058703,
+      "grad_norm": 6.0,
+      "learning_rate": 4.1933150743976e-05,
+      "loss": 0.1852,
+      "step": 3700
+    },
+    {
+      "epoch": 0.26921123285683185,
+      "grad_norm": 3.578125,
+      "learning_rate": 4.189087208442084e-05,
+      "loss": 0.2231,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2699368696030767,
+      "grad_norm": 4.15625,
+      "learning_rate": 4.184850435804672e-05,
+      "loss": 0.2219,
+      "step": 3720
+    },
+    {
+      "epoch": 0.27066250634932154,
+      "grad_norm": 6.375,
+      "learning_rate": 4.180604778826176e-05,
+      "loss": 0.2189,
+      "step": 3730
+    },
+    {
+      "epoch": 0.27138814309556636,
+      "grad_norm": 6.25,
+      "learning_rate": 4.176350259894263e-05,
+      "loss": 0.1797,
+      "step": 3740
+    },
+    {
+      "epoch": 0.2721137798418112,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.172086901443325e-05,
+      "loss": 0.2735,
+      "step": 3750
+    },
+    {
+      "epoch": 0.272839416588056,
+      "grad_norm": 5.46875,
+      "learning_rate": 4.167814725954367e-05,
+      "loss": 0.2413,
+      "step": 3760
+    },
+    {
+      "epoch": 0.27356505333430087,
+      "grad_norm": 8.625,
+      "learning_rate": 4.163533755954885e-05,
+      "loss": 0.2151,
+      "step": 3770
+    },
+    {
+      "epoch": 0.2742906900805457,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.159244014018754e-05,
+      "loss": 0.1747,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2750163268267905,
+      "grad_norm": 5.84375,
+      "learning_rate": 4.154945522766098e-05,
+      "loss": 0.2007,
+      "step": 3790
+    },
+    {
+      "epoch": 0.2757419635730353,
+      "grad_norm": 6.0,
+      "learning_rate": 4.15063830486318e-05,
+      "loss": 0.1511,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2764676003192802,
+      "grad_norm": 5.5625,
+      "learning_rate": 4.14632238302228e-05,
+      "loss": 0.1725,
+      "step": 3810
+    },
+    {
+      "epoch": 0.277193237065525,
+      "grad_norm": 6.375,
+      "learning_rate": 4.141997780001573e-05,
+      "loss": 0.1666,
+      "step": 3820
+    },
+    {
+      "epoch": 0.27791887381176983,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.1376645186050105e-05,
+      "loss": 0.2431,
+      "step": 3830
+    },
+    {
+      "epoch": 0.27864451055801465,
+      "grad_norm": 6.15625,
+      "learning_rate": 4.133322621682201e-05,
+      "loss": 0.2868,
+      "step": 3840
+    },
+    {
+      "epoch": 0.27937014730425946,
+      "grad_norm": 1.625,
+      "learning_rate": 4.128972112128288e-05,
+      "loss": 0.2607,
+      "step": 3850
+    },
+    {
+      "epoch": 0.28009578405050434,
+      "grad_norm": 6.71875,
+      "learning_rate": 4.124613012883831e-05,
+      "loss": 0.1274,
+      "step": 3860
+    },
+    {
+      "epoch": 0.28082142079674915,
+      "grad_norm": 5.15625,
+      "learning_rate": 4.1202453469346826e-05,
+      "loss": 0.1535,
+      "step": 3870
+    },
+    {
+      "epoch": 0.281547057542994,
+      "grad_norm": 3.625,
+      "learning_rate": 4.115869137311869e-05,
+      "loss": 0.1437,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2822726942892388,
+      "grad_norm": 4.125,
+      "learning_rate": 4.111484407091467e-05,
+      "loss": 0.2132,
+      "step": 3890
+    },
+    {
+      "epoch": 0.28299833103548366,
+      "grad_norm": 5.84375,
+      "learning_rate": 4.107091179394484e-05,
+      "loss": 0.2423,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2837239677817285,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.1026894773867355e-05,
+      "loss": 0.2818,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2844496045279733,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.098279324278721e-05,
+      "loss": 0.2143,
+      "step": 3920
+    },
+    {
+      "epoch": 0.2851752412742181,
+      "grad_norm": 7.375,
+      "learning_rate": 4.0938607433255056e-05,
+      "loss": 0.2649,
+      "step": 3930
+    },
+    {
+      "epoch": 0.28590087802046293,
+      "grad_norm": 5.34375,
+      "learning_rate": 4.0894337578265924e-05,
+      "loss": 0.1729,
+      "step": 3940
+    },
+    {
+      "epoch": 0.2866265147667078,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.084998391125804e-05,
+      "loss": 0.1692,
+      "step": 3950
+    },
+    {
+      "epoch": 0.2873521515129526,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.0805546666111575e-05,
+      "loss": 0.2736,
+      "step": 3960
+    },
+    {
+      "epoch": 0.28807778825919744,
+      "grad_norm": 11.0625,
+      "learning_rate": 4.0761026077147425e-05,
+      "loss": 0.2362,
+      "step": 3970
+    },
+    {
+      "epoch": 0.28880342500544226,
+      "grad_norm": 4.21875,
+      "learning_rate": 4.071642237912593e-05,
+      "loss": 0.1906,
+      "step": 3980
+    },
+    {
+      "epoch": 0.2895290617516871,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.0671735807245704e-05,
+      "loss": 0.1604,
+      "step": 3990
+    },
+    {
+      "epoch": 0.29025469849793195,
+      "grad_norm": 6.53125,
+      "learning_rate": 4.0626966597142354e-05,
+      "loss": 0.172,
+      "step": 4000
+    },
+    {
+      "epoch": 0.29098033524417677,
+      "grad_norm": 8.0,
+      "learning_rate": 4.0582114984887225e-05,
+      "loss": 0.2528,
+      "step": 4010
+    },
+    {
+      "epoch": 0.2917059719904216,
+      "grad_norm": 2.65625,
+      "learning_rate": 4.05371812069862e-05,
+      "loss": 0.199,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2924316087366664,
+      "grad_norm": 5.25,
+      "learning_rate": 4.049216550037841e-05,
+      "loss": 0.1516,
+      "step": 4030
+    },
+    {
+      "epoch": 0.2931572454829113,
+      "grad_norm": 2.40625,
+      "learning_rate": 4.0447068102435015e-05,
+      "loss": 0.1504,
+      "step": 4040
+    },
+    {
+      "epoch": 0.2938828822291561,
+      "grad_norm": 6.9375,
+      "learning_rate": 4.0401889250957926e-05,
+      "loss": 0.2432,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2946085189754009,
+      "grad_norm": 8.8125,
+      "learning_rate": 4.035662918417857e-05,
+      "loss": 0.244,
+      "step": 4060
+    },
+    {
+      "epoch": 0.29533415572164573,
+      "grad_norm": 2.25,
+      "learning_rate": 4.031128814075665e-05,
+      "loss": 0.1402,
+      "step": 4070
+    },
+    {
+      "epoch": 0.29605979246789055,
+      "grad_norm": 8.625,
+      "learning_rate": 4.026586635977882e-05,
+      "loss": 0.2012,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2967854292141354,
+      "grad_norm": 10.75,
+      "learning_rate": 4.02203640807575e-05,
+      "loss": 0.1722,
+      "step": 4090
+    },
+    {
+      "epoch": 0.29751106596038024,
+      "grad_norm": 6.625,
+      "learning_rate": 4.017478154362959e-05,
+      "loss": 0.1729,
+      "step": 4100
+    },
+    {
+      "epoch": 0.29823670270662506,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.012911898875516e-05,
+      "loss": 0.1613,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2989623394528699,
+      "grad_norm": 6.625,
+      "learning_rate": 4.008337665691626e-05,
+      "loss": 0.1887,
+      "step": 4120
+    },
+    {
+      "epoch": 0.29968797619911475,
+      "grad_norm": 5.90625,
+      "learning_rate": 4.0037554789315585e-05,
+      "loss": 0.1853,
+      "step": 4130
+    },
+    {
+      "epoch": 0.30041361294535956,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.999165362757524e-05,
+      "loss": 0.2209,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3011392496916044,
+      "grad_norm": 3.859375,
+      "learning_rate": 3.994567341373544e-05,
+      "loss": 0.256,
+      "step": 4150
+    },
+    {
+      "epoch": 0.3018648864378492,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.9899614390253264e-05,
+      "loss": 0.1986,
+      "step": 4160
+    },
+    {
+      "epoch": 0.302590523184094,
+      "grad_norm": 3.25,
+      "learning_rate": 3.9853476800001356e-05,
+      "loss": 0.1695,
+      "step": 4170
+    },
+    {
+      "epoch": 0.3033161599303389,
+      "grad_norm": 3.125,
+      "learning_rate": 3.980726088626664e-05,
+      "loss": 0.1863,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3040417966765837,
+      "grad_norm": 6.5,
+      "learning_rate": 3.976096689274904e-05,
+      "loss": 0.1599,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3047674334228285,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.9714595063560235e-05,
+      "loss": 0.1489,
+      "step": 4200
+    },
+    {
+      "epoch": 0.30549307016907334,
+      "grad_norm": 0.74609375,
+      "learning_rate": 3.9668145643222296e-05,
+      "loss": 0.1681,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3062187069153182,
+      "grad_norm": 6.3125,
+      "learning_rate": 3.962161887666645e-05,
+      "loss": 0.1861,
+      "step": 4220
+    },
+    {
+      "epoch": 0.30694434366156303,
+      "grad_norm": 9.9375,
+      "learning_rate": 3.9575015009231806e-05,
+      "loss": 0.3147,
+      "step": 4230
+    },
+    {
+      "epoch": 0.30766998040780785,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.952833428666398e-05,
+      "loss": 0.2436,
+      "step": 4240
+    },
+    {
+      "epoch": 0.30839561715405267,
+      "grad_norm": 6.25,
+      "learning_rate": 3.9481576955113897e-05,
+      "loss": 0.2173,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3091212539002975,
+      "grad_norm": 4.78125,
+      "learning_rate": 3.943474326113642e-05,
+      "loss": 0.2422,
+      "step": 4260
+    },
+    {
+      "epoch": 0.30984689064654236,
+      "grad_norm": 6.96875,
+      "learning_rate": 3.938783345168909e-05,
+      "loss": 0.2201,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3105725273927872,
+      "grad_norm": 3.703125,
+      "learning_rate": 3.9340847774130796e-05,
+      "loss": 0.1594,
+      "step": 4280
+    },
+    {
+      "epoch": 0.311298164139032,
+      "grad_norm": 3.390625,
+      "learning_rate": 3.92937864762205e-05,
+      "loss": 0.1946,
+      "step": 4290
+    },
+    {
+      "epoch": 0.3120238008852768,
+      "grad_norm": 2.515625,
+      "learning_rate": 3.92466498061159e-05,
+      "loss": 0.1694,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3127494376315217,
+      "grad_norm": 2.140625,
+      "learning_rate": 3.919943801237217e-05,
+      "loss": 0.2164,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3134750743777665,
+      "grad_norm": 11.0625,
+      "learning_rate": 3.915215134394057e-05,
+      "loss": 0.2352,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3142007111240113,
+      "grad_norm": 6.34375,
+      "learning_rate": 3.910479005016722e-05,
+      "loss": 0.2234,
+      "step": 4330
+    },
+    {
+      "epoch": 0.31492634787025614,
+      "grad_norm": 12.125,
+      "learning_rate": 3.905735438079171e-05,
+      "loss": 0.1806,
+      "step": 4340
+    },
+    {
+      "epoch": 0.31565198461650096,
+      "grad_norm": 5.8125,
+      "learning_rate": 3.900984458594586e-05,
+      "loss": 0.197,
+      "step": 4350
+    },
+    {
+      "epoch": 0.31637762136274583,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.8962260916152315e-05,
+      "loss": 0.143,
+      "step": 4360
+    },
+    {
+      "epoch": 0.31710325810899065,
+      "grad_norm": 6.59375,
+      "learning_rate": 3.8914603622323295e-05,
+      "loss": 0.2315,
+      "step": 4370
+    },
+    {
+      "epoch": 0.31782889485523547,
+      "grad_norm": 0.6640625,
+      "learning_rate": 3.886687295575925e-05,
+      "loss": 0.1557,
+      "step": 4380
+    },
+    {
+      "epoch": 0.3185545316014803,
+      "grad_norm": 4.8125,
+      "learning_rate": 3.88190691681475e-05,
+      "loss": 0.1443,
+      "step": 4390
+    },
+    {
+      "epoch": 0.3192801683477251,
+      "grad_norm": 4.65625,
+      "learning_rate": 3.877119251156097e-05,
+      "loss": 0.1125,
+      "step": 4400
+    },
+    {
+      "epoch": 0.32000580509397,
+      "grad_norm": 9.375,
+      "learning_rate": 3.872324323845681e-05,
+      "loss": 0.2663,
+      "step": 4410
+    },
+    {
+      "epoch": 0.3207314418402148,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.8675221601675085e-05,
+      "loss": 0.2016,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3214570785864596,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.862712785443744e-05,
+      "loss": 0.2524,
+      "step": 4430
+    },
+    {
+      "epoch": 0.32218271533270443,
+      "grad_norm": 7.875,
+      "learning_rate": 3.8578962250345765e-05,
+      "loss": 0.2159,
+      "step": 4440
+    },
+    {
+      "epoch": 0.3229083520789493,
+      "grad_norm": 7.125,
+      "learning_rate": 3.853072504338087e-05,
+      "loss": 0.1617,
+      "step": 4450
+    },
+    {
+      "epoch": 0.3236339888251941,
+      "grad_norm": 7.5,
+      "learning_rate": 3.848241648790109e-05,
+      "loss": 0.2983,
+      "step": 4460
+    },
+    {
+      "epoch": 0.32435962557143894,
+      "grad_norm": 2.125,
+      "learning_rate": 3.843403683864104e-05,
+      "loss": 0.1618,
+      "step": 4470
+    },
+    {
+      "epoch": 0.32508526231768375,
+      "grad_norm": 2.46875,
+      "learning_rate": 3.838558635071018e-05,
+      "loss": 0.1574,
+      "step": 4480
+    },
+    {
+      "epoch": 0.32581089906392857,
+      "grad_norm": 7.5625,
+      "learning_rate": 3.8337065279591536e-05,
+      "loss": 0.1802,
+      "step": 4490
+    },
+    {
+      "epoch": 0.32653653581017344,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.828847388114027e-05,
+      "loss": 0.1615,
+      "step": 4500
+    },
+    {
+      "epoch": 0.32726217255641826,
+      "grad_norm": 6.3125,
+      "learning_rate": 3.823981241158247e-05,
+      "loss": 0.1669,
+      "step": 4510
+    },
+    {
+      "epoch": 0.3279878093026631,
+      "grad_norm": 5.65625,
+      "learning_rate": 3.8191081127513637e-05,
+      "loss": 0.1979,
+      "step": 4520
+    },
+    {
+      "epoch": 0.3287134460489079,
+      "grad_norm": 3.828125,
+      "learning_rate": 3.8142280285897446e-05,
+      "loss": 0.1953,
+      "step": 4530
+    },
+    {
+      "epoch": 0.32943908279515277,
+      "grad_norm": 2.265625,
+      "learning_rate": 3.809341014406436e-05,
+      "loss": 0.1882,
+      "step": 4540
+    },
+    {
+      "epoch": 0.3301647195413976,
+      "grad_norm": 6.125,
+      "learning_rate": 3.8044470959710246e-05,
+      "loss": 0.2276,
+      "step": 4550
+    },
+    {
+      "epoch": 0.3308903562876424,
+      "grad_norm": 11.3125,
+      "learning_rate": 3.799546299089504e-05,
+      "loss": 0.2742,
+      "step": 4560
+    },
+    {
+      "epoch": 0.3316159930338872,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.7946386496041394e-05,
+      "loss": 0.1428,
+      "step": 4570
+    },
+    {
+      "epoch": 0.33234162978013204,
+      "grad_norm": 6.625,
+      "learning_rate": 3.78972417339333e-05,
+      "loss": 0.1944,
+      "step": 4580
+    },
+    {
+      "epoch": 0.3330672665263769,
+      "grad_norm": 2.953125,
+      "learning_rate": 3.784802896371473e-05,
+      "loss": 0.2013,
+      "step": 4590
+    },
+    {
+      "epoch": 0.33379290327262173,
+      "grad_norm": 9.375,
+      "learning_rate": 3.779874844488825e-05,
+      "loss": 0.2329,
+      "step": 4600
+    },
+    {
+      "epoch": 0.33451854001886655,
+      "grad_norm": 7.34375,
+      "learning_rate": 3.774940043731371e-05,
+      "loss": 0.1168,
+      "step": 4610
+    },
+    {
+      "epoch": 0.33524417676511137,
+      "grad_norm": 3.359375,
+      "learning_rate": 3.769998520120678e-05,
+      "loss": 0.1817,
+      "step": 4620
+    },
+    {
+      "epoch": 0.33596981351135624,
+      "grad_norm": 5.625,
+      "learning_rate": 3.765050299713768e-05,
+      "loss": 0.1902,
+      "step": 4630
+    },
+    {
+      "epoch": 0.33669545025760106,
+      "grad_norm": 5.65625,
+      "learning_rate": 3.760095408602973e-05,
+      "loss": 0.1813,
+      "step": 4640
+    },
+    {
+      "epoch": 0.3374210870038459,
+      "grad_norm": 3.546875,
+      "learning_rate": 3.7551338729158e-05,
+      "loss": 0.1854,
+      "step": 4650
+    },
+    {
+      "epoch": 0.3381467237500907,
+      "grad_norm": 5.375,
+      "learning_rate": 3.750165718814794e-05,
+      "loss": 0.1983,
+      "step": 4660
+    },
+    {
+      "epoch": 0.3388723604963355,
+      "grad_norm": 0.9609375,
+      "learning_rate": 3.7451909724974e-05,
+      "loss": 0.147,
+      "step": 4670
+    },
+    {
+      "epoch": 0.3395979972425804,
+      "grad_norm": 7.78125,
+      "learning_rate": 3.740209660195823e-05,
+      "loss": 0.186,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3403236339888252,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.735221808176892e-05,
+      "loss": 0.1677,
+      "step": 4690
+    },
+    {
+      "epoch": 0.34104927073507,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.73022744274192e-05,
+      "loss": 0.1673,
+      "step": 4700
+    },
+    {
+      "epoch": 0.34177490748131484,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.725226590226565e-05,
+      "loss": 0.134,
+      "step": 4710
+    },
+    {
+      "epoch": 0.3425005442275597,
+      "grad_norm": 9.8125,
+      "learning_rate": 3.7202192770006935e-05,
+      "loss": 0.2233,
+      "step": 4720
+    },
+    {
+      "epoch": 0.34322618097380453,
+      "grad_norm": 4.5,
+      "learning_rate": 3.715205529468238e-05,
+      "loss": 0.1979,
+      "step": 4730
+    },
+    {
+      "epoch": 0.34395181772004935,
+      "grad_norm": 2.796875,
+      "learning_rate": 3.71018537406706e-05,
+      "loss": 0.1824,
+      "step": 4740
+    },
+    {
+      "epoch": 0.34467745446629416,
+      "grad_norm": 5.65625,
+      "learning_rate": 3.705158837268813e-05,
+      "loss": 0.2925,
+      "step": 4750
+    },
+    {
+      "epoch": 0.345403091212539,
+      "grad_norm": 5.75,
+      "learning_rate": 3.700125945578795e-05,
+      "loss": 0.2262,
+      "step": 4760
+    },
+    {
+      "epoch": 0.34612872795878386,
+      "grad_norm": 3.03125,
+      "learning_rate": 3.695086725535818e-05,
+      "loss": 0.1748,
+      "step": 4770
+    },
+    {
+      "epoch": 0.3468543647050287,
+      "grad_norm": 6.9375,
+      "learning_rate": 3.6900412037120625e-05,
+      "loss": 0.254,
+      "step": 4780
+    },
+    {
+      "epoch": 0.3475800014512735,
+      "grad_norm": 8.375,
+      "learning_rate": 3.684989406712937e-05,
+      "loss": 0.1996,
+      "step": 4790
+    },
+    {
+      "epoch": 0.3483056381975183,
+      "grad_norm": 5.15625,
+      "learning_rate": 3.679931361176942e-05,
+      "loss": 0.2098,
+      "step": 4800
+    },
+    {
+      "epoch": 0.3490312749437631,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.6748670937755266e-05,
+      "loss": 0.1342,
+      "step": 4810
+    },
+    {
+      "epoch": 0.349756911690008,
+      "grad_norm": 6.0625,
+      "learning_rate": 3.669796631212948e-05,
+      "loss": 0.2532,
+      "step": 4820
+    },
+    {
+      "epoch": 0.3504825484362528,
+      "grad_norm": 1.96875,
+      "learning_rate": 3.6647200002261284e-05,
+      "loss": 0.1901,
+      "step": 4830
+    },
+    {
+      "epoch": 0.35120818518249763,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3.659637227584521e-05,
+      "loss": 0.2048,
+      "step": 4840
+    },
+    {
+      "epoch": 0.35193382192874245,
+      "grad_norm": 3.78125,
+      "learning_rate": 3.6545483400899614e-05,
+      "loss": 0.1548,
+      "step": 4850
+    },
+    {
+      "epoch": 0.3526594586749873,
+      "grad_norm": 8.875,
+      "learning_rate": 3.6494533645765306e-05,
+      "loss": 0.2062,
+      "step": 4860
+    },
+    {
+      "epoch": 0.35338509542123214,
+      "grad_norm": 2.375,
+      "learning_rate": 3.644352327910411e-05,
+      "loss": 0.1933,
+      "step": 4870
+    },
+    {
+      "epoch": 0.35411073216747696,
+      "grad_norm": 11.4375,
+      "learning_rate": 3.6392452569897485e-05,
+      "loss": 0.1588,
+      "step": 4880
+    },
+    {
+      "epoch": 0.3548363689137218,
+      "grad_norm": 9.5625,
+      "learning_rate": 3.634132178744503e-05,
+      "loss": 0.2152,
+      "step": 4890
+    },
+    {
+      "epoch": 0.3555620056599666,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.629013120136316e-05,
+      "loss": 0.1734,
+      "step": 4900
+    },
+    {
+      "epoch": 0.35628764240621147,
+      "grad_norm": 3.96875,
+      "learning_rate": 3.6238881081583636e-05,
+      "loss": 0.2473,
+      "step": 4910
+    },
+    {
+      "epoch": 0.3570132791524563,
+      "grad_norm": 7.0,
+      "learning_rate": 3.618757169835212e-05,
+      "loss": 0.3071,
+      "step": 4920
+    },
+    {
+      "epoch": 0.3577389158987011,
+      "grad_norm": 7.8125,
+      "learning_rate": 3.61362033222268e-05,
+      "loss": 0.2066,
+      "step": 4930
+    },
+    {
+      "epoch": 0.3584645526449459,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.608477622407692e-05,
+      "loss": 0.2513,
+      "step": 4940
+    },
+    {
+      "epoch": 0.3591901893911908,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.603329067508137e-05,
+      "loss": 0.2052,
+      "step": 4950
+    },
+    {
+      "epoch": 0.3599158261374356,
+      "grad_norm": 5.9375,
+      "learning_rate": 3.598174694672727e-05,
+      "loss": 0.2427,
+      "step": 4960
+    },
+    {
+      "epoch": 0.36064146288368043,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.5930145310808514e-05,
+      "loss": 0.1953,
+      "step": 4970
+    },
+    {
+      "epoch": 0.36136709962992525,
+      "grad_norm": 5.15625,
+      "learning_rate": 3.587848603942434e-05,
+      "loss": 0.1994,
+      "step": 4980
+    },
+    {
+      "epoch": 0.36209273637617007,
+      "grad_norm": 5.8125,
+      "learning_rate": 3.582676940497792e-05,
+      "loss": 0.2595,
+      "step": 4990
+    },
+    {
+      "epoch": 0.36281837312241494,
+      "grad_norm": 6.5625,
+      "learning_rate": 3.57749956801749e-05,
+      "loss": 0.1806,
+      "step": 5000
+    },
+    {
+      "epoch": 0.36354400986865976,
+      "grad_norm": 5.4375,
+      "learning_rate": 3.572316513802196e-05,
+      "loss": 0.158,
+      "step": 5010
+    },
+    {
+      "epoch": 0.3642696466149046,
+      "grad_norm": 9.75,
+      "learning_rate": 3.567127805182538e-05,
+      "loss": 0.2198,
+      "step": 5020
+    },
+    {
+      "epoch": 0.3649952833611494,
+      "grad_norm": 7.8125,
+      "learning_rate": 3.5619334695189606e-05,
+      "loss": 0.2098,
+      "step": 5030
+    },
+    {
+      "epoch": 0.36572092010739427,
+      "grad_norm": 9.5625,
+      "learning_rate": 3.556733534201581e-05,
+      "loss": 0.1712,
+      "step": 5040
+    },
+    {
+      "epoch": 0.3664465568536391,
+      "grad_norm": 8.75,
+      "learning_rate": 3.551528026650042e-05,
+      "loss": 0.2482,
+      "step": 5050
+    },
+    {
+      "epoch": 0.3671721935998839,
+      "grad_norm": 6.09375,
+      "learning_rate": 3.5463169743133716e-05,
+      "loss": 0.1205,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3678978303461287,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.541100404669835e-05,
+      "loss": 0.2395,
+      "step": 5070
+    },
+    {
+      "epoch": 0.36862346709237354,
+      "grad_norm": 7.21875,
+      "learning_rate": 3.535878345226789e-05,
+      "loss": 0.1655,
+      "step": 5080
+    },
+    {
+      "epoch": 0.3693491038386184,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.530650823520539e-05,
+      "loss": 0.315,
+      "step": 5090
+    },
+    {
+      "epoch": 0.3700747405848632,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.525417867116195e-05,
+      "loss": 0.1425,
+      "step": 5100
+    },
+    {
+      "epoch": 0.37080037733110804,
+      "grad_norm": 2.875,
+      "learning_rate": 3.5201795036075233e-05,
+      "loss": 0.2102,
+      "step": 5110
+    },
+    {
+      "epoch": 0.37152601407735286,
+      "grad_norm": 6.5625,
+      "learning_rate": 3.514935760616803e-05,
+      "loss": 0.2712,
+      "step": 5120
+    },
+    {
+      "epoch": 0.3722516508235977,
+      "grad_norm": 2.65625,
+      "learning_rate": 3.509686665794679e-05,
+      "loss": 0.1846,
+      "step": 5130
+    },
+    {
+      "epoch": 0.37297728756984255,
+      "grad_norm": 6.34375,
+      "learning_rate": 3.504432246820015e-05,
+      "loss": 0.1419,
+      "step": 5140
+    },
+    {
+      "epoch": 0.37370292431608737,
+      "grad_norm": 5.34375,
+      "learning_rate": 3.499172531399755e-05,
+      "loss": 0.1889,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3744285610623322,
+      "grad_norm": 11.75,
+      "learning_rate": 3.493907547268764e-05,
+      "loss": 0.2545,
+      "step": 5160
+    },
+    {
+      "epoch": 0.375154197808577,
+      "grad_norm": 4.59375,
+      "learning_rate": 3.488637322189695e-05,
+      "loss": 0.1573,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3758798345548219,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.483361883952835e-05,
+      "loss": 0.1936,
+      "step": 5180
+    },
+    {
+      "epoch": 0.3766054713010667,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.47808126037596e-05,
+      "loss": 0.1964,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3773311080473115,
+      "grad_norm": 9.125,
+      "learning_rate": 3.4727954793041886e-05,
+      "loss": 0.2488,
+      "step": 5200
+    },
+    {
+      "epoch": 0.37805674479355633,
+      "grad_norm": 11.125,
+      "learning_rate": 3.4675045686098374e-05,
+      "loss": 0.1915,
+      "step": 5210
+    },
+    {
+      "epoch": 0.37878238153980115,
+      "grad_norm": 2.40625,
+      "learning_rate": 3.462208556192268e-05,
+      "loss": 0.1983,
+      "step": 5220
+    },
+    {
+      "epoch": 0.379508018286046,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.456907469977747e-05,
+      "loss": 0.2398,
+      "step": 5230
+    },
+    {
+      "epoch": 0.38023365503229084,
+      "grad_norm": 6.59375,
+      "learning_rate": 3.451601337919295e-05,
+      "loss": 0.254,
+      "step": 5240
+    },
+    {
+      "epoch": 0.38095929177853566,
+      "grad_norm": 3.03125,
+      "learning_rate": 3.446290187996539e-05,
+      "loss": 0.197,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3816849285247805,
+      "grad_norm": 7.28125,
+      "learning_rate": 3.440974048215564e-05,
+      "loss": 0.2394,
+      "step": 5260
+    },
+    {
+      "epoch": 0.38241056527102535,
+      "grad_norm": 4.625,
+      "learning_rate": 3.4356529466087714e-05,
+      "loss": 0.1847,
+      "step": 5270
+    },
+    {
+      "epoch": 0.38313620201727017,
+      "grad_norm": 10.875,
+      "learning_rate": 3.430326911234722e-05,
+      "loss": 0.2947,
+      "step": 5280
+    },
+    {
+      "epoch": 0.383861838763515,
+      "grad_norm": 7.78125,
+      "learning_rate": 3.4249959701779957e-05,
+      "loss": 0.2134,
+      "step": 5290
+    },
+    {
+      "epoch": 0.3845874755097598,
+      "grad_norm": 9.625,
+      "learning_rate": 3.419660151549038e-05,
+      "loss": 0.1939,
+      "step": 5300
+    },
+    {
+      "epoch": 0.3853131122560046,
+      "grad_norm": 8.4375,
+      "learning_rate": 3.4143194834840164e-05,
+      "loss": 0.2139,
+      "step": 5310
+    },
+    {
+      "epoch": 0.3860387490022495,
+      "grad_norm": 9.0625,
+      "learning_rate": 3.408973994144669e-05,
+      "loss": 0.1718,
+      "step": 5320
+    },
+    {
+      "epoch": 0.3867643857484943,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.4036237117181566e-05,
+      "loss": 0.2513,
+      "step": 5330
+    },
+    {
+      "epoch": 0.38749002249473913,
+      "grad_norm": 7.5625,
+      "learning_rate": 3.398268664416916e-05,
+      "loss": 0.133,
+      "step": 5340
+    },
+    {
+      "epoch": 0.38821565924098395,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.392908880478505e-05,
+      "loss": 0.2633,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3889412959872288,
+      "grad_norm": 4.78125,
+      "learning_rate": 3.3875443881654635e-05,
+      "loss": 0.1924,
+      "step": 5360
+    },
+    {
+      "epoch": 0.38966693273347364,
+      "grad_norm": 2.84375,
+      "learning_rate": 3.3821752157651565e-05,
+      "loss": 0.2403,
+      "step": 5370
+    },
+    {
+      "epoch": 0.39039256947971845,
+      "grad_norm": 8.0,
+      "learning_rate": 3.376801391589626e-05,
+      "loss": 0.196,
+      "step": 5380
+    },
+    {
+      "epoch": 0.39111820622596327,
+      "grad_norm": 8.0,
+      "learning_rate": 3.371422943975447e-05,
+      "loss": 0.2293,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3918438429722081,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.3660399012835705e-05,
+      "loss": 0.1403,
+      "step": 5400
+    },
+    {
+      "epoch": 0.39256947971845296,
+      "grad_norm": 6.84375,
+      "learning_rate": 3.360652291899179e-05,
+      "loss": 0.2718,
+      "step": 5410
+    },
+    {
+      "epoch": 0.3932951164646978,
+      "grad_norm": 7.96875,
+      "learning_rate": 3.355260144231536e-05,
+      "loss": 0.2116,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3940207532109426,
+      "grad_norm": 9.5625,
+      "learning_rate": 3.349863486713835e-05,
+      "loss": 0.2809,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3947463899571874,
+      "grad_norm": 4.375,
+      "learning_rate": 3.3444623478030516e-05,
+      "loss": 0.2067,
+      "step": 5440
+    },
+    {
+      "epoch": 0.3954720267034323,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.33905675597979e-05,
+      "loss": 0.2171,
+      "step": 5450
+    },
+    {
+      "epoch": 0.3961976634496771,
+      "grad_norm": 8.8125,
+      "learning_rate": 3.333646739748137e-05,
+      "loss": 0.1717,
+      "step": 5460
+    },
+    {
+      "epoch": 0.3969233001959219,
+      "grad_norm": 3.53125,
+      "learning_rate": 3.328232327635508e-05,
+      "loss": 0.2451,
+      "step": 5470
+    },
+    {
+      "epoch": 0.39764893694216674,
+      "grad_norm": 6.53125,
+      "learning_rate": 3.3228135481924996e-05,
+      "loss": 0.2168,
+      "step": 5480
+    },
+    {
+      "epoch": 0.39837457368841156,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.3173904299927355e-05,
+      "loss": 0.1968,
+      "step": 5490
+    },
+    {
+      "epoch": 0.39910021043465643,
+      "grad_norm": 7.90625,
+      "learning_rate": 3.311963001632722e-05,
+      "loss": 0.1992,
+      "step": 5500
+    },
+    {
+      "epoch": 0.39982584718090125,
+      "grad_norm": 6.90625,
+      "learning_rate": 3.306531291731688e-05,
+      "loss": 0.1882,
+      "step": 5510
+    },
+    {
+      "epoch": 0.40055148392714607,
+      "grad_norm": 4.15625,
+      "learning_rate": 3.3010953289314435e-05,
+      "loss": 0.2068,
+      "step": 5520
+    },
+    {
+      "epoch": 0.4012771206733909,
+      "grad_norm": 3.15625,
+      "learning_rate": 3.295655141896222e-05,
+      "loss": 0.2522,
+      "step": 5530
+    },
+    {
+      "epoch": 0.4020027574196357,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.290210759312534e-05,
+      "loss": 0.1814,
+      "step": 5540
+    },
+    {
+      "epoch": 0.4027283941658806,
+      "grad_norm": 3.546875,
+      "learning_rate": 3.284762209889009e-05,
+      "loss": 0.2211,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4034540309121254,
+      "grad_norm": 6.9375,
+      "learning_rate": 3.2793095223562545e-05,
+      "loss": 0.149,
+      "step": 5560
+    },
+    {
+      "epoch": 0.4041796676583702,
+      "grad_norm": 2.671875,
+      "learning_rate": 3.2738527254666946e-05,
+      "loss": 0.1788,
+      "step": 5570
+    },
+    {
+      "epoch": 0.40490530440461503,
+      "grad_norm": 7.0625,
+      "learning_rate": 3.268391847994422e-05,
+      "loss": 0.1292,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4056309411508599,
+      "grad_norm": 7.5625,
+      "learning_rate": 3.26292691873505e-05,
+      "loss": 0.2032,
+      "step": 5590
+    },
+    {
+      "epoch": 0.4063565778971047,
+      "grad_norm": 3.640625,
+      "learning_rate": 3.257457966505553e-05,
+      "loss": 0.2506,
+      "step": 5600
+    },
+    {
+      "epoch": 0.40708221464334954,
+      "grad_norm": 4.9375,
+      "learning_rate": 3.251985020144123e-05,
+      "loss": 0.1476,
+      "step": 5610
+    },
+    {
+      "epoch": 0.40780785138959436,
+      "grad_norm": 4.625,
+      "learning_rate": 3.24650810851001e-05,
+      "loss": 0.2202,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4085334881358392,
+      "grad_norm": 10.8125,
+      "learning_rate": 3.241027260483374e-05,
+      "loss": 0.1855,
+      "step": 5630
+    },
+    {
+      "epoch": 0.40925912488208405,
+      "grad_norm": 4.15625,
+      "learning_rate": 3.235542504965134e-05,
+      "loss": 0.2204,
+      "step": 5640
+    },
+    {
+      "epoch": 0.40998476162832886,
+      "grad_norm": 2.0625,
+      "learning_rate": 3.230053870876811e-05,
+      "loss": 0.2307,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4107103983745737,
+      "grad_norm": 5.46875,
+      "learning_rate": 3.22456138716038e-05,
+      "loss": 0.1216,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4114360351208185,
+      "grad_norm": 5.8125,
+      "learning_rate": 3.219065082778113e-05,
+      "loss": 0.1524,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4121616718670634,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.2135649867124296e-05,
+      "loss": 0.1362,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4128873086133082,
+      "grad_norm": 5.59375,
+      "learning_rate": 3.2080611279657444e-05,
+      "loss": 0.1211,
+      "step": 5690
+    },
+    {
+      "epoch": 0.413612945359553,
+      "grad_norm": 8.1875,
+      "learning_rate": 3.202553535560312e-05,
+      "loss": 0.23,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4143385821057978,
+      "grad_norm": 5.71875,
+      "learning_rate": 3.197042238538075e-05,
+      "loss": 0.2395,
+      "step": 5710
+    },
+    {
+      "epoch": 0.41506421885204264,
+      "grad_norm": 5.71875,
+      "learning_rate": 3.19152726596051e-05,
+      "loss": 0.2042,
+      "step": 5720
+    },
+    {
+      "epoch": 0.4157898555982875,
+      "grad_norm": 2.21875,
+      "learning_rate": 3.186008646908476e-05,
+      "loss": 0.1732,
+      "step": 5730
+    },
+    {
+      "epoch": 0.41651549234453233,
+      "grad_norm": 8.125,
+      "learning_rate": 3.18048641048206e-05,
+      "loss": 0.2344,
+      "step": 5740
+    },
+    {
+      "epoch": 0.41724112909077715,
+      "grad_norm": 5.40625,
+      "learning_rate": 3.1749605858004225e-05,
+      "loss": 0.1514,
+      "step": 5750
+    },
+    {
+      "epoch": 0.41796676583702197,
+      "grad_norm": 10.1875,
+      "learning_rate": 3.169431202001648e-05,
+      "loss": 0.2349,
+      "step": 5760
+    },
+    {
+      "epoch": 0.41869240258326684,
+      "grad_norm": 11.875,
+      "learning_rate": 3.1638982882425864e-05,
+      "loss": 0.2691,
+      "step": 5770
+    },
+    {
+      "epoch": 0.41941803932951166,
+      "grad_norm": 13.0,
+      "learning_rate": 3.158361873698699e-05,
+      "loss": 0.1772,
+      "step": 5780
+    },
+    {
+      "epoch": 0.4201436760757565,
+      "grad_norm": 5.4375,
+      "learning_rate": 3.152821987563912e-05,
+      "loss": 0.1288,
+      "step": 5790
+    },
+    {
+      "epoch": 0.4208693128220013,
+      "grad_norm": 5.75,
+      "learning_rate": 3.147278659050455e-05,
+      "loss": 0.2165,
+      "step": 5800
+    },
+    {
+      "epoch": 0.4215949495682461,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.141731917388709e-05,
+      "loss": 0.2248,
+      "step": 5810
+    },
+    {
+      "epoch": 0.422320586314491,
+      "grad_norm": 8.75,
+      "learning_rate": 3.1361817918270534e-05,
+      "loss": 0.2254,
+      "step": 5820
+    },
+    {
+      "epoch": 0.4230462230607358,
+      "grad_norm": 5.53125,
+      "learning_rate": 3.130628311631711e-05,
+      "loss": 0.1899,
+      "step": 5830
+    },
+    {
+      "epoch": 0.4237718598069806,
+      "grad_norm": 2.65625,
+      "learning_rate": 3.1250715060865934e-05,
+      "loss": 0.1949,
+      "step": 5840
+    },
+    {
+      "epoch": 0.42449749655322544,
+      "grad_norm": 7.15625,
+      "learning_rate": 3.11951140449315e-05,
+      "loss": 0.2476,
+      "step": 5850
+    },
+    {
+      "epoch": 0.4252231332994703,
+      "grad_norm": 7.03125,
+      "learning_rate": 3.113948036170206e-05,
+      "loss": 0.1336,
+      "step": 5860
+    },
+    {
+      "epoch": 0.42594877004571513,
+      "grad_norm": 8.8125,
+      "learning_rate": 3.108381430453816e-05,
+      "loss": 0.2252,
+      "step": 5870
+    },
+    {
+      "epoch": 0.42667440679195995,
+      "grad_norm": 4.09375,
+      "learning_rate": 3.102811616697102e-05,
+      "loss": 0.2027,
+      "step": 5880
+    },
+    {
+      "epoch": 0.42740004353820477,
+      "grad_norm": 10.1875,
+      "learning_rate": 3.0972386242701065e-05,
+      "loss": 0.1673,
+      "step": 5890
+    },
+    {
+      "epoch": 0.4281256802844496,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.09166248255963e-05,
+      "loss": 0.1324,
+      "step": 5900
+    },
+    {
+      "epoch": 0.42885131703069446,
+      "grad_norm": 9.0625,
+      "learning_rate": 3.08608322096908e-05,
+      "loss": 0.3027,
+      "step": 5910
+    },
+    {
+      "epoch": 0.4295769537769393,
+      "grad_norm": 4.78125,
+      "learning_rate": 3.080500868918317e-05,
+      "loss": 0.2427,
+      "step": 5920
+    },
+    {
+      "epoch": 0.4303025905231841,
+      "grad_norm": 10.0,
+      "learning_rate": 3.074915455843496e-05,
+      "loss": 0.2219,
+      "step": 5930
+    },
+    {
+      "epoch": 0.4310282272694289,
+      "grad_norm": 7.40625,
+      "learning_rate": 3.069327011196913e-05,
+      "loss": 0.2701,
+      "step": 5940
+    },
+    {
+      "epoch": 0.4317538640156737,
+      "grad_norm": 9.25,
+      "learning_rate": 3.0637355644468524e-05,
+      "loss": 0.1589,
+      "step": 5950
+    },
+    {
+      "epoch": 0.4324795007619186,
+      "grad_norm": 9.0625,
+      "learning_rate": 3.058141145077425e-05,
+      "loss": 0.2039,
+      "step": 5960
+    },
+    {
+      "epoch": 0.4332051375081634,
+      "grad_norm": 3.90625,
+      "learning_rate": 3.0525437825884193e-05,
+      "loss": 0.1976,
+      "step": 5970
+    },
+    {
+      "epoch": 0.43393077425440824,
+      "grad_norm": 2.6875,
+      "learning_rate": 3.0469435064951424e-05,
+      "loss": 0.2134,
+      "step": 5980
+    },
+    {
+      "epoch": 0.43465641100065305,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.041340346328263e-05,
+      "loss": 0.1619,
+      "step": 5990
+    },
+    {
+      "epoch": 0.4353820477468979,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.0357343316336617e-05,
+      "loss": 0.1579,
+      "step": 6000
+    },
+    {
+      "epoch": 0.43610768449314274,
+      "grad_norm": 8.5625,
+      "learning_rate": 3.0301254919722667e-05,
+      "loss": 0.1502,
+      "step": 6010
+    },
+    {
+      "epoch": 0.43683332123938756,
+      "grad_norm": 4.40625,
+      "learning_rate": 3.0245138569199073e-05,
+      "loss": 0.1887,
+      "step": 6020
+    },
+    {
+      "epoch": 0.4375589579856324,
+      "grad_norm": 5.625,
+      "learning_rate": 3.018899456067148e-05,
+      "loss": 0.1968,
+      "step": 6030
+    },
+    {
+      "epoch": 0.4382845947318772,
+      "grad_norm": 5.03125,
+      "learning_rate": 3.0132823190191416e-05,
+      "loss": 0.1644,
+      "step": 6040
+    },
+    {
+      "epoch": 0.43901023147812207,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.0076624753954684e-05,
+      "loss": 0.1948,
+      "step": 6050
+    },
+    {
+      "epoch": 0.4397358682243669,
+      "grad_norm": 10.3125,
+      "learning_rate": 3.0020399548299777e-05,
+      "loss": 0.2043,
+      "step": 6060
+    },
+    {
+      "epoch": 0.4404615049706117,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.9964147869706378e-05,
+      "loss": 0.1854,
+      "step": 6070
+    },
+    {
+      "epoch": 0.4411871417168565,
+      "grad_norm": 2.890625,
+      "learning_rate": 2.990787001479376e-05,
+      "loss": 0.1294,
+      "step": 6080
+    },
+    {
+      "epoch": 0.4419127784631014,
+      "grad_norm": 5.59375,
+      "learning_rate": 2.9851566280319216e-05,
+      "loss": 0.2319,
+      "step": 6090
+    },
+    {
+      "epoch": 0.4426384152093462,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.9795236963176493e-05,
+      "loss": 0.142,
+      "step": 6100
+    },
+    {
+      "epoch": 0.44336405195559103,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.973888236039426e-05,
+      "loss": 0.1708,
+      "step": 6110
+    },
+    {
+      "epoch": 0.44408968870183585,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.9682502769134513e-05,
+      "loss": 0.233,
+      "step": 6120
+    },
+    {
+      "epoch": 0.44481532544808067,
+      "grad_norm": 10.125,
+      "learning_rate": 2.9626098486691e-05,
+      "loss": 0.211,
+      "step": 6130
+    },
+    {
+      "epoch": 0.44554096219432554,
+      "grad_norm": 6.21875,
+      "learning_rate": 2.9569669810487695e-05,
+      "loss": 0.2481,
+      "step": 6140
+    },
+    {
+      "epoch": 0.44626659894057036,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.9513217038077167e-05,
+      "loss": 0.1694,
+      "step": 6150
+    },
+    {
+      "epoch": 0.4469922356868152,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.9456740467139072e-05,
+      "loss": 0.1763,
+      "step": 6160
+    },
+    {
+      "epoch": 0.44771787243306,
+      "grad_norm": 8.875,
+      "learning_rate": 2.940024039547855e-05,
+      "loss": 0.2495,
+      "step": 6170
+    },
+    {
+      "epoch": 0.44844350917930487,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.9343717121024665e-05,
+      "loss": 0.3127,
+      "step": 6180
+    },
+    {
+      "epoch": 0.4491691459255497,
+      "grad_norm": 5.21875,
+      "learning_rate": 2.9287170941828824e-05,
+      "loss": 0.1294,
+      "step": 6190
+    },
+    {
+      "epoch": 0.4498947826717945,
+      "grad_norm": 9.875,
+      "learning_rate": 2.9230602156063215e-05,
+      "loss": 0.25,
+      "step": 6200
+    },
+    {
+      "epoch": 0.4506204194180393,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.917401106201923e-05,
+      "loss": 0.1202,
+      "step": 6210
+    },
+    {
+      "epoch": 0.45134605616428414,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.91173979581059e-05,
+      "loss": 0.1891,
+      "step": 6220
+    },
+    {
+      "epoch": 0.452071692910529,
+      "grad_norm": 9.4375,
+      "learning_rate": 2.9060763142848336e-05,
+      "loss": 0.1584,
+      "step": 6230
+    },
+    {
+      "epoch": 0.45279732965677383,
+      "grad_norm": 5.25,
+      "learning_rate": 2.900410691488607e-05,
+      "loss": 0.2097,
+      "step": 6240
+    },
+    {
+      "epoch": 0.45352296640301865,
+      "grad_norm": 6.4375,
+      "learning_rate": 2.894742957297161e-05,
+      "loss": 0.209,
+      "step": 6250
+    },
+    {
+      "epoch": 0.45424860314926346,
+      "grad_norm": 9.5625,
+      "learning_rate": 2.889073141596878e-05,
+      "loss": 0.2306,
+      "step": 6260
+    },
+    {
+      "epoch": 0.4549742398955083,
+      "grad_norm": 5.84375,
+      "learning_rate": 2.8834012742851146e-05,
+      "loss": 0.2608,
+      "step": 6270
+    },
+    {
+      "epoch": 0.45569987664175315,
+      "grad_norm": 5.40625,
+      "learning_rate": 2.8777273852700488e-05,
+      "loss": 0.1725,
+      "step": 6280
+    },
+    {
+      "epoch": 0.45642551338799797,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.8720515044705154e-05,
+      "loss": 0.1559,
+      "step": 6290
+    },
+    {
+      "epoch": 0.4571511501342428,
+      "grad_norm": 5.09375,
+      "learning_rate": 2.866373661815855e-05,
+      "loss": 0.1735,
+      "step": 6300
+    },
+    {
+      "epoch": 0.4578767868804876,
+      "grad_norm": 9.125,
+      "learning_rate": 2.860693887245752e-05,
+      "loss": 0.2272,
+      "step": 6310
+    },
+    {
+      "epoch": 0.4586024236267325,
+      "grad_norm": 4.5,
+      "learning_rate": 2.8550122107100797e-05,
+      "loss": 0.2697,
+      "step": 6320
+    },
+    {
+      "epoch": 0.4593280603729773,
+      "grad_norm": 5.25,
+      "learning_rate": 2.849328662168737e-05,
+      "loss": 0.2354,
+      "step": 6330
+    },
+    {
+      "epoch": 0.4600536971192221,
+      "grad_norm": 5.46875,
+      "learning_rate": 2.843643271591497e-05,
+      "loss": 0.251,
+      "step": 6340
+    },
+    {
+      "epoch": 0.46077933386546693,
+      "grad_norm": 3.5625,
+      "learning_rate": 2.837956068957845e-05,
+      "loss": 0.1532,
+      "step": 6350
+    },
+    {
+      "epoch": 0.46150497061171175,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.8322670842568217e-05,
+      "loss": 0.1489,
+      "step": 6360
+    },
+    {
+      "epoch": 0.4622306073579566,
+      "grad_norm": 2.0625,
+      "learning_rate": 2.8265763474868644e-05,
+      "loss": 0.2271,
+      "step": 6370
+    },
+    {
+      "epoch": 0.46295624410420144,
+      "grad_norm": 3.53125,
+      "learning_rate": 2.8208838886556495e-05,
+      "loss": 0.2302,
+      "step": 6380
+    },
+    {
+      "epoch": 0.46368188085044626,
+      "grad_norm": 8.0625,
+      "learning_rate": 2.815189737779934e-05,
+      "loss": 0.2382,
+      "step": 6390
+    },
+    {
+      "epoch": 0.4644075175966911,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.8094939248853968e-05,
+      "loss": 0.1812,
+      "step": 6400
+    },
+    {
+      "epoch": 0.46513315434293595,
+      "grad_norm": 11.625,
+      "learning_rate": 2.8037964800064814e-05,
+      "loss": 0.2234,
+      "step": 6410
+    },
+    {
+      "epoch": 0.46585879108918077,
+      "grad_norm": 5.375,
+      "learning_rate": 2.7980974331862363e-05,
+      "loss": 0.1407,
+      "step": 6420
+    },
+    {
+      "epoch": 0.4665844278354256,
+      "grad_norm": 3.125,
+      "learning_rate": 2.7923968144761564e-05,
+      "loss": 0.1561,
+      "step": 6430
+    },
+    {
+      "epoch": 0.4673100645816704,
+      "grad_norm": 5.09375,
+      "learning_rate": 2.7866946539360265e-05,
+      "loss": 0.3223,
+      "step": 6440
+    },
+    {
+      "epoch": 0.4680357013279152,
+      "grad_norm": 4.46875,
+      "learning_rate": 2.7809909816337626e-05,
+      "loss": 0.1804,
+      "step": 6450
+    },
+    {
+      "epoch": 0.4687613380741601,
+      "grad_norm": 9.9375,
+      "learning_rate": 2.7752858276452494e-05,
+      "loss": 0.2061,
+      "step": 6460
+    },
+    {
+      "epoch": 0.4694869748204049,
+      "grad_norm": 3.96875,
+      "learning_rate": 2.7695792220541872e-05,
+      "loss": 0.2215,
+      "step": 6470
+    },
+    {
+      "epoch": 0.47021261156664973,
+      "grad_norm": 5.53125,
+      "learning_rate": 2.7638711949519296e-05,
+      "loss": 0.2286,
+      "step": 6480
+    },
+    {
+      "epoch": 0.47093824831289455,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.758161776437326e-05,
+      "loss": 0.2407,
+      "step": 6490
+    },
+    {
+      "epoch": 0.4716638850591394,
+      "grad_norm": 6.0,
+      "learning_rate": 2.7524509966165622e-05,
+      "loss": 0.2642,
+      "step": 6500
+    },
+    {
+      "epoch": 0.47238952180538424,
+      "grad_norm": 5.0625,
+      "learning_rate": 2.7467388856030048e-05,
+      "loss": 0.1785,
+      "step": 6510
+    },
+    {
+      "epoch": 0.47311515855162906,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.741025473517036e-05,
+      "loss": 0.1266,
+      "step": 6520
+    },
+    {
+      "epoch": 0.4738407952978739,
+      "grad_norm": 7.25,
+      "learning_rate": 2.7353107904859016e-05,
+      "loss": 0.1669,
+      "step": 6530
+    },
+    {
+      "epoch": 0.4745664320441187,
+      "grad_norm": 2.828125,
+      "learning_rate": 2.7295948666435482e-05,
+      "loss": 0.1958,
+      "step": 6540
+    },
+    {
+      "epoch": 0.47529206879036356,
+      "grad_norm": 0.8671875,
+      "learning_rate": 2.723877732130466e-05,
+      "loss": 0.1029,
+      "step": 6550
+    },
+    {
+      "epoch": 0.4760177055366084,
+      "grad_norm": 6.78125,
+      "learning_rate": 2.7181594170935275e-05,
+      "loss": 0.224,
+      "step": 6560
+    },
+    {
+      "epoch": 0.4767433422828532,
+      "grad_norm": 5.59375,
+      "learning_rate": 2.712439951685833e-05,
+      "loss": 0.1665,
+      "step": 6570
+    },
+    {
+      "epoch": 0.477468979029098,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.706719366066544e-05,
+      "loss": 0.1881,
+      "step": 6580
+    },
+    {
+      "epoch": 0.4781946157753429,
+      "grad_norm": 6.40625,
+      "learning_rate": 2.7009976904007346e-05,
+      "loss": 0.2291,
+      "step": 6590
+    },
+    {
+      "epoch": 0.4789202525215877,
+      "grad_norm": 8.5,
+      "learning_rate": 2.6952749548592236e-05,
+      "loss": 0.1509,
+      "step": 6600
+    },
+    {
+      "epoch": 0.4796458892678325,
+      "grad_norm": 10.0625,
+      "learning_rate": 2.689551189618419e-05,
+      "loss": 0.2517,
+      "step": 6610
+    },
+    {
+      "epoch": 0.48037152601407734,
+      "grad_norm": 4.84375,
+      "learning_rate": 2.6838264248601573e-05,
+      "loss": 0.1647,
+      "step": 6620
+    },
+    {
+      "epoch": 0.48109716276032216,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.67810069077155e-05,
+      "loss": 0.1343,
+      "step": 6630
+    },
+    {
+      "epoch": 0.48182279950656703,
+      "grad_norm": 7.8125,
+      "learning_rate": 2.6723740175448136e-05,
+      "loss": 0.2078,
+      "step": 6640
+    },
+    {
+      "epoch": 0.48254843625281185,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.666646435377122e-05,
+      "loss": 0.1813,
+      "step": 6650
+    },
+    {
+      "epoch": 0.48327407299905667,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.6609179744704393e-05,
+      "loss": 0.1698,
+      "step": 6660
+    },
+    {
+      "epoch": 0.4839997097453015,
+      "grad_norm": 8.0625,
+      "learning_rate": 2.655188665031364e-05,
+      "loss": 0.122,
+      "step": 6670
+    },
+    {
+      "epoch": 0.4847253464915463,
+      "grad_norm": 8.5625,
+      "learning_rate": 2.6494585372709683e-05,
+      "loss": 0.1926,
+      "step": 6680
+    },
+    {
+      "epoch": 0.4854509832377912,
+      "grad_norm": 7.28125,
+      "learning_rate": 2.6437276214046418e-05,
+      "loss": 0.2151,
+      "step": 6690
+    },
+    {
+      "epoch": 0.486176619984036,
+      "grad_norm": 5.5625,
+      "learning_rate": 2.637995947651927e-05,
+      "loss": 0.1465,
+      "step": 6700
+    },
+    {
+      "epoch": 0.4869022567302808,
+      "grad_norm": 2.484375,
+      "learning_rate": 2.6322635462363637e-05,
+      "loss": 0.2109,
+      "step": 6710
+    },
+    {
+      "epoch": 0.48762789347652563,
+      "grad_norm": 8.8125,
+      "learning_rate": 2.62653044738533e-05,
+      "loss": 0.25,
+      "step": 6720
+    },
+    {
+      "epoch": 0.4883535302227705,
+      "grad_norm": 7.34375,
+      "learning_rate": 2.620796681329879e-05,
+      "loss": 0.2333,
+      "step": 6730
+    },
+    {
+      "epoch": 0.4890791669690153,
+      "grad_norm": 5.5625,
+      "learning_rate": 2.6150622783045842e-05,
+      "loss": 0.2297,
+      "step": 6740
+    },
+    {
+      "epoch": 0.48980480371526014,
+      "grad_norm": 2.75,
+      "learning_rate": 2.609327268547378e-05,
+      "loss": 0.1751,
+      "step": 6750
+    },
+    {
+      "epoch": 0.49053044046150496,
+      "grad_norm": 4.46875,
+      "learning_rate": 2.6035916822993918e-05,
+      "loss": 0.1583,
+      "step": 6760
+    },
+    {
+      "epoch": 0.4912560772077498,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.5978555498047942e-05,
+      "loss": 0.1878,
+      "step": 6770
+    },
+    {
+      "epoch": 0.49198171395399465,
+      "grad_norm": 2.640625,
+      "learning_rate": 2.592118901310639e-05,
+      "loss": 0.1792,
+      "step": 6780
+    },
+    {
+      "epoch": 0.49270735070023947,
+      "grad_norm": 2.75,
+      "learning_rate": 2.5863817670666967e-05,
+      "loss": 0.1612,
+      "step": 6790
+    },
+    {
+      "epoch": 0.4934329874464843,
+      "grad_norm": 6.96875,
+      "learning_rate": 2.580644177325302e-05,
+      "loss": 0.2044,
+      "step": 6800
+    },
+    {
+      "epoch": 0.4941586241927291,
+      "grad_norm": 6.8125,
+      "learning_rate": 2.5749061623411896e-05,
+      "loss": 0.2831,
+      "step": 6810
+    },
+    {
+      "epoch": 0.494884260938974,
+      "grad_norm": 7.46875,
+      "learning_rate": 2.5691677523713387e-05,
+      "loss": 0.1796,
+      "step": 6820
+    },
+    {
+      "epoch": 0.4956098976852188,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.5634289776748077e-05,
+      "loss": 0.1737,
+      "step": 6830
+    },
+    {
+      "epoch": 0.4963355344314636,
+      "grad_norm": 7.5,
+      "learning_rate": 2.5576898685125822e-05,
+      "loss": 0.2062,
+      "step": 6840
+    },
+    {
+      "epoch": 0.4970611711777084,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.5519504551474095e-05,
+      "loss": 0.1243,
+      "step": 6850
+    },
+    {
+      "epoch": 0.49778680792395325,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.5462107678436396e-05,
+      "loss": 0.209,
+      "step": 6860
+    },
+    {
+      "epoch": 0.4985124446701981,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.54047083686707e-05,
+      "loss": 0.2307,
+      "step": 6870
+    },
+    {
+      "epoch": 0.49923808141644294,
+      "grad_norm": 5.96875,
+      "learning_rate": 2.5347306924847825e-05,
+      "loss": 0.1643,
+      "step": 6880
+    },
+    {
+      "epoch": 0.49996371816268775,
+      "grad_norm": 5.5,
+      "learning_rate": 2.5289903649649816e-05,
+      "loss": 0.212,
+      "step": 6890
+    },
+    {
+      "epoch": 0.5006893549089326,
+      "grad_norm": 6.21875,
+      "learning_rate": 2.52324988457684e-05,
+      "loss": 0.1689,
+      "step": 6900
+    },
+    {
+      "epoch": 0.5014149916551774,
+      "grad_norm": 2.21875,
+      "learning_rate": 2.5175092815903362e-05,
+      "loss": 0.1937,
+      "step": 6910
+    },
+    {
+      "epoch": 0.5021406284014223,
+      "grad_norm": 5.09375,
+      "learning_rate": 2.511768586276093e-05,
+      "loss": 0.1965,
+      "step": 6920
+    },
+    {
+      "epoch": 0.5028662651476671,
+      "grad_norm": 5.09375,
+      "learning_rate": 2.5060278289052247e-05,
+      "loss": 0.1705,
+      "step": 6930
+    },
+    {
+      "epoch": 0.5035919018939119,
+      "grad_norm": 5.25,
+      "learning_rate": 2.500287039749168e-05,
+      "loss": 0.2236,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5043175386401567,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.4945462490795287e-05,
+      "loss": 0.1183,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5050431753864015,
+      "grad_norm": 4.8125,
+      "learning_rate": 2.488805487167922e-05,
+      "loss": 0.2477,
+      "step": 6960
+    },
+    {
+      "epoch": 0.5057688121326464,
+      "grad_norm": 6.28125,
+      "learning_rate": 2.4830647842858106e-05,
+      "loss": 0.1972,
+      "step": 6970
+    },
+    {
+      "epoch": 0.5064944488788913,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.477324170704345e-05,
+      "loss": 0.2312,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5072200856251361,
+      "grad_norm": 4.09375,
+      "learning_rate": 2.4715836766942065e-05,
+      "loss": 0.1873,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5079457223713809,
+      "grad_norm": 8.125,
+      "learning_rate": 2.465843332525444e-05,
+      "loss": 0.1655,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5086713591176257,
+      "grad_norm": 6.46875,
+      "learning_rate": 2.460103168467319e-05,
+      "loss": 0.1788,
+      "step": 7010
+    },
+    {
+      "epoch": 0.5093969958638705,
+      "grad_norm": 6.375,
+      "learning_rate": 2.4543632147881393e-05,
+      "loss": 0.1737,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5101226326101154,
+      "grad_norm": 5.0,
+      "learning_rate": 2.448623501755107e-05,
+      "loss": 0.1951,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5108482693563602,
+      "grad_norm": 4.375,
+      "learning_rate": 2.442884059634153e-05,
+      "loss": 0.1205,
+      "step": 7040
+    },
+    {
+      "epoch": 0.511573906102605,
+      "grad_norm": 6.25,
+      "learning_rate": 2.437144918689781e-05,
+      "loss": 0.2093,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5122995428488498,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.4314061091849063e-05,
+      "loss": 0.1537,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5130251795950947,
+      "grad_norm": 4.34375,
+      "learning_rate": 2.4256676613806957e-05,
+      "loss": 0.2419,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5137508163413396,
+      "grad_norm": 2.609375,
+      "learning_rate": 2.41992960553641e-05,
+      "loss": 0.1428,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5144764530875844,
+      "grad_norm": 3.953125,
+      "learning_rate": 2.4141919719092424e-05,
+      "loss": 0.1902,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5152020898338292,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.4084547907541598e-05,
+      "loss": 0.181,
+      "step": 7100
+    },
+    {
+      "epoch": 0.515927726580074,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.4027180923237423e-05,
+      "loss": 0.2185,
+      "step": 7110
+    },
+    {
+      "epoch": 0.5166533633263188,
+      "grad_norm": 3.75,
+      "learning_rate": 2.3969819068680258e-05,
+      "loss": 0.1705,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5173790000725637,
+      "grad_norm": 5.9375,
+      "learning_rate": 2.3912462646343413e-05,
+      "loss": 0.1645,
+      "step": 7130
+    },
+    {
+      "epoch": 0.5181046368188085,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.3855111958671546e-05,
+      "loss": 0.1872,
+      "step": 7140
+    },
+    {
+      "epoch": 0.5188302735650533,
+      "grad_norm": 6.75,
+      "learning_rate": 2.3797767308079087e-05,
+      "loss": 0.1644,
+      "step": 7150
+    },
+    {
+      "epoch": 0.5195559103112982,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3740428996948604e-05,
+      "loss": 0.2008,
+      "step": 7160
+    },
+    {
+      "epoch": 0.520281547057543,
+      "grad_norm": 9.6875,
+      "learning_rate": 2.3683097327629276e-05,
+      "loss": 0.1746,
+      "step": 7170
+    },
+    {
+      "epoch": 0.5210071838037879,
+      "grad_norm": 4.1875,
+      "learning_rate": 2.3625772602435236e-05,
+      "loss": 0.2044,
+      "step": 7180
+    },
+    {
+      "epoch": 0.5217328205500327,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.3568455123644008e-05,
+      "loss": 0.2134,
+      "step": 7190
+    },
+    {
+      "epoch": 0.5224584572962775,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.351114519349488e-05,
+      "loss": 0.1605,
+      "step": 7200
+    },
+    {
+      "epoch": 0.5231840940425223,
+      "grad_norm": 51.0,
+      "learning_rate": 2.3453843114187373e-05,
+      "loss": 0.2345,
+      "step": 7210
+    },
+    {
+      "epoch": 0.5239097307887671,
+      "grad_norm": 12.3125,
+      "learning_rate": 2.3396549187879595e-05,
+      "loss": 0.2352,
+      "step": 7220
+    },
+    {
+      "epoch": 0.5246353675350119,
+      "grad_norm": 10.5,
+      "learning_rate": 2.3339263716686656e-05,
+      "loss": 0.1863,
+      "step": 7230
+    },
+    {
+      "epoch": 0.5253610042812568,
+      "grad_norm": 13.6875,
+      "learning_rate": 2.3281987002679087e-05,
+      "loss": 0.2164,
+      "step": 7240
+    },
+    {
+      "epoch": 0.5260866410275016,
+      "grad_norm": 5.75,
+      "learning_rate": 2.322471934788124e-05,
+      "loss": 0.228,
+      "step": 7250
+    },
+    {
+      "epoch": 0.5268122777737465,
+      "grad_norm": 2.015625,
+      "learning_rate": 2.3167461054269697e-05,
+      "loss": 0.1691,
+      "step": 7260
+    },
+    {
+      "epoch": 0.5275379145199913,
+      "grad_norm": 7.90625,
+      "learning_rate": 2.311021242377169e-05,
+      "loss": 0.2007,
+      "step": 7270
+    },
+    {
+      "epoch": 0.5282635512662361,
+      "grad_norm": 2.21875,
+      "learning_rate": 2.305297375826347e-05,
+      "loss": 0.2599,
+      "step": 7280
+    },
+    {
+      "epoch": 0.528989188012481,
+      "grad_norm": 6.46875,
+      "learning_rate": 2.2995745359568783e-05,
+      "loss": 0.1182,
+      "step": 7290
+    },
+    {
+      "epoch": 0.5297148247587258,
+      "grad_norm": 7.40625,
+      "learning_rate": 2.293852752945718e-05,
+      "loss": 0.1907,
+      "step": 7300
+    },
+    {
+      "epoch": 0.5304404615049706,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.2881320569642537e-05,
+      "loss": 0.1526,
+      "step": 7310
+    },
+    {
+      "epoch": 0.5311660982512154,
+      "grad_norm": 3.890625,
+      "learning_rate": 2.2824124781781374e-05,
+      "loss": 0.1932,
+      "step": 7320
+    },
+    {
+      "epoch": 0.5318917349974602,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.2766940467471327e-05,
+      "loss": 0.1716,
+      "step": 7330
+    },
+    {
+      "epoch": 0.532617371743705,
+      "grad_norm": 1.3203125,
+      "learning_rate": 2.2709767928249516e-05,
+      "loss": 0.121,
+      "step": 7340
+    },
+    {
+      "epoch": 0.53334300848995,
+      "grad_norm": 9.3125,
+      "learning_rate": 2.2652607465590965e-05,
+      "loss": 0.2614,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5340686452361948,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.259545938090704e-05,
+      "loss": 0.1436,
+      "step": 7360
+    },
+    {
+      "epoch": 0.5347942819824396,
+      "grad_norm": 8.9375,
+      "learning_rate": 2.2538323975543804e-05,
+      "loss": 0.2048,
+      "step": 7370
+    },
+    {
+      "epoch": 0.5355199187286844,
+      "grad_norm": 7.34375,
+      "learning_rate": 2.2481201550780504e-05,
+      "loss": 0.1784,
+      "step": 7380
+    },
+    {
+      "epoch": 0.5362455554749292,
+      "grad_norm": 6.03125,
+      "learning_rate": 2.2424092407827888e-05,
+      "loss": 0.2452,
+      "step": 7390
+    },
+    {
+      "epoch": 0.5369711922211741,
+      "grad_norm": 6.5625,
+      "learning_rate": 2.2366996847826712e-05,
+      "loss": 0.1983,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5376968289674189,
+      "grad_norm": 4.875,
+      "learning_rate": 2.230991517184608e-05,
+      "loss": 0.1566,
+      "step": 7410
+    },
+    {
+      "epoch": 0.5384224657136637,
+      "grad_norm": 11.4375,
+      "learning_rate": 2.22528476808819e-05,
+      "loss": 0.1845,
+      "step": 7420
+    },
+    {
+      "epoch": 0.5391481024599085,
+      "grad_norm": 8.125,
+      "learning_rate": 2.2195794675855285e-05,
+      "loss": 0.2368,
+      "step": 7430
+    },
+    {
+      "epoch": 0.5398737392061534,
+      "grad_norm": 9.1875,
+      "learning_rate": 2.213875645761094e-05,
+      "loss": 0.149,
+      "step": 7440
+    },
+    {
+      "epoch": 0.5405993759523983,
+      "grad_norm": 3.84375,
+      "learning_rate": 2.2081733326915618e-05,
+      "loss": 0.1656,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5413250126986431,
+      "grad_norm": 5.625,
+      "learning_rate": 2.202472558445652e-05,
+      "loss": 0.1711,
+      "step": 7460
+    },
+    {
+      "epoch": 0.5420506494448879,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.1967733530839675e-05,
+      "loss": 0.1567,
+      "step": 7470
+    },
+    {
+      "epoch": 0.5427762861911327,
+      "grad_norm": 11.375,
+      "learning_rate": 2.1910757466588417e-05,
+      "loss": 0.2384,
+      "step": 7480
+    },
+    {
+      "epoch": 0.5435019229373775,
+      "grad_norm": 0.8046875,
+      "learning_rate": 2.1853797692141737e-05,
+      "loss": 0.1774,
+      "step": 7490
+    },
+    {
+      "epoch": 0.5442275596836224,
+      "grad_norm": 3.375,
+      "learning_rate": 2.1796854507852744e-05,
+      "loss": 0.1116,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5449531964298672,
+      "grad_norm": 5.65625,
+      "learning_rate": 2.1739928213987067e-05,
+      "loss": 0.2961,
+      "step": 7510
+    },
+    {
+      "epoch": 0.545678833176112,
+      "grad_norm": 4.90625,
+      "learning_rate": 2.1683019110721275e-05,
+      "loss": 0.1824,
+      "step": 7520
+    },
+    {
+      "epoch": 0.5464044699223569,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.1626127498141276e-05,
+      "loss": 0.1549,
+      "step": 7530
+    },
+    {
+      "epoch": 0.5471301066686017,
+      "grad_norm": 3.90625,
+      "learning_rate": 2.1569253676240746e-05,
+      "loss": 0.2524,
+      "step": 7540
+    },
+    {
+      "epoch": 0.5478557434148466,
+      "grad_norm": 2.5625,
+      "learning_rate": 2.151239794491958e-05,
+      "loss": 0.2434,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5485813801610914,
+      "grad_norm": 3.78125,
+      "learning_rate": 2.1455560603982243e-05,
+      "loss": 0.1429,
+      "step": 7560
+    },
+    {
+      "epoch": 0.5493070169073362,
+      "grad_norm": 5.625,
+      "learning_rate": 2.1398741953136253e-05,
+      "loss": 0.1429,
+      "step": 7570
+    },
+    {
+      "epoch": 0.550032653653581,
+      "grad_norm": 7.5,
+      "learning_rate": 2.1341942291990547e-05,
+      "loss": 0.1771,
+      "step": 7580
+    },
+    {
+      "epoch": 0.5507582903998258,
+      "grad_norm": 5.53125,
+      "learning_rate": 2.1285161920053954e-05,
+      "loss": 0.1868,
+      "step": 7590
+    },
+    {
+      "epoch": 0.5514839271460706,
+      "grad_norm": 3.90625,
+      "learning_rate": 2.1228401136733565e-05,
+      "loss": 0.1809,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5522095638923155,
+      "grad_norm": 8.75,
+      "learning_rate": 2.1171660241333215e-05,
+      "loss": 0.1386,
+      "step": 7610
+    },
+    {
+      "epoch": 0.5529352006385604,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.111493953305183e-05,
+      "loss": 0.11,
+      "step": 7620
+    },
+    {
+      "epoch": 0.5536608373848052,
+      "grad_norm": 3.609375,
+      "learning_rate": 2.1058239310981903e-05,
+      "loss": 0.1261,
+      "step": 7630
+    },
+    {
+      "epoch": 0.55438647413105,
+      "grad_norm": 7.375,
+      "learning_rate": 2.100155987410792e-05,
+      "loss": 0.2369,
+      "step": 7640
+    },
+    {
+      "epoch": 0.5551121108772948,
+      "grad_norm": 6.09375,
+      "learning_rate": 2.0944901521304726e-05,
+      "loss": 0.2068,
+      "step": 7650
+    },
+    {
+      "epoch": 0.5558377476235397,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.088826455133603e-05,
+      "loss": 0.2407,
+      "step": 7660
+    },
+    {
+      "epoch": 0.5565633843697845,
+      "grad_norm": 5.03125,
+      "learning_rate": 2.0831649262852748e-05,
+      "loss": 0.2114,
+      "step": 7670
+    },
+    {
+      "epoch": 0.5572890211160293,
+      "grad_norm": 8.125,
+      "learning_rate": 2.07750559543915e-05,
+      "loss": 0.2068,
+      "step": 7680
+    },
+    {
+      "epoch": 0.5580146578622741,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.0718484924372986e-05,
+      "loss": 0.2249,
+      "step": 7690
+    },
+    {
+      "epoch": 0.5587402946085189,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.0661936471100435e-05,
+      "loss": 0.2185,
+      "step": 7700
+    },
+    {
+      "epoch": 0.5594659313547639,
+      "grad_norm": 8.625,
+      "learning_rate": 2.0605410892758037e-05,
+      "loss": 0.2553,
+      "step": 7710
+    },
+    {
+      "epoch": 0.5601915681010087,
+      "grad_norm": 6.875,
+      "learning_rate": 2.054890848740934e-05,
+      "loss": 0.1944,
+      "step": 7720
+    },
+    {
+      "epoch": 0.5609172048472535,
+      "grad_norm": 1.2421875,
+      "learning_rate": 2.0492429552995715e-05,
+      "loss": 0.1832,
+      "step": 7730
+    },
+    {
+      "epoch": 0.5616428415934983,
+      "grad_norm": 11.1875,
+      "learning_rate": 2.0435974387334764e-05,
+      "loss": 0.1693,
+      "step": 7740
+    },
+    {
+      "epoch": 0.5623684783397431,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.0379543288118753e-05,
+      "loss": 0.2008,
+      "step": 7750
+    },
+    {
+      "epoch": 0.563094115085988,
+      "grad_norm": 7.75,
+      "learning_rate": 2.032313655291306e-05,
+      "loss": 0.1959,
+      "step": 7760
+    },
+    {
+      "epoch": 0.5638197518322328,
+      "grad_norm": 4.125,
+      "learning_rate": 2.0266754479154556e-05,
+      "loss": 0.1698,
+      "step": 7770
+    },
+    {
+      "epoch": 0.5645453885784776,
+      "grad_norm": 6.4375,
+      "learning_rate": 2.0210397364150093e-05,
+      "loss": 0.1593,
+      "step": 7780
+    },
+    {
+      "epoch": 0.5652710253247224,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.0154065505074916e-05,
+      "loss": 0.1852,
+      "step": 7790
+    },
+    {
+      "epoch": 0.5659966620709673,
+      "grad_norm": 2.65625,
+      "learning_rate": 2.0097759198971096e-05,
+      "loss": 0.1614,
+      "step": 7800
+    },
+    {
+      "epoch": 0.5667222988172121,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.0041478742745937e-05,
+      "loss": 0.1781,
+      "step": 7810
+    },
+    {
+      "epoch": 0.567447935563457,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.998522443317046e-05,
+      "loss": 0.1959,
+      "step": 7820
+    },
+    {
+      "epoch": 0.5681735723097018,
+      "grad_norm": 8.625,
+      "learning_rate": 1.992899656687781e-05,
+      "loss": 0.2641,
+      "step": 7830
+    },
+    {
+      "epoch": 0.5688992090559466,
+      "grad_norm": 6.75,
+      "learning_rate": 1.987279544036168e-05,
+      "loss": 0.1612,
+      "step": 7840
+    },
+    {
+      "epoch": 0.5696248458021914,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.981662134997478e-05,
+      "loss": 0.136,
+      "step": 7850
+    },
+    {
+      "epoch": 0.5703504825484362,
+      "grad_norm": 10.625,
+      "learning_rate": 1.9760474591927234e-05,
+      "loss": 0.1985,
+      "step": 7860
+    },
+    {
+      "epoch": 0.571076119294681,
+      "grad_norm": 4.75,
+      "learning_rate": 1.9704355462285057e-05,
+      "loss": 0.1835,
+      "step": 7870
+    },
+    {
+      "epoch": 0.5718017560409259,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.9648264256968577e-05,
+      "loss": 0.1418,
+      "step": 7880
+    },
+    {
+      "epoch": 0.5725273927871708,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.9592201271750878e-05,
+      "loss": 0.2299,
+      "step": 7890
+    },
+    {
+      "epoch": 0.5732530295334156,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.9536166802256222e-05,
+      "loss": 0.1803,
+      "step": 7900
+    },
+    {
+      "epoch": 0.5739786662796604,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.9480161143958518e-05,
+      "loss": 0.1992,
+      "step": 7910
+    },
+    {
+      "epoch": 0.5747043030259052,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.942418459217976e-05,
+      "loss": 0.2701,
+      "step": 7920
+    },
+    {
+      "epoch": 0.5754299397721501,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.936823744208843e-05,
+      "loss": 0.1887,
+      "step": 7930
+    },
+    {
+      "epoch": 0.5761555765183949,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.931231998869801e-05,
+      "loss": 0.1242,
+      "step": 7940
+    },
+    {
+      "epoch": 0.5768812132646397,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.9256432526865357e-05,
+      "loss": 0.1883,
+      "step": 7950
+    },
+    {
+      "epoch": 0.5776068500108845,
+      "grad_norm": 5.90625,
+      "learning_rate": 1.9200575351289207e-05,
+      "loss": 0.1577,
+      "step": 7960
+    },
+    {
+      "epoch": 0.5783324867571293,
+      "grad_norm": 11.875,
+      "learning_rate": 1.914474875650857e-05,
+      "loss": 0.1777,
+      "step": 7970
+    },
+    {
+      "epoch": 0.5790581235033742,
+      "grad_norm": 5.6875,
+      "learning_rate": 1.9088953036901222e-05,
+      "loss": 0.1961,
+      "step": 7980
+    },
+    {
+      "epoch": 0.5797837602496191,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.903318848668212e-05,
+      "loss": 0.1668,
+      "step": 7990
+    },
+    {
+      "epoch": 0.5805093969958639,
+      "grad_norm": 6.5,
+      "learning_rate": 1.8977455399901862e-05,
+      "loss": 0.1485,
+      "step": 8000
+    },
+    {
+      "epoch": 0.5812350337421087,
+      "grad_norm": 2.875,
+      "learning_rate": 1.8921754070445148e-05,
+      "loss": 0.1537,
+      "step": 8010
+    },
+    {
+      "epoch": 0.5819606704883535,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.8866084792029208e-05,
+      "loss": 0.1492,
+      "step": 8020
+    },
+    {
+      "epoch": 0.5826863072345984,
+      "grad_norm": 3.703125,
+      "learning_rate": 1.8810447858202266e-05,
+      "loss": 0.1362,
+      "step": 8030
+    },
+    {
+      "epoch": 0.5834119439808432,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.8754843562342012e-05,
+      "loss": 0.2008,
+      "step": 8040
+    },
+    {
+      "epoch": 0.584137580727088,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.8699272197653992e-05,
+      "loss": 0.124,
+      "step": 8050
+    },
+    {
+      "epoch": 0.5848632174733328,
+      "grad_norm": 10.0,
+      "learning_rate": 1.8643734057170125e-05,
+      "loss": 0.1914,
+      "step": 8060
+    },
+    {
+      "epoch": 0.5855888542195776,
+      "grad_norm": 8.625,
+      "learning_rate": 1.858822943374715e-05,
+      "loss": 0.228,
+      "step": 8070
+    },
+    {
+      "epoch": 0.5863144909658226,
+      "grad_norm": 9.25,
+      "learning_rate": 1.853275862006505e-05,
+      "loss": 0.2581,
+      "step": 8080
+    },
+    {
+      "epoch": 0.5870401277120674,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.847732190862553e-05,
+      "loss": 0.1816,
+      "step": 8090
+    },
+    {
+      "epoch": 0.5877657644583122,
+      "grad_norm": 8.5,
+      "learning_rate": 1.8421919591750487e-05,
+      "loss": 0.2577,
+      "step": 8100
+    },
+    {
+      "epoch": 0.588491401204557,
+      "grad_norm": 5.5,
+      "learning_rate": 1.8366551961580415e-05,
+      "loss": 0.1603,
+      "step": 8110
+    },
+    {
+      "epoch": 0.5892170379508018,
+      "grad_norm": 6.375,
+      "learning_rate": 1.8311219310072948e-05,
+      "loss": 0.2544,
+      "step": 8120
+    },
+    {
+      "epoch": 0.5899426746970466,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.8255921929001254e-05,
+      "loss": 0.1548,
+      "step": 8130
+    },
+    {
+      "epoch": 0.5906683114432915,
+      "grad_norm": 4.28125,
+      "learning_rate": 1.8200660109952504e-05,
+      "loss": 0.1921,
+      "step": 8140
+    },
+    {
+      "epoch": 0.5913939481895363,
+      "grad_norm": 3.625,
+      "learning_rate": 1.8145434144326364e-05,
+      "loss": 0.2011,
+      "step": 8150
+    },
+    {
+      "epoch": 0.5921195849357811,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.8090244323333444e-05,
+      "loss": 0.1765,
+      "step": 8160
+    },
+    {
+      "epoch": 0.592845221682026,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.8035090937993754e-05,
+      "loss": 0.1672,
+      "step": 8170
+    },
+    {
+      "epoch": 0.5935708584282708,
+      "grad_norm": 5.53125,
+      "learning_rate": 1.7979974279135174e-05,
+      "loss": 0.2116,
+      "step": 8180
+    },
+    {
+      "epoch": 0.5942964951745157,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.7924894637391943e-05,
+      "loss": 0.2011,
+      "step": 8190
+    },
+    {
+      "epoch": 0.5950221319207605,
+      "grad_norm": 4.15625,
+      "learning_rate": 1.7869852303203065e-05,
+      "loss": 0.1901,
+      "step": 8200
+    },
+    {
+      "epoch": 0.5957477686670053,
+      "grad_norm": 8.375,
+      "learning_rate": 1.7814847566810856e-05,
+      "loss": 0.2242,
+      "step": 8210
+    },
+    {
+      "epoch": 0.5964734054132501,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.775988071825937e-05,
+      "loss": 0.1885,
+      "step": 8220
+    },
+    {
+      "epoch": 0.5971990421594949,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.770495204739287e-05,
+      "loss": 0.1521,
+      "step": 8230
+    },
+    {
+      "epoch": 0.5979246789057397,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.765006184385428e-05,
+      "loss": 0.2152,
+      "step": 8240
+    },
+    {
+      "epoch": 0.5986503156519846,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.759521039708372e-05,
+      "loss": 0.17,
+      "step": 8250
+    },
+    {
+      "epoch": 0.5993759523982295,
+      "grad_norm": 6.0,
+      "learning_rate": 1.754039799631694e-05,
+      "loss": 0.1901,
+      "step": 8260
+    },
+    {
+      "epoch": 0.6001015891444743,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.7485624930583775e-05,
+      "loss": 0.201,
+      "step": 8270
+    },
+    {
+      "epoch": 0.6008272258907191,
+      "grad_norm": 5.6875,
+      "learning_rate": 1.743089148870666e-05,
+      "loss": 0.1556,
+      "step": 8280
+    },
+    {
+      "epoch": 0.601552862636964,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.7376197959299094e-05,
+      "loss": 0.1426,
+      "step": 8290
+    },
+    {
+      "epoch": 0.6022784993832088,
+      "grad_norm": 10.625,
+      "learning_rate": 1.7321544630764088e-05,
+      "loss": 0.1558,
+      "step": 8300
+    },
+    {
+      "epoch": 0.6030041361294536,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.7266931791292708e-05,
+      "loss": 0.1678,
+      "step": 8310
+    },
+    {
+      "epoch": 0.6037297728756984,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.7212359728862486e-05,
+      "loss": 0.2296,
+      "step": 8320
+    },
+    {
+      "epoch": 0.6044554096219432,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.7157828731235958e-05,
+      "loss": 0.1766,
+      "step": 8330
+    },
+    {
+      "epoch": 0.605181046368188,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.7103339085959102e-05,
+      "loss": 0.1777,
+      "step": 8340
+    },
+    {
+      "epoch": 0.605906683114433,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.704889108035986e-05,
+      "loss": 0.2133,
+      "step": 8350
+    },
+    {
+      "epoch": 0.6066323198606778,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.6994485001546588e-05,
+      "loss": 0.239,
+      "step": 8360
+    },
+    {
+      "epoch": 0.6073579566069226,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.694012113640657e-05,
+      "loss": 0.1416,
+      "step": 8370
+    },
+    {
+      "epoch": 0.6080835933531674,
+      "grad_norm": 2.59375,
+      "learning_rate": 1.6885799771604498e-05,
+      "loss": 0.1283,
+      "step": 8380
+    },
+    {
+      "epoch": 0.6088092300994122,
+      "grad_norm": 5.40625,
+      "learning_rate": 1.6831521193580944e-05,
+      "loss": 0.3281,
+      "step": 8390
+    },
+    {
+      "epoch": 0.609534866845657,
+      "grad_norm": 7.59375,
+      "learning_rate": 1.677728568855087e-05,
+      "loss": 0.2566,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6102605035919019,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.6723093542502112e-05,
+      "loss": 0.1879,
+      "step": 8410
+    },
+    {
+      "epoch": 0.6109861403381467,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.6668945041193867e-05,
+      "loss": 0.1202,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6117117770843915,
+      "grad_norm": 9.75,
+      "learning_rate": 1.6614840470155187e-05,
+      "loss": 0.1762,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6124374138306364,
+      "grad_norm": 1.25,
+      "learning_rate": 1.656078011468348e-05,
+      "loss": 0.1612,
+      "step": 8440
+    },
+    {
+      "epoch": 0.6131630505768813,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.6506764259843004e-05,
+      "loss": 0.1288,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6138886873231261,
+      "grad_norm": 5.75,
+      "learning_rate": 1.645279319046336e-05,
+      "loss": 0.1915,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6146143240693709,
+      "grad_norm": 4.0,
+      "learning_rate": 1.6398867191138005e-05,
+      "loss": 0.1868,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6153399608156157,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.6344986546222706e-05,
+      "loss": 0.2249,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6160655975618605,
+      "grad_norm": 9.125,
+      "learning_rate": 1.6291151539834104e-05,
+      "loss": 0.2665,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6167912343081053,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.6237362455848176e-05,
+      "loss": 0.2199,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6175168710543502,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.6183619577898746e-05,
+      "loss": 0.1631,
+      "step": 8510
+    },
+    {
+      "epoch": 0.618242507800595,
+      "grad_norm": 2.625,
+      "learning_rate": 1.6129923189375966e-05,
+      "loss": 0.1938,
+      "step": 8520
+    },
+    {
+      "epoch": 0.6189681445468399,
+      "grad_norm": 4.28125,
+      "learning_rate": 1.6076273573424886e-05,
+      "loss": 0.1662,
+      "step": 8530
+    },
+    {
+      "epoch": 0.6196937812930847,
+      "grad_norm": 7.5,
+      "learning_rate": 1.6022671012943906e-05,
+      "loss": 0.2011,
+      "step": 8540
+    },
+    {
+      "epoch": 0.6204194180393295,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5969115790583284e-05,
+      "loss": 0.1496,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6211450547855744,
+      "grad_norm": 6.75,
+      "learning_rate": 1.591560818874368e-05,
+      "loss": 0.228,
+      "step": 8560
+    },
+    {
+      "epoch": 0.6218706915318192,
+      "grad_norm": 5.375,
+      "learning_rate": 1.5862148489574635e-05,
+      "loss": 0.2003,
+      "step": 8570
+    },
+    {
+      "epoch": 0.622596328278064,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.5808736974973105e-05,
+      "loss": 0.127,
+      "step": 8580
+    },
+    {
+      "epoch": 0.6233219650243088,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.5755373926581963e-05,
+      "loss": 0.2397,
+      "step": 8590
+    },
+    {
+      "epoch": 0.6240476017705536,
+      "grad_norm": 5.125,
+      "learning_rate": 1.5702059625788512e-05,
+      "loss": 0.1294,
+      "step": 8600
+    },
+    {
+      "epoch": 0.6247732385167984,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.5648794353723e-05,
+      "loss": 0.1277,
+      "step": 8610
+    },
+    {
+      "epoch": 0.6254988752630434,
+      "grad_norm": 7.5,
+      "learning_rate": 1.5595578391257153e-05,
+      "loss": 0.2329,
+      "step": 8620
+    },
+    {
+      "epoch": 0.6262245120092882,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.5542412019002684e-05,
+      "loss": 0.3033,
+      "step": 8630
+    },
+    {
+      "epoch": 0.626950148755533,
+      "grad_norm": 8.375,
+      "learning_rate": 1.5489295517309808e-05,
+      "loss": 0.2644,
+      "step": 8640
+    },
+    {
+      "epoch": 0.6276757855017778,
+      "grad_norm": 5.15625,
+      "learning_rate": 1.543622916626577e-05,
+      "loss": 0.2097,
+      "step": 8650
+    },
+    {
+      "epoch": 0.6284014222480226,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5383213245693373e-05,
+      "loss": 0.1933,
+      "step": 8660
+    },
+    {
+      "epoch": 0.6291270589942675,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.5330248035149484e-05,
+      "loss": 0.1481,
+      "step": 8670
+    },
+    {
+      "epoch": 0.6298526957405123,
+      "grad_norm": 12.0,
+      "learning_rate": 1.5277333813923582e-05,
+      "loss": 0.1359,
+      "step": 8680
+    },
+    {
+      "epoch": 0.6305783324867571,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.5224470861036278e-05,
+      "loss": 0.2397,
+      "step": 8690
+    },
+    {
+      "epoch": 0.6313039692330019,
+      "grad_norm": 3.0,
+      "learning_rate": 1.5171659455237836e-05,
+      "loss": 0.1727,
+      "step": 8700
+    },
+    {
+      "epoch": 0.6320296059792467,
+      "grad_norm": 5.15625,
+      "learning_rate": 1.5118899875006692e-05,
+      "loss": 0.1759,
+      "step": 8710
+    },
+    {
+      "epoch": 0.6327552427254917,
+      "grad_norm": 5.75,
+      "learning_rate": 1.5066192398548041e-05,
+      "loss": 0.2449,
+      "step": 8720
+    },
+    {
+      "epoch": 0.6334808794717365,
+      "grad_norm": 7.84375,
+      "learning_rate": 1.5013537303792286e-05,
+      "loss": 0.1926,
+      "step": 8730
+    },
+    {
+      "epoch": 0.6342065162179813,
+      "grad_norm": 8.0,
+      "learning_rate": 1.4960934868393655e-05,
+      "loss": 0.1831,
+      "step": 8740
+    },
+    {
+      "epoch": 0.6349321529642261,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.4908385369728689e-05,
+      "loss": 0.1875,
+      "step": 8750
+    },
+    {
+      "epoch": 0.6356577897104709,
+      "grad_norm": 10.0,
+      "learning_rate": 1.485588908489477e-05,
+      "loss": 0.2325,
+      "step": 8760
+    },
+    {
+      "epoch": 0.6363834264567158,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.48034462907087e-05,
+      "loss": 0.2062,
+      "step": 8770
+    },
+    {
+      "epoch": 0.6371090632029606,
+      "grad_norm": 7.6875,
+      "learning_rate": 1.475105726370522e-05,
+      "loss": 0.1271,
+      "step": 8780
+    },
+    {
+      "epoch": 0.6378346999492054,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.4698722280135551e-05,
+      "loss": 0.1704,
+      "step": 8790
+    },
+    {
+      "epoch": 0.6385603366954502,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.4646441615965922e-05,
+      "loss": 0.1826,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6392859734416951,
+      "grad_norm": 1.6640625,
+      "learning_rate": 1.4594215546876149e-05,
+      "loss": 0.145,
+      "step": 8810
+    },
+    {
+      "epoch": 0.64001161018794,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.4542044348258172e-05,
+      "loss": 0.1631,
+      "step": 8820
+    },
+    {
+      "epoch": 0.6407372469341848,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.4489928295214577e-05,
+      "loss": 0.2639,
+      "step": 8830
+    },
+    {
+      "epoch": 0.6414628836804296,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.4437867662557156e-05,
+      "loss": 0.2504,
+      "step": 8840
+    },
+    {
+      "epoch": 0.6421885204266744,
+      "grad_norm": 5.0625,
+      "learning_rate": 1.4385862724805505e-05,
+      "loss": 0.137,
+      "step": 8850
+    },
+    {
+      "epoch": 0.6429141571729192,
+      "grad_norm": 2.75,
+      "learning_rate": 1.4333913756185496e-05,
+      "loss": 0.1598,
+      "step": 8860
+    },
+    {
+      "epoch": 0.643639793919164,
+      "grad_norm": 2.546875,
+      "learning_rate": 1.4282021030627878e-05,
+      "loss": 0.1439,
+      "step": 8870
+    },
+    {
+      "epoch": 0.6443654306654089,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.4230184821766867e-05,
+      "loss": 0.1037,
+      "step": 8880
+    },
+    {
+      "epoch": 0.6450910674116537,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.41784054029386e-05,
+      "loss": 0.1449,
+      "step": 8890
+    },
+    {
+      "epoch": 0.6458167041578986,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4126683047179781e-05,
+      "loss": 0.1113,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6465423409041434,
+      "grad_norm": 6.5625,
+      "learning_rate": 1.4075018027226245e-05,
+      "loss": 0.1827,
+      "step": 8910
+    },
+    {
+      "epoch": 0.6472679776503882,
+      "grad_norm": 10.25,
+      "learning_rate": 1.4023410615511456e-05,
+      "loss": 0.2141,
+      "step": 8920
+    },
+    {
+      "epoch": 0.647993614396633,
+      "grad_norm": 5.90625,
+      "learning_rate": 1.3971861084165108e-05,
+      "loss": 0.1588,
+      "step": 8930
+    },
+    {
+      "epoch": 0.6487192511428779,
+      "grad_norm": 5.75,
+      "learning_rate": 1.3920369705011716e-05,
+      "loss": 0.1935,
+      "step": 8940
+    },
+    {
+      "epoch": 0.6494448878891227,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.386893674956912e-05,
+      "loss": 0.2316,
+      "step": 8950
+    },
+    {
+      "epoch": 0.6501705246353675,
+      "grad_norm": 9.375,
+      "learning_rate": 1.3817562489047095e-05,
+      "loss": 0.1794,
+      "step": 8960
+    },
+    {
+      "epoch": 0.6508961613816123,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.3766247194345944e-05,
+      "loss": 0.1333,
+      "step": 8970
+    },
+    {
+      "epoch": 0.6516217981278571,
+      "grad_norm": 4.25,
+      "learning_rate": 1.3714991136054998e-05,
+      "loss": 0.2158,
+      "step": 8980
+    },
+    {
+      "epoch": 0.6523474348741021,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3663794584451251e-05,
+      "loss": 0.1486,
+      "step": 8990
+    },
+    {
+      "epoch": 0.6530730716203469,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.3612657809497909e-05,
+      "loss": 0.182,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6537987083665917,
+      "grad_norm": 5.875,
+      "learning_rate": 1.356158108084297e-05,
+      "loss": 0.1069,
+      "step": 9010
+    },
+    {
+      "epoch": 0.6545243451128365,
+      "grad_norm": 9.0,
+      "learning_rate": 1.3510564667817798e-05,
+      "loss": 0.2026,
+      "step": 9020
+    },
+    {
+      "epoch": 0.6552499818590813,
+      "grad_norm": 8.875,
+      "learning_rate": 1.3459608839435737e-05,
+      "loss": 0.2062,
+      "step": 9030
+    },
+    {
+      "epoch": 0.6559756186053262,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.3408713864390629e-05,
+      "loss": 0.2019,
+      "step": 9040
+    },
+    {
+      "epoch": 0.656701255351571,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.3357880011055437e-05,
+      "loss": 0.1006,
+      "step": 9050
+    },
+    {
+      "epoch": 0.6574268920978158,
+      "grad_norm": 2.25,
+      "learning_rate": 1.3307107547480858e-05,
+      "loss": 0.1384,
+      "step": 9060
+    },
+    {
+      "epoch": 0.6581525288440606,
+      "grad_norm": 12.625,
+      "learning_rate": 1.3256396741393829e-05,
+      "loss": 0.1971,
+      "step": 9070
+    },
+    {
+      "epoch": 0.6588781655903055,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3205747860196197e-05,
+      "loss": 0.1604,
+      "step": 9080
+    },
+    {
+      "epoch": 0.6596038023365504,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.3155161170963254e-05,
+      "loss": 0.2457,
+      "step": 9090
+    },
+    {
+      "epoch": 0.6603294390827952,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.310463694044236e-05,
+      "loss": 0.12,
+      "step": 9100
+    },
+    {
+      "epoch": 0.66105507582904,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.3054175435051518e-05,
+      "loss": 0.1872,
+      "step": 9110
+    },
+    {
+      "epoch": 0.6617807125752848,
+      "grad_norm": 0.625,
+      "learning_rate": 1.3003776920877996e-05,
+      "loss": 0.2188,
+      "step": 9120
+    },
+    {
+      "epoch": 0.6625063493215296,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.295344166367688e-05,
+      "loss": 0.1991,
+      "step": 9130
+    },
+    {
+      "epoch": 0.6632319860677744,
+      "grad_norm": 3.609375,
+      "learning_rate": 1.2903169928869702e-05,
+      "loss": 0.18,
+      "step": 9140
+    },
+    {
+      "epoch": 0.6639576228140193,
+      "grad_norm": 7.15625,
+      "learning_rate": 1.2852961981543057e-05,
+      "loss": 0.1694,
+      "step": 9150
+    },
+    {
+      "epoch": 0.6646832595602641,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.280281808644716e-05,
+      "loss": 0.1939,
+      "step": 9160
+    },
+    {
+      "epoch": 0.665408896306509,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.2752738507994483e-05,
+      "loss": 0.1745,
+      "step": 9170
+    },
+    {
+      "epoch": 0.6661345330527538,
+      "grad_norm": 5.75,
+      "learning_rate": 1.2702723510258349e-05,
+      "loss": 0.1604,
+      "step": 9180
+    },
+    {
+      "epoch": 0.6668601697989986,
+      "grad_norm": 8.875,
+      "learning_rate": 1.2652773356971537e-05,
+      "loss": 0.1785,
+      "step": 9190
+    },
+    {
+      "epoch": 0.6675858065452435,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.2602888311524918e-05,
+      "loss": 0.223,
+      "step": 9200
+    },
+    {
+      "epoch": 0.6683114432914883,
+      "grad_norm": 6.0,
+      "learning_rate": 1.255306863696602e-05,
+      "loss": 0.1985,
+      "step": 9210
+    },
+    {
+      "epoch": 0.6690370800377331,
+      "grad_norm": 6.875,
+      "learning_rate": 1.2503314595997684e-05,
+      "loss": 0.1434,
+      "step": 9220
+    },
+    {
+      "epoch": 0.6697627167839779,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.2453626450976633e-05,
+      "loss": 0.1696,
+      "step": 9230
+    },
+    {
+      "epoch": 0.6704883535302227,
+      "grad_norm": 5.75,
+      "learning_rate": 1.2404004463912155e-05,
+      "loss": 0.233,
+      "step": 9240
+    },
+    {
+      "epoch": 0.6712139902764676,
+      "grad_norm": 5.15625,
+      "learning_rate": 1.2354448896464654e-05,
+      "loss": 0.2062,
+      "step": 9250
+    },
+    {
+      "epoch": 0.6719396270227125,
+      "grad_norm": 6.0625,
+      "learning_rate": 1.2304960009944305e-05,
+      "loss": 0.2004,
+      "step": 9260
+    },
+    {
+      "epoch": 0.6726652637689573,
+      "grad_norm": 4.75,
+      "learning_rate": 1.2255538065309671e-05,
+      "loss": 0.1774,
+      "step": 9270
+    },
+    {
+      "epoch": 0.6733909005152021,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.2206183323166314e-05,
+      "loss": 0.1483,
+      "step": 9280
+    },
+    {
+      "epoch": 0.6741165372614469,
+      "grad_norm": 4.0,
+      "learning_rate": 1.2156896043765465e-05,
+      "loss": 0.1664,
+      "step": 9290
+    },
+    {
+      "epoch": 0.6748421740076918,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2107676487002581e-05,
+      "loss": 0.2178,
+      "step": 9300
+    },
+    {
+      "epoch": 0.6755678107539366,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.2058524912416027e-05,
+      "loss": 0.2353,
+      "step": 9310
+    },
+    {
+      "epoch": 0.6762934475001814,
+      "grad_norm": 5.8125,
+      "learning_rate": 1.200944157918571e-05,
+      "loss": 0.2511,
+      "step": 9320
+    },
+    {
+      "epoch": 0.6770190842464262,
+      "grad_norm": 14.25,
+      "learning_rate": 1.196042674613167e-05,
+      "loss": 0.2214,
+      "step": 9330
+    },
+    {
+      "epoch": 0.677744720992671,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.1911480671712751e-05,
+      "loss": 0.1299,
+      "step": 9340
+    },
+    {
+      "epoch": 0.678470357738916,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.1862603614025228e-05,
+      "loss": 0.1668,
+      "step": 9350
+    },
+    {
+      "epoch": 0.6791959944851608,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.1813795830801441e-05,
+      "loss": 0.2543,
+      "step": 9360
+    },
+    {
+      "epoch": 0.6799216312314056,
+      "grad_norm": 5.25,
+      "learning_rate": 1.1765057579408442e-05,
+      "loss": 0.1904,
+      "step": 9370
+    },
+    {
+      "epoch": 0.6806472679776504,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.1716389116846652e-05,
+      "loss": 0.2138,
+      "step": 9380
+    },
+    {
+      "epoch": 0.6813729047238952,
+      "grad_norm": 6.0,
+      "learning_rate": 1.1667790699748466e-05,
+      "loss": 0.1659,
+      "step": 9390
+    },
+    {
+      "epoch": 0.68209854147014,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.1619262584376931e-05,
+      "loss": 0.197,
+      "step": 9400
+    },
+    {
+      "epoch": 0.6828241782163849,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.157080502662441e-05,
+      "loss": 0.1717,
+      "step": 9410
+    },
+    {
+      "epoch": 0.6835498149626297,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.152241828201118e-05,
+      "loss": 0.1194,
+      "step": 9420
+    },
+    {
+      "epoch": 0.6842754517088745,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.147410260568412e-05,
+      "loss": 0.208,
+      "step": 9430
+    },
+    {
+      "epoch": 0.6850010884551194,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.1425858252415387e-05,
+      "loss": 0.1023,
+      "step": 9440
+    },
+    {
+      "epoch": 0.6857267252013642,
+      "grad_norm": 5.125,
+      "learning_rate": 1.1377685476601026e-05,
+      "loss": 0.1496,
+      "step": 9450
+    },
+    {
+      "epoch": 0.6864523619476091,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.1329584532259626e-05,
+      "loss": 0.1409,
+      "step": 9460
+    },
+    {
+      "epoch": 0.6871779986938539,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.1281555673031052e-05,
+      "loss": 0.1495,
+      "step": 9470
+    },
+    {
+      "epoch": 0.6879036354400987,
+      "grad_norm": 3.921875,
+      "learning_rate": 1.1233599152175025e-05,
+      "loss": 0.15,
+      "step": 9480
+    },
+    {
+      "epoch": 0.6886292721863435,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.1185715222569823e-05,
+      "loss": 0.2487,
+      "step": 9490
+    },
+    {
+      "epoch": 0.6893549089325883,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.113790413671097e-05,
+      "loss": 0.1871,
+      "step": 9500
+    },
+    {
+      "epoch": 0.6900805456788331,
+      "grad_norm": 8.25,
+      "learning_rate": 1.1090166146709848e-05,
+      "loss": 0.1591,
+      "step": 9510
+    },
+    {
+      "epoch": 0.690806182425078,
+      "grad_norm": 9.25,
+      "learning_rate": 1.104250150429241e-05,
+      "loss": 0.1563,
+      "step": 9520
+    },
+    {
+      "epoch": 0.6915318191713228,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.0994910460797856e-05,
+      "loss": 0.1366,
+      "step": 9530
+    },
+    {
+      "epoch": 0.6922574559175677,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.0947393267177286e-05,
+      "loss": 0.1957,
+      "step": 9540
+    },
+    {
+      "epoch": 0.6929830926638125,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.0899950173992348e-05,
+      "loss": 0.1197,
+      "step": 9550
+    },
+    {
+      "epoch": 0.6937087294100573,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.0852581431414019e-05,
+      "loss": 0.1813,
+      "step": 9560
+    },
+    {
+      "epoch": 0.6944343661563022,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.080528728922117e-05,
+      "loss": 0.1226,
+      "step": 9570
+    },
+    {
+      "epoch": 0.695160002902547,
+      "grad_norm": 16.0,
+      "learning_rate": 1.0758067996799308e-05,
+      "loss": 0.2175,
+      "step": 9580
+    },
+    {
+      "epoch": 0.6958856396487918,
+      "grad_norm": 5.375,
+      "learning_rate": 1.0710923803139275e-05,
+      "loss": 0.1879,
+      "step": 9590
+    },
+    {
+      "epoch": 0.6966112763950366,
+      "grad_norm": 5.59375,
+      "learning_rate": 1.0663854956835884e-05,
+      "loss": 0.1496,
+      "step": 9600
+    },
+    {
+      "epoch": 0.6973369131412814,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.061686170608664e-05,
+      "loss": 0.2154,
+      "step": 9610
+    },
+    {
+      "epoch": 0.6980625498875263,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.0569944298690452e-05,
+      "loss": 0.2213,
+      "step": 9620
+    },
+    {
+      "epoch": 0.6987881866337712,
+      "grad_norm": 4.75,
+      "learning_rate": 1.0523102982046268e-05,
+      "loss": 0.1694,
+      "step": 9630
+    },
+    {
+      "epoch": 0.699513823380016,
+      "grad_norm": 3.6875,
+      "learning_rate": 1.0476338003151822e-05,
+      "loss": 0.1486,
+      "step": 9640
+    },
+    {
+      "epoch": 0.7002394601262608,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.042964960860231e-05,
+      "loss": 0.221,
+      "step": 9650
+    },
+    {
+      "epoch": 0.7009650968725056,
+      "grad_norm": 6.0,
+      "learning_rate": 1.0383038044589094e-05,
+      "loss": 0.2547,
+      "step": 9660
+    },
+    {
+      "epoch": 0.7016907336187505,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.0336503556898391e-05,
+      "loss": 0.1532,
+      "step": 9670
+    },
+    {
+      "epoch": 0.7024163703649953,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.0290046390910019e-05,
+      "loss": 0.1184,
+      "step": 9680
+    },
+    {
+      "epoch": 0.7031420071112401,
+      "grad_norm": 2.65625,
+      "learning_rate": 1.024366679159604e-05,
+      "loss": 0.1041,
+      "step": 9690
+    },
+    {
+      "epoch": 0.7038676438574849,
+      "grad_norm": 2.625,
+      "learning_rate": 1.0197365003519513e-05,
+      "loss": 0.1293,
+      "step": 9700
+    },
+    {
+      "epoch": 0.7045932806037297,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.0151141270833209e-05,
+      "loss": 0.2057,
+      "step": 9710
+    },
+    {
+      "epoch": 0.7053189173499747,
+      "grad_norm": 3.5,
+      "learning_rate": 1.0104995837278284e-05,
+      "loss": 0.1669,
+      "step": 9720
+    },
+    {
+      "epoch": 0.7060445540962195,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.0058928946183022e-05,
+      "loss": 0.0862,
+      "step": 9730
+    },
+    {
+      "epoch": 0.7067701908424643,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.0012940840461554e-05,
+      "loss": 0.1759,
+      "step": 9740
+    },
+    {
+      "epoch": 0.7074958275887091,
+      "grad_norm": 3.5,
+      "learning_rate": 9.967031762612566e-06,
+      "loss": 0.1729,
+      "step": 9750
+    },
+    {
+      "epoch": 0.7082214643349539,
+      "grad_norm": 7.71875,
+      "learning_rate": 9.921201954718013e-06,
+      "loss": 0.2038,
+      "step": 9760
+    },
+    {
+      "epoch": 0.7089471010811987,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.875451658441886e-06,
+      "loss": 0.1564,
+      "step": 9770
+    },
+    {
+      "epoch": 0.7096727378274436,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.829781115028872e-06,
+      "loss": 0.158,
+      "step": 9780
+    },
+    {
+      "epoch": 0.7103983745736884,
+      "grad_norm": 6.0625,
+      "learning_rate": 9.784190565303122e-06,
+      "loss": 0.2061,
+      "step": 9790
+    },
+    {
+      "epoch": 0.7111240113199332,
+      "grad_norm": 4.78125,
+      "learning_rate": 9.738680249666996e-06,
+      "loss": 0.1574,
+      "step": 9800
+    },
+    {
+      "epoch": 0.7118496480661781,
+      "grad_norm": 11.5,
+      "learning_rate": 9.693250408099746e-06,
+      "loss": 0.2073,
+      "step": 9810
+    },
+    {
+      "epoch": 0.7125752848124229,
+      "grad_norm": 7.125,
+      "learning_rate": 9.647901280156294e-06,
+      "loss": 0.1601,
+      "step": 9820
+    },
+    {
+      "epoch": 0.7133009215586678,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.602633104965946e-06,
+      "loss": 0.2388,
+      "step": 9830
+    },
+    {
+      "epoch": 0.7140265583049126,
+      "grad_norm": 2.671875,
+      "learning_rate": 9.557446121231145e-06,
+      "loss": 0.1949,
+      "step": 9840
+    },
+    {
+      "epoch": 0.7147521950511574,
+      "grad_norm": 6.6875,
+      "learning_rate": 9.512340567226189e-06,
+      "loss": 0.2427,
+      "step": 9850
+    },
+    {
+      "epoch": 0.7154778317974022,
+      "grad_norm": 9.6875,
+      "learning_rate": 9.467316680796023e-06,
+      "loss": 0.2315,
+      "step": 9860
+    },
+    {
+      "epoch": 0.716203468543647,
+      "grad_norm": 10.0,
+      "learning_rate": 9.422374699354922e-06,
+      "loss": 0.2032,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7169291052898918,
+      "grad_norm": 10.3125,
+      "learning_rate": 9.377514859885278e-06,
+      "loss": 0.2365,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7176547420361367,
+      "grad_norm": 2.796875,
+      "learning_rate": 9.332737398936356e-06,
+      "loss": 0.2075,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7183803787823816,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.288042552623019e-06,
+      "loss": 0.2223,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7191060155286264,
+      "grad_norm": 10.8125,
+      "learning_rate": 9.243430556624492e-06,
+      "loss": 0.2116,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7198316522748712,
+      "grad_norm": 4.25,
+      "learning_rate": 9.198901646183155e-06,
+      "loss": 0.2051,
+      "step": 9920
+    },
+    {
+      "epoch": 0.720557289021116,
+      "grad_norm": 9.625,
+      "learning_rate": 9.154456056103217e-06,
+      "loss": 0.2134,
+      "step": 9930
+    },
+    {
+      "epoch": 0.7212829257673609,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.110094020749591e-06,
+      "loss": 0.1396,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7220085625136057,
+      "grad_norm": 7.21875,
+      "learning_rate": 9.065815774046563e-06,
+      "loss": 0.1919,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7227341992598505,
+      "grad_norm": 6.9375,
+      "learning_rate": 9.021621549476608e-06,
+      "loss": 0.2076,
+      "step": 9960
+    },
+    {
+      "epoch": 0.7234598360060953,
+      "grad_norm": 5.0625,
+      "learning_rate": 8.977511580079137e-06,
+      "loss": 0.1682,
+      "step": 9970
+    },
+    {
+      "epoch": 0.7241854727523401,
+      "grad_norm": 4.625,
+      "learning_rate": 8.933486098449298e-06,
+      "loss": 0.1318,
+      "step": 9980
+    },
+    {
+      "epoch": 0.7249111094985851,
+      "grad_norm": 5.125,
+      "learning_rate": 8.889545336736706e-06,
+      "loss": 0.1931,
+      "step": 9990
+    },
+    {
+      "epoch": 0.7256367462448299,
+      "grad_norm": 2.296875,
+      "learning_rate": 8.845689526644247e-06,
+      "loss": 0.1896,
+      "step": 10000
+    },
+    {
+      "epoch": 0.7263623829910747,
+      "grad_norm": 4.1875,
+      "learning_rate": 8.801918899426872e-06,
+      "loss": 0.1065,
+      "step": 10010
+    },
+    {
+      "epoch": 0.7270880197373195,
+      "grad_norm": 6.6875,
+      "learning_rate": 8.758233685890313e-06,
+      "loss": 0.1789,
+      "step": 10020
+    },
+    {
+      "epoch": 0.7278136564835643,
+      "grad_norm": 7.65625,
+      "learning_rate": 8.714634116389955e-06,
+      "loss": 0.1374,
+      "step": 10030
+    },
+    {
+      "epoch": 0.7285392932298091,
+      "grad_norm": 4.78125,
+      "learning_rate": 8.671120420829548e-06,
+      "loss": 0.1704,
+      "step": 10040
+    },
+    {
+      "epoch": 0.729264929976054,
+      "grad_norm": 5.9375,
+      "learning_rate": 8.627692828660025e-06,
+      "loss": 0.2178,
+      "step": 10050
+    },
+    {
+      "epoch": 0.7299905667222988,
+      "grad_norm": 4.78125,
+      "learning_rate": 8.58435156887831e-06,
+      "loss": 0.1657,
+      "step": 10060
+    },
+    {
+      "epoch": 0.7307162034685436,
+      "grad_norm": 7.46875,
+      "learning_rate": 8.541096870026064e-06,
+      "loss": 0.1946,
+      "step": 10070
+    },
+    {
+      "epoch": 0.7314418402147885,
+      "grad_norm": 4.59375,
+      "learning_rate": 8.497928960188518e-06,
+      "loss": 0.157,
+      "step": 10080
+    },
+    {
+      "epoch": 0.7321674769610333,
+      "grad_norm": 10.0,
+      "learning_rate": 8.454848066993246e-06,
+      "loss": 0.1984,
+      "step": 10090
+    },
+    {
+      "epoch": 0.7328931137072782,
+      "grad_norm": 15.125,
+      "learning_rate": 8.411854417609007e-06,
+      "loss": 0.2094,
+      "step": 10100
+    },
+    {
+      "epoch": 0.733618750453523,
+      "grad_norm": 2.734375,
+      "learning_rate": 8.36894823874449e-06,
+      "loss": 0.1452,
+      "step": 10110
+    },
+    {
+      "epoch": 0.7343443871997678,
+      "grad_norm": 6.375,
+      "learning_rate": 8.326129756647146e-06,
+      "loss": 0.1803,
+      "step": 10120
+    },
+    {
+      "epoch": 0.7350700239460126,
+      "grad_norm": 1.421875,
+      "learning_rate": 8.283399197102007e-06,
+      "loss": 0.1944,
+      "step": 10130
+    },
+    {
+      "epoch": 0.7357956606922574,
+      "grad_norm": 5.78125,
+      "learning_rate": 8.240756785430464e-06,
+      "loss": 0.2076,
+      "step": 10140
+    },
+    {
+      "epoch": 0.7365212974385023,
+      "grad_norm": 6.0625,
+      "learning_rate": 8.198202746489133e-06,
+      "loss": 0.2228,
+      "step": 10150
+    },
+    {
+      "epoch": 0.7372469341847471,
+      "grad_norm": 8.125,
+      "learning_rate": 8.155737304668602e-06,
+      "loss": 0.1868,
+      "step": 10160
+    },
+    {
+      "epoch": 0.737972570930992,
+      "grad_norm": 2.21875,
+      "learning_rate": 8.113360683892282e-06,
+      "loss": 0.2,
+      "step": 10170
+    },
+    {
+      "epoch": 0.7386982076772368,
+      "grad_norm": 10.0625,
+      "learning_rate": 8.071073107615235e-06,
+      "loss": 0.1695,
+      "step": 10180
+    },
+    {
+      "epoch": 0.7394238444234816,
+      "grad_norm": 2.703125,
+      "learning_rate": 8.028874798822989e-06,
+      "loss": 0.142,
+      "step": 10190
+    },
+    {
+      "epoch": 0.7401494811697265,
+      "grad_norm": 3.96875,
+      "learning_rate": 7.986765980030348e-06,
+      "loss": 0.153,
+      "step": 10200
+    },
+    {
+      "epoch": 0.7408751179159713,
+      "grad_norm": 6.90625,
+      "learning_rate": 7.944746873280223e-06,
+      "loss": 0.2906,
+      "step": 10210
+    },
+    {
+      "epoch": 0.7416007546622161,
+      "grad_norm": 6.0625,
+      "learning_rate": 7.902817700142484e-06,
+      "loss": 0.1856,
+      "step": 10220
+    },
+    {
+      "epoch": 0.7423263914084609,
+      "grad_norm": 10.375,
+      "learning_rate": 7.860978681712753e-06,
+      "loss": 0.1286,
+      "step": 10230
+    },
+    {
+      "epoch": 0.7430520281547057,
+      "grad_norm": 9.8125,
+      "learning_rate": 7.819230038611285e-06,
+      "loss": 0.2498,
+      "step": 10240
+    },
+    {
+      "epoch": 0.7437776649009505,
+      "grad_norm": 4.90625,
+      "learning_rate": 7.777571990981764e-06,
+      "loss": 0.1425,
+      "step": 10250
+    },
+    {
+      "epoch": 0.7445033016471954,
+      "grad_norm": 8.5,
+      "learning_rate": 7.736004758490143e-06,
+      "loss": 0.2038,
+      "step": 10260
+    },
+    {
+      "epoch": 0.7452289383934403,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.69452856032354e-06,
+      "loss": 0.24,
+      "step": 10270
+    },
+    {
+      "epoch": 0.7459545751396851,
+      "grad_norm": 5.8125,
+      "learning_rate": 7.653143615189007e-06,
+      "loss": 0.2234,
+      "step": 10280
+    },
+    {
+      "epoch": 0.7466802118859299,
+      "grad_norm": 4.03125,
+      "learning_rate": 7.611850141312424e-06,
+      "loss": 0.1559,
+      "step": 10290
+    },
+    {
+      "epoch": 0.7474058486321747,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.570648356437335e-06,
+      "loss": 0.2113,
+      "step": 10300
+    },
+    {
+      "epoch": 0.7481314853784196,
+      "grad_norm": 7.9375,
+      "learning_rate": 7.5295384778238024e-06,
+      "loss": 0.1753,
+      "step": 10310
+    },
+    {
+      "epoch": 0.7488571221246644,
+      "grad_norm": 1.9296875,
+      "learning_rate": 7.488520722247252e-06,
+      "loss": 0.1852,
+      "step": 10320
+    },
+    {
+      "epoch": 0.7495827588709092,
+      "grad_norm": 9.25,
+      "learning_rate": 7.44759530599736e-06,
+      "loss": 0.1824,
+      "step": 10330
+    },
+    {
+      "epoch": 0.750308395617154,
+      "grad_norm": 2.140625,
+      "learning_rate": 7.406762444876872e-06,
+      "loss": 0.1725,
+      "step": 10340
+    },
+    {
+      "epoch": 0.7510340323633988,
+      "grad_norm": 6.3125,
+      "learning_rate": 7.366022354200477e-06,
+      "loss": 0.1907,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7517596691096438,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.325375248793703e-06,
+      "loss": 0.2317,
+      "step": 10360
+    },
+    {
+      "epoch": 0.7524853058558886,
+      "grad_norm": 8.75,
+      "learning_rate": 7.284821342991743e-06,
+      "loss": 0.1978,
+      "step": 10370
+    },
+    {
+      "epoch": 0.7532109426021334,
+      "grad_norm": 7.65625,
+      "learning_rate": 7.244360850638332e-06,
+      "loss": 0.2462,
+      "step": 10380
+    },
+    {
+      "epoch": 0.7539365793483782,
+      "grad_norm": 2.0,
+      "learning_rate": 7.20399398508467e-06,
+      "loss": 0.124,
+      "step": 10390
+    },
+    {
+      "epoch": 0.754662216094623,
+      "grad_norm": 6.21875,
+      "learning_rate": 7.1637209591882e-06,
+      "loss": 0.2236,
+      "step": 10400
+    },
+    {
+      "epoch": 0.7553878528408678,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.123541985311569e-06,
+      "loss": 0.1782,
+      "step": 10410
+    },
+    {
+      "epoch": 0.7561134895871127,
+      "grad_norm": 1.046875,
+      "learning_rate": 7.083457275321501e-06,
+      "loss": 0.1548,
+      "step": 10420
+    },
+    {
+      "epoch": 0.7568391263333575,
+      "grad_norm": 7.84375,
+      "learning_rate": 7.043467040587626e-06,
+      "loss": 0.2244,
+      "step": 10430
+    },
+    {
+      "epoch": 0.7575647630796023,
+      "grad_norm": 5.75,
+      "learning_rate": 7.0035714919814064e-06,
+      "loss": 0.2411,
+      "step": 10440
+    },
+    {
+      "epoch": 0.7582903998258472,
+      "grad_norm": 2.671875,
+      "learning_rate": 6.963770839875042e-06,
+      "loss": 0.1699,
+      "step": 10450
+    },
+    {
+      "epoch": 0.759016036572092,
+      "grad_norm": 3.453125,
+      "learning_rate": 6.924065294140306e-06,
+      "loss": 0.196,
+      "step": 10460
+    },
+    {
+      "epoch": 0.7597416733183369,
+      "grad_norm": 4.34375,
+      "learning_rate": 6.884455064147474e-06,
+      "loss": 0.1544,
+      "step": 10470
+    },
+    {
+      "epoch": 0.7604673100645817,
+      "grad_norm": 16.875,
+      "learning_rate": 6.844940358764246e-06,
+      "loss": 0.2223,
+      "step": 10480
+    },
+    {
+      "epoch": 0.7611929468108265,
+      "grad_norm": 6.375,
+      "learning_rate": 6.805521386354563e-06,
+      "loss": 0.1572,
+      "step": 10490
+    },
+    {
+      "epoch": 0.7619185835570713,
+      "grad_norm": 10.625,
+      "learning_rate": 6.766198354777589e-06,
+      "loss": 0.2224,
+      "step": 10500
+    },
+    {
+      "epoch": 0.7626442203033161,
+      "grad_norm": 6.53125,
+      "learning_rate": 6.726971471386592e-06,
+      "loss": 0.2001,
+      "step": 10510
+    },
+    {
+      "epoch": 0.763369857049561,
+      "grad_norm": 5.4375,
+      "learning_rate": 6.6878409430278264e-06,
+      "loss": 0.2215,
+      "step": 10520
+    },
+    {
+      "epoch": 0.7640954937958058,
+      "grad_norm": 2.328125,
+      "learning_rate": 6.6488069760394595e-06,
+      "loss": 0.1484,
+      "step": 10530
+    },
+    {
+      "epoch": 0.7648211305420507,
+      "grad_norm": 8.3125,
+      "learning_rate": 6.6098697762504995e-06,
+      "loss": 0.3203,
+      "step": 10540
+    },
+    {
+      "epoch": 0.7655467672882955,
+      "grad_norm": 8.5625,
+      "learning_rate": 6.5710295489796795e-06,
+      "loss": 0.1871,
+      "step": 10550
+    },
+    {
+      "epoch": 0.7662724040345403,
+      "grad_norm": 9.875,
+      "learning_rate": 6.5322864990343814e-06,
+      "loss": 0.2259,
+      "step": 10560
+    },
+    {
+      "epoch": 0.7669980407807852,
+      "grad_norm": 9.6875,
+      "learning_rate": 6.4936408307096045e-06,
+      "loss": 0.1612,
+      "step": 10570
+    },
+    {
+      "epoch": 0.76772367752703,
+      "grad_norm": 5.40625,
+      "learning_rate": 6.455092747786795e-06,
+      "loss": 0.2162,
+      "step": 10580
+    },
+    {
+      "epoch": 0.7684493142732748,
+      "grad_norm": 5.46875,
+      "learning_rate": 6.4166424535328424e-06,
+      "loss": 0.1547,
+      "step": 10590
+    },
+    {
+      "epoch": 0.7691749510195196,
+      "grad_norm": 10.9375,
+      "learning_rate": 6.378290150699018e-06,
+      "loss": 0.2242,
+      "step": 10600
+    },
+    {
+      "epoch": 0.7699005877657644,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.340036041519839e-06,
+      "loss": 0.1452,
+      "step": 10610
+    },
+    {
+      "epoch": 0.7706262245120092,
+      "grad_norm": 4.5625,
+      "learning_rate": 6.301880327712053e-06,
+      "loss": 0.1707,
+      "step": 10620
+    },
+    {
+      "epoch": 0.7713518612582542,
+      "grad_norm": 4.9375,
+      "learning_rate": 6.263823210473576e-06,
+      "loss": 0.1881,
+      "step": 10630
+    },
+    {
+      "epoch": 0.772077498004499,
+      "grad_norm": 5.875,
+      "learning_rate": 6.2258648904824e-06,
+      "loss": 0.1559,
+      "step": 10640
+    },
+    {
+      "epoch": 0.7728031347507438,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.188005567895547e-06,
+      "loss": 0.1434,
+      "step": 10650
+    },
+    {
+      "epoch": 0.7735287714969886,
+      "grad_norm": 5.5625,
+      "learning_rate": 6.15024544234804e-06,
+      "loss": 0.1507,
+      "step": 10660
+    },
+    {
+      "epoch": 0.7742544082432334,
+      "grad_norm": 9.25,
+      "learning_rate": 6.112584712951821e-06,
+      "loss": 0.1173,
+      "step": 10670
+    },
+    {
+      "epoch": 0.7749800449894783,
+      "grad_norm": 9.0625,
+      "learning_rate": 6.075023578294678e-06,
+      "loss": 0.2496,
+      "step": 10680
+    },
+    {
+      "epoch": 0.7757056817357231,
+      "grad_norm": 8.1875,
+      "learning_rate": 6.037562236439282e-06,
+      "loss": 0.2226,
+      "step": 10690
+    },
+    {
+      "epoch": 0.7764313184819679,
+      "grad_norm": 4.9375,
+      "learning_rate": 6.0002008849220496e-06,
+      "loss": 0.1602,
+      "step": 10700
+    },
+    {
+      "epoch": 0.7771569552282127,
+      "grad_norm": 15.75,
+      "learning_rate": 5.96293972075215e-06,
+      "loss": 0.3226,
+      "step": 10710
+    },
+    {
+      "epoch": 0.7778825919744576,
+      "grad_norm": 6.125,
+      "learning_rate": 5.92577894041047e-06,
+      "loss": 0.209,
+      "step": 10720
+    },
+    {
+      "epoch": 0.7786082287207025,
+      "grad_norm": 3.453125,
+      "learning_rate": 5.888718739848553e-06,
+      "loss": 0.2234,
+      "step": 10730
+    },
+    {
+      "epoch": 0.7793338654669473,
+      "grad_norm": 3.03125,
+      "learning_rate": 5.851759314487573e-06,
+      "loss": 0.1759,
+      "step": 10740
+    },
+    {
+      "epoch": 0.7800595022131921,
+      "grad_norm": 2.46875,
+      "learning_rate": 5.814900859217326e-06,
+      "loss": 0.181,
+      "step": 10750
+    },
+    {
+      "epoch": 0.7807851389594369,
+      "grad_norm": 4.96875,
+      "learning_rate": 5.7781435683951725e-06,
+      "loss": 0.1762,
+      "step": 10760
+    },
+    {
+      "epoch": 0.7815107757056817,
+      "grad_norm": 3.078125,
+      "learning_rate": 5.741487635845027e-06,
+      "loss": 0.2239,
+      "step": 10770
+    },
+    {
+      "epoch": 0.7822364124519265,
+      "grad_norm": 0.984375,
+      "learning_rate": 5.704933254856334e-06,
+      "loss": 0.1361,
+      "step": 10780
+    },
+    {
+      "epoch": 0.7829620491981714,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.668480618183053e-06,
+      "loss": 0.1579,
+      "step": 10790
+    },
+    {
+      "epoch": 0.7836876859444162,
+      "grad_norm": 5.375,
+      "learning_rate": 5.632129918042628e-06,
+      "loss": 0.2017,
+      "step": 10800
+    },
+    {
+      "epoch": 0.7844133226906611,
+      "grad_norm": 8.1875,
+      "learning_rate": 5.595881346115006e-06,
+      "loss": 0.1875,
+      "step": 10810
+    },
+    {
+      "epoch": 0.7851389594369059,
+      "grad_norm": 4.15625,
+      "learning_rate": 5.559735093541582e-06,
+      "loss": 0.1596,
+      "step": 10820
+    },
+    {
+      "epoch": 0.7858645961831507,
+      "grad_norm": 3.59375,
+      "learning_rate": 5.5236913509242175e-06,
+      "loss": 0.2001,
+      "step": 10830
+    },
+    {
+      "epoch": 0.7865902329293956,
+      "grad_norm": 10.125,
+      "learning_rate": 5.487750308324247e-06,
+      "loss": 0.2378,
+      "step": 10840
+    },
+    {
+      "epoch": 0.7873158696756404,
+      "grad_norm": 6.875,
+      "learning_rate": 5.451912155261443e-06,
+      "loss": 0.1904,
+      "step": 10850
+    },
+    {
+      "epoch": 0.7880415064218852,
+      "grad_norm": 2.40625,
+      "learning_rate": 5.416177080713042e-06,
+      "loss": 0.1203,
+      "step": 10860
+    },
+    {
+      "epoch": 0.78876714316813,
+      "grad_norm": 7.25,
+      "learning_rate": 5.380545273112733e-06,
+      "loss": 0.2244,
+      "step": 10870
+    },
+    {
+      "epoch": 0.7894927799143748,
+      "grad_norm": 3.859375,
+      "learning_rate": 5.345016920349671e-06,
+      "loss": 0.2093,
+      "step": 10880
+    },
+    {
+      "epoch": 0.7902184166606196,
+      "grad_norm": 7.40625,
+      "learning_rate": 5.309592209767506e-06,
+      "loss": 0.229,
+      "step": 10890
+    },
+    {
+      "epoch": 0.7909440534068646,
+      "grad_norm": 5.65625,
+      "learning_rate": 5.274271328163349e-06,
+      "loss": 0.1831,
+      "step": 10900
+    },
+    {
+      "epoch": 0.7916696901531094,
+      "grad_norm": 7.625,
+      "learning_rate": 5.239054461786827e-06,
+      "loss": 0.1408,
+      "step": 10910
+    },
+    {
+      "epoch": 0.7923953268993542,
+      "grad_norm": 1.9453125,
+      "learning_rate": 5.2039417963390764e-06,
+      "loss": 0.121,
+      "step": 10920
+    },
+    {
+      "epoch": 0.793120963645599,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.168933516971791e-06,
+      "loss": 0.1046,
+      "step": 10930
+    },
+    {
+      "epoch": 0.7938466003918438,
+      "grad_norm": 13.1875,
+      "learning_rate": 5.134029808286217e-06,
+      "loss": 0.1921,
+      "step": 10940
+    },
+    {
+      "epoch": 0.7945722371380887,
+      "grad_norm": 0.5859375,
+      "learning_rate": 5.099230854332196e-06,
+      "loss": 0.1492,
+      "step": 10950
+    },
+    {
+      "epoch": 0.7952978738843335,
+      "grad_norm": 8.25,
+      "learning_rate": 5.0645368386071816e-06,
+      "loss": 0.1717,
+      "step": 10960
+    },
+    {
+      "epoch": 0.7960235106305783,
+      "grad_norm": 6.625,
+      "learning_rate": 5.029947944055285e-06,
+      "loss": 0.2406,
+      "step": 10970
+    },
+    {
+      "epoch": 0.7967491473768231,
+      "grad_norm": 2.515625,
+      "learning_rate": 4.995464353066323e-06,
+      "loss": 0.2153,
+      "step": 10980
+    },
+    {
+      "epoch": 0.7974747841230679,
+      "grad_norm": 5.78125,
+      "learning_rate": 4.961086247474814e-06,
+      "loss": 0.1761,
+      "step": 10990
+    },
+    {
+      "epoch": 0.7982004208693129,
+      "grad_norm": 7.5625,
+      "learning_rate": 4.9268138085590495e-06,
+      "loss": 0.1547,
+      "step": 11000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 13781,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 5,
+  "trial_name": null,
+  "trial_params": null
+}