diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4383 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5735918320523116,
+  "eval_steps": 10,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001147183664104623,
+      "eval_loss": 1.873344898223877,
+      "eval_runtime": 12.7647,
+      "eval_samples_per_second": 515.797,
+      "eval_steps_per_second": 8.069,
+      "step": 10
+    },
+    {
+      "epoch": 0.002294367328209246,
+      "eval_loss": 1.8726389408111572,
+      "eval_runtime": 12.8667,
+      "eval_samples_per_second": 511.709,
+      "eval_steps_per_second": 8.005,
+      "step": 20
+    },
+    {
+      "epoch": 0.0034415509923138693,
+      "eval_loss": 1.8714078664779663,
+      "eval_runtime": 12.9103,
+      "eval_samples_per_second": 509.979,
+      "eval_steps_per_second": 7.978,
+      "step": 30
+    },
+    {
+      "epoch": 0.004588734656418492,
+      "eval_loss": 1.8696790933609009,
+      "eval_runtime": 12.947,
+      "eval_samples_per_second": 508.534,
+      "eval_steps_per_second": 7.955,
+      "step": 40
+    },
+    {
+      "epoch": 0.0057359183205231154,
+      "eval_loss": 1.8675329685211182,
+      "eval_runtime": 12.9458,
+      "eval_samples_per_second": 508.582,
+      "eval_steps_per_second": 7.956,
+      "step": 50
+    },
+    {
+      "epoch": 0.006883101984627739,
+      "eval_loss": 1.8649154901504517,
+      "eval_runtime": 13.0432,
+      "eval_samples_per_second": 504.785,
+      "eval_steps_per_second": 7.897,
+      "step": 60
+    },
+    {
+      "epoch": 0.008030285648732363,
+      "eval_loss": 1.8619294166564941,
+      "eval_runtime": 13.0638,
+      "eval_samples_per_second": 503.988,
+      "eval_steps_per_second": 7.884,
+      "step": 70
+    },
+    {
+      "epoch": 0.009177469312836984,
+      "eval_loss": 1.8583979606628418,
+      "eval_runtime": 13.0482,
+      "eval_samples_per_second": 504.592,
+      "eval_steps_per_second": 7.894,
+      "step": 80
+    },
+    {
+      "epoch": 0.010324652976941608,
+      "eval_loss": 1.85438871383667,
+      "eval_runtime": 13.0615,
+      "eval_samples_per_second": 504.075,
+      "eval_steps_per_second": 7.886,
+      "step": 90
+    },
+    {
+      "epoch": 0.011471836641046231,
+      "grad_norm": 9.938580513000488,
+      "learning_rate": 3.8226299694189603e-07,
+      "loss": 3.1046,
+      "step": 100
+    },
+    {
+      "epoch": 0.011471836641046231,
+      "eval_loss": 1.849947214126587,
+      "eval_runtime": 13.0663,
+      "eval_samples_per_second": 503.89,
+      "eval_steps_per_second": 7.883,
+      "step": 100
+    },
+    {
+      "epoch": 0.012619020305150854,
+      "eval_loss": 1.8451412916183472,
+      "eval_runtime": 12.9771,
+      "eval_samples_per_second": 507.357,
+      "eval_steps_per_second": 7.937,
+      "step": 110
+    },
+    {
+      "epoch": 0.013766203969255477,
+      "eval_loss": 1.8399487733840942,
+      "eval_runtime": 13.0209,
+      "eval_samples_per_second": 505.648,
+      "eval_steps_per_second": 7.91,
+      "step": 120
+    },
+    {
+      "epoch": 0.0149133876333601,
+      "eval_loss": 1.8342881202697754,
+      "eval_runtime": 13.0369,
+      "eval_samples_per_second": 505.028,
+      "eval_steps_per_second": 7.901,
+      "step": 130
+    },
+    {
+      "epoch": 0.016060571297464726,
+      "eval_loss": 1.8283486366271973,
+      "eval_runtime": 13.0149,
+      "eval_samples_per_second": 505.88,
+      "eval_steps_per_second": 7.914,
+      "step": 140
+    },
+    {
+      "epoch": 0.017207754961569347,
+      "eval_loss": 1.822334885597229,
+      "eval_runtime": 13.0213,
+      "eval_samples_per_second": 505.632,
+      "eval_steps_per_second": 7.91,
+      "step": 150
+    },
+    {
+      "epoch": 0.01835493862567397,
+      "eval_loss": 1.8158738613128662,
+      "eval_runtime": 13.0599,
+      "eval_samples_per_second": 504.14,
+      "eval_steps_per_second": 7.887,
+      "step": 160
+    },
+    {
+      "epoch": 0.019502122289778594,
+      "eval_loss": 1.8090614080429077,
+      "eval_runtime": 13.034,
+      "eval_samples_per_second": 505.14,
+      "eval_steps_per_second": 7.902,
+      "step": 170
+    },
+    {
+      "epoch": 0.020649305953883215,
+      "eval_loss": 1.8015782833099365,
+      "eval_runtime": 13.0665,
+      "eval_samples_per_second": 503.885,
+      "eval_steps_per_second": 7.883,
+      "step": 180
+    },
+    {
+      "epoch": 0.02179648961798784,
+      "eval_loss": 1.793796420097351,
+      "eval_runtime": 13.0555,
+      "eval_samples_per_second": 504.31,
+      "eval_steps_per_second": 7.889,
+      "step": 190
+    },
+    {
+      "epoch": 0.022943673282092462,
+      "grad_norm": 4.906337738037109,
+      "learning_rate": 7.645259938837921e-07,
+      "loss": 3.0303,
+      "step": 200
+    },
+    {
+      "epoch": 0.022943673282092462,
+      "eval_loss": 1.785815715789795,
+      "eval_runtime": 12.9925,
+      "eval_samples_per_second": 506.754,
+      "eval_steps_per_second": 7.928,
+      "step": 200
+    },
+    {
+      "epoch": 0.024090856946197087,
+      "eval_loss": 1.7775053977966309,
+      "eval_runtime": 13.0639,
+      "eval_samples_per_second": 503.986,
+      "eval_steps_per_second": 7.884,
+      "step": 210
+    },
+    {
+      "epoch": 0.025238040610301708,
+      "eval_loss": 1.7692992687225342,
+      "eval_runtime": 13.0129,
+      "eval_samples_per_second": 505.96,
+      "eval_steps_per_second": 7.915,
+      "step": 220
+    },
+    {
+      "epoch": 0.026385224274406333,
+      "eval_loss": 1.760453224182129,
+      "eval_runtime": 13.0078,
+      "eval_samples_per_second": 506.158,
+      "eval_steps_per_second": 7.918,
+      "step": 230
+    },
+    {
+      "epoch": 0.027532407938510955,
+      "eval_loss": 1.751396656036377,
+      "eval_runtime": 12.9957,
+      "eval_samples_per_second": 506.628,
+      "eval_steps_per_second": 7.926,
+      "step": 240
+    },
+    {
+      "epoch": 0.02867959160261558,
+      "eval_loss": 1.7417218685150146,
+      "eval_runtime": 12.9774,
+      "eval_samples_per_second": 507.344,
+      "eval_steps_per_second": 7.937,
+      "step": 250
+    },
+    {
+      "epoch": 0.0298267752667202,
+      "eval_loss": 1.7319914102554321,
+      "eval_runtime": 13.0219,
+      "eval_samples_per_second": 505.611,
+      "eval_steps_per_second": 7.91,
+      "step": 260
+    },
+    {
+      "epoch": 0.030973958930824826,
+      "eval_loss": 1.7227253913879395,
+      "eval_runtime": 13.0026,
+      "eval_samples_per_second": 506.361,
+      "eval_steps_per_second": 7.922,
+      "step": 270
+    },
+    {
+      "epoch": 0.03212114259492945,
+      "eval_loss": 1.7133797407150269,
+      "eval_runtime": 12.9757,
+      "eval_samples_per_second": 507.409,
+      "eval_steps_per_second": 7.938,
+      "step": 280
+    },
+    {
+      "epoch": 0.03326832625903407,
+      "eval_loss": 1.704041600227356,
+      "eval_runtime": 12.9845,
+      "eval_samples_per_second": 507.065,
+      "eval_steps_per_second": 7.933,
+      "step": 290
+    },
+    {
+      "epoch": 0.034415509923138694,
+      "grad_norm": 4.665822505950928,
+      "learning_rate": 1.1467889908256882e-06,
+      "loss": 2.9459,
+      "step": 300
+    },
+    {
+      "epoch": 0.034415509923138694,
+      "eval_loss": 1.6940686702728271,
+      "eval_runtime": 13.0019,
+      "eval_samples_per_second": 506.387,
+      "eval_steps_per_second": 7.922,
+      "step": 300
+    },
+    {
+      "epoch": 0.035562693587243316,
+      "eval_loss": 1.683342695236206,
+      "eval_runtime": 13.0065,
+      "eval_samples_per_second": 506.209,
+      "eval_steps_per_second": 7.919,
+      "step": 310
+    },
+    {
+      "epoch": 0.03670987725134794,
+      "eval_loss": 1.6724653244018555,
+      "eval_runtime": 13.0129,
+      "eval_samples_per_second": 505.96,
+      "eval_steps_per_second": 7.915,
+      "step": 320
+    },
+    {
+      "epoch": 0.037857060915452566,
+      "eval_loss": 1.6614341735839844,
+      "eval_runtime": 12.9921,
+      "eval_samples_per_second": 506.769,
+      "eval_steps_per_second": 7.928,
+      "step": 330
+    },
+    {
+      "epoch": 0.03900424457955719,
+      "eval_loss": 1.6510112285614014,
+      "eval_runtime": 13.0242,
+      "eval_samples_per_second": 505.52,
+      "eval_steps_per_second": 7.908,
+      "step": 340
+    },
+    {
+      "epoch": 0.04015142824366181,
+      "eval_loss": 1.6401513814926147,
+      "eval_runtime": 12.9214,
+      "eval_samples_per_second": 509.542,
+      "eval_steps_per_second": 7.971,
+      "step": 350
+    },
+    {
+      "epoch": 0.04129861190776643,
+      "eval_loss": 1.6295816898345947,
+      "eval_runtime": 12.9563,
+      "eval_samples_per_second": 508.171,
+      "eval_steps_per_second": 7.95,
+      "step": 360
+    },
+    {
+      "epoch": 0.04244579557187106,
+      "eval_loss": 1.6187150478363037,
+      "eval_runtime": 12.9758,
+      "eval_samples_per_second": 507.405,
+      "eval_steps_per_second": 7.938,
+      "step": 370
+    },
+    {
+      "epoch": 0.04359297923597568,
+      "eval_loss": 1.607272982597351,
+      "eval_runtime": 12.9876,
+      "eval_samples_per_second": 506.947,
+      "eval_steps_per_second": 7.931,
+      "step": 380
+    },
+    {
+      "epoch": 0.0447401629000803,
+      "eval_loss": 1.5961676836013794,
+      "eval_runtime": 12.9782,
+      "eval_samples_per_second": 507.313,
+      "eval_steps_per_second": 7.936,
+      "step": 390
+    },
+    {
+      "epoch": 0.045887346564184923,
+      "grad_norm": 4.870114326477051,
+      "learning_rate": 1.5290519877675841e-06,
+      "loss": 2.7813,
+      "step": 400
+    },
+    {
+      "epoch": 0.045887346564184923,
+      "eval_loss": 1.5848218202590942,
+      "eval_runtime": 12.9783,
+      "eval_samples_per_second": 507.309,
+      "eval_steps_per_second": 7.936,
+      "step": 400
+    },
+    {
+      "epoch": 0.04703453022828955,
+      "eval_loss": 1.5734797716140747,
+      "eval_runtime": 12.9739,
+      "eval_samples_per_second": 507.482,
+      "eval_steps_per_second": 7.939,
+      "step": 410
+    },
+    {
+      "epoch": 0.04818171389239417,
+      "eval_loss": 1.562021255493164,
+      "eval_runtime": 12.9388,
+      "eval_samples_per_second": 508.855,
+      "eval_steps_per_second": 7.961,
+      "step": 420
+    },
+    {
+      "epoch": 0.049328897556498795,
+      "eval_loss": 1.5495364665985107,
+      "eval_runtime": 12.9412,
+      "eval_samples_per_second": 508.764,
+      "eval_steps_per_second": 7.959,
+      "step": 430
+    },
+    {
+      "epoch": 0.050476081220603417,
+      "eval_loss": 1.5375314950942993,
+      "eval_runtime": 12.9686,
+      "eval_samples_per_second": 507.687,
+      "eval_steps_per_second": 7.942,
+      "step": 440
+    },
+    {
+      "epoch": 0.051623264884708045,
+      "eval_loss": 1.525598168373108,
+      "eval_runtime": 12.9695,
+      "eval_samples_per_second": 507.651,
+      "eval_steps_per_second": 7.942,
+      "step": 450
+    },
+    {
+      "epoch": 0.052770448548812667,
+      "eval_loss": 1.5132672786712646,
+      "eval_runtime": 12.8961,
+      "eval_samples_per_second": 510.543,
+      "eval_steps_per_second": 7.987,
+      "step": 460
+    },
+    {
+      "epoch": 0.05391763221291729,
+      "eval_loss": 1.5012215375900269,
+      "eval_runtime": 12.9428,
+      "eval_samples_per_second": 508.7,
+      "eval_steps_per_second": 7.958,
+      "step": 470
+    },
+    {
+      "epoch": 0.05506481587702191,
+      "eval_loss": 1.4892219305038452,
+      "eval_runtime": 12.9208,
+      "eval_samples_per_second": 509.567,
+      "eval_steps_per_second": 7.972,
+      "step": 480
+    },
+    {
+      "epoch": 0.05621199954112653,
+      "eval_loss": 1.4768636226654053,
+      "eval_runtime": 12.9423,
+      "eval_samples_per_second": 508.721,
+      "eval_steps_per_second": 7.958,
+      "step": 490
+    },
+    {
+      "epoch": 0.05735918320523116,
+      "grad_norm": 4.155641555786133,
+      "learning_rate": 1.9113149847094803e-06,
+      "loss": 2.6308,
+      "step": 500
+    },
+    {
+      "epoch": 0.05735918320523116,
+      "eval_loss": 1.4640088081359863,
+      "eval_runtime": 12.8729,
+      "eval_samples_per_second": 511.462,
+      "eval_steps_per_second": 8.001,
+      "step": 500
+    },
+    {
+      "epoch": 0.05850636686933578,
+      "eval_loss": 1.4513096809387207,
+      "eval_runtime": 12.9653,
+      "eval_samples_per_second": 507.817,
+      "eval_steps_per_second": 7.944,
+      "step": 510
+    },
+    {
+      "epoch": 0.0596535505334404,
+      "eval_loss": 1.439149260520935,
+      "eval_runtime": 12.9443,
+      "eval_samples_per_second": 508.639,
+      "eval_steps_per_second": 7.957,
+      "step": 520
+    },
+    {
+      "epoch": 0.060800734197545024,
+      "eval_loss": 1.426237940788269,
+      "eval_runtime": 12.9496,
+      "eval_samples_per_second": 508.433,
+      "eval_steps_per_second": 7.954,
+      "step": 530
+    },
+    {
+      "epoch": 0.06194791786164965,
+      "eval_loss": 1.4129557609558105,
+      "eval_runtime": 12.9822,
+      "eval_samples_per_second": 507.157,
+      "eval_steps_per_second": 7.934,
+      "step": 540
+    },
+    {
+      "epoch": 0.06309510152575427,
+      "eval_loss": 1.3997886180877686,
+      "eval_runtime": 12.9979,
+      "eval_samples_per_second": 506.542,
+      "eval_steps_per_second": 7.924,
+      "step": 550
+    },
+    {
+      "epoch": 0.0642422851898589,
+      "eval_loss": 1.3873906135559082,
+      "eval_runtime": 12.9378,
+      "eval_samples_per_second": 508.895,
+      "eval_steps_per_second": 7.961,
+      "step": 560
+    },
+    {
+      "epoch": 0.06538946885396352,
+      "eval_loss": 1.3751789331436157,
+      "eval_runtime": 12.9624,
+      "eval_samples_per_second": 507.932,
+      "eval_steps_per_second": 7.946,
+      "step": 570
+    },
+    {
+      "epoch": 0.06653665251806815,
+      "eval_loss": 1.3620370626449585,
+      "eval_runtime": 12.9498,
+      "eval_samples_per_second": 508.426,
+      "eval_steps_per_second": 7.954,
+      "step": 580
+    },
+    {
+      "epoch": 0.06768383618217276,
+      "eval_loss": 1.3485124111175537,
+      "eval_runtime": 12.9759,
+      "eval_samples_per_second": 507.403,
+      "eval_steps_per_second": 7.938,
+      "step": 590
+    },
+    {
+      "epoch": 0.06883101984627739,
+      "grad_norm": 5.262124061584473,
+      "learning_rate": 2.2935779816513764e-06,
+      "loss": 2.4452,
+      "step": 600
+    },
+    {
+      "epoch": 0.06883101984627739,
+      "eval_loss": 1.3349775075912476,
+      "eval_runtime": 12.9631,
+      "eval_samples_per_second": 507.902,
+      "eval_steps_per_second": 7.946,
+      "step": 600
+    },
+    {
+      "epoch": 0.06997820351038202,
+      "eval_loss": 1.3213400840759277,
+      "eval_runtime": 12.9619,
+      "eval_samples_per_second": 507.951,
+      "eval_steps_per_second": 7.946,
+      "step": 610
+    },
+    {
+      "epoch": 0.07112538717448663,
+      "eval_loss": 1.308822512626648,
+      "eval_runtime": 12.9652,
+      "eval_samples_per_second": 507.82,
+      "eval_steps_per_second": 7.944,
+      "step": 620
+    },
+    {
+      "epoch": 0.07227257083859126,
+      "eval_loss": 1.296485185623169,
+      "eval_runtime": 13.0441,
+      "eval_samples_per_second": 504.75,
+      "eval_steps_per_second": 7.896,
+      "step": 630
+    },
+    {
+      "epoch": 0.07341975450269587,
+      "eval_loss": 1.283867597579956,
+      "eval_runtime": 12.9822,
+      "eval_samples_per_second": 507.154,
+      "eval_steps_per_second": 7.934,
+      "step": 640
+    },
+    {
+      "epoch": 0.0745669381668005,
+      "eval_loss": 1.2713148593902588,
+      "eval_runtime": 12.9775,
+      "eval_samples_per_second": 507.338,
+      "eval_steps_per_second": 7.937,
+      "step": 650
+    },
+    {
+      "epoch": 0.07571412183090513,
+      "eval_loss": 1.2591922283172607,
+      "eval_runtime": 13.0227,
+      "eval_samples_per_second": 505.578,
+      "eval_steps_per_second": 7.909,
+      "step": 660
+    },
+    {
+      "epoch": 0.07686130549500975,
+      "eval_loss": 1.246610164642334,
+      "eval_runtime": 13.014,
+      "eval_samples_per_second": 505.917,
+      "eval_steps_per_second": 7.915,
+      "step": 670
+    },
+    {
+      "epoch": 0.07800848915911437,
+      "eval_loss": 1.2331972122192383,
+      "eval_runtime": 12.9634,
+      "eval_samples_per_second": 507.891,
+      "eval_steps_per_second": 7.945,
+      "step": 680
+    },
+    {
+      "epoch": 0.079155672823219,
+      "eval_loss": 1.2203081846237183,
+      "eval_runtime": 12.9664,
+      "eval_samples_per_second": 507.775,
+      "eval_steps_per_second": 7.944,
+      "step": 690
+    },
+    {
+      "epoch": 0.08030285648732362,
+      "grad_norm": 3.824066400527954,
+      "learning_rate": 2.6758409785932725e-06,
+      "loss": 2.2626,
+      "step": 700
+    },
+    {
+      "epoch": 0.08030285648732362,
+      "eval_loss": 1.207729697227478,
+      "eval_runtime": 13.037,
+      "eval_samples_per_second": 505.025,
+      "eval_steps_per_second": 7.901,
+      "step": 700
+    },
+    {
+      "epoch": 0.08145004015142825,
+      "eval_loss": 1.195885419845581,
+      "eval_runtime": 12.9625,
+      "eval_samples_per_second": 507.928,
+      "eval_steps_per_second": 7.946,
+      "step": 710
+    },
+    {
+      "epoch": 0.08259722381553286,
+      "eval_loss": 1.1840639114379883,
+      "eval_runtime": 13.0124,
+      "eval_samples_per_second": 505.98,
+      "eval_steps_per_second": 7.916,
+      "step": 720
+    },
+    {
+      "epoch": 0.08374440747963749,
+      "eval_loss": 1.1725258827209473,
+      "eval_runtime": 12.9851,
+      "eval_samples_per_second": 507.043,
+      "eval_steps_per_second": 7.932,
+      "step": 730
+    },
+    {
+      "epoch": 0.08489159114374212,
+      "eval_loss": 1.1619255542755127,
+      "eval_runtime": 13.1041,
+      "eval_samples_per_second": 502.438,
+      "eval_steps_per_second": 7.86,
+      "step": 740
+    },
+    {
+      "epoch": 0.08603877480784673,
+      "eval_loss": 1.1515777111053467,
+      "eval_runtime": 12.9648,
+      "eval_samples_per_second": 507.837,
+      "eval_steps_per_second": 7.945,
+      "step": 750
+    },
+    {
+      "epoch": 0.08718595847195136,
+      "eval_loss": 1.1416434049606323,
+      "eval_runtime": 13.0349,
+      "eval_samples_per_second": 505.104,
+      "eval_steps_per_second": 7.902,
+      "step": 760
+    },
+    {
+      "epoch": 0.08833314213605599,
+      "eval_loss": 1.1320044994354248,
+      "eval_runtime": 13.0667,
+      "eval_samples_per_second": 503.875,
+      "eval_steps_per_second": 7.883,
+      "step": 770
+    },
+    {
+      "epoch": 0.0894803258001606,
+      "eval_loss": 1.1226693391799927,
+      "eval_runtime": 13.3034,
+      "eval_samples_per_second": 494.912,
+      "eval_steps_per_second": 7.742,
+      "step": 780
+    },
+    {
+      "epoch": 0.09062750946426523,
+      "eval_loss": 1.113792061805725,
+      "eval_runtime": 12.9793,
+      "eval_samples_per_second": 507.271,
+      "eval_steps_per_second": 7.936,
+      "step": 790
+    },
+    {
+      "epoch": 0.09177469312836985,
+      "grad_norm": 3.358140707015991,
+      "learning_rate": 3.0581039755351682e-06,
+      "loss": 2.0044,
+      "step": 800
+    },
+    {
+      "epoch": 0.09177469312836985,
+      "eval_loss": 1.1052675247192383,
+      "eval_runtime": 13.0193,
+      "eval_samples_per_second": 505.709,
+      "eval_steps_per_second": 7.911,
+      "step": 800
+    },
+    {
+      "epoch": 0.09292187679247448,
+      "eval_loss": 1.0964922904968262,
+      "eval_runtime": 13.1289,
+      "eval_samples_per_second": 501.489,
+      "eval_steps_per_second": 7.845,
+      "step": 810
+    },
+    {
+      "epoch": 0.0940690604565791,
+      "eval_loss": 1.0879426002502441,
+      "eval_runtime": 12.9727,
+      "eval_samples_per_second": 507.526,
+      "eval_steps_per_second": 7.94,
+      "step": 820
+    },
+    {
+      "epoch": 0.09521624412068372,
+      "eval_loss": 1.0795575380325317,
+      "eval_runtime": 13.1138,
+      "eval_samples_per_second": 502.068,
+      "eval_steps_per_second": 7.854,
+      "step": 830
+    },
+    {
+      "epoch": 0.09636342778478835,
+      "eval_loss": 1.0718028545379639,
+      "eval_runtime": 12.9813,
+      "eval_samples_per_second": 507.191,
+      "eval_steps_per_second": 7.934,
+      "step": 840
+    },
+    {
+      "epoch": 0.09751061144889296,
+      "eval_loss": 1.0643888711929321,
+      "eval_runtime": 13.0171,
+      "eval_samples_per_second": 505.795,
+      "eval_steps_per_second": 7.913,
+      "step": 850
+    },
+    {
+      "epoch": 0.09865779511299759,
+      "eval_loss": 1.0564391613006592,
+      "eval_runtime": 13.0169,
+      "eval_samples_per_second": 505.806,
+      "eval_steps_per_second": 7.913,
+      "step": 860
+    },
+    {
+      "epoch": 0.09980497877710222,
+      "eval_loss": 1.0490267276763916,
+      "eval_runtime": 13.0703,
+      "eval_samples_per_second": 503.738,
+      "eval_steps_per_second": 7.88,
+      "step": 870
+    },
+    {
+      "epoch": 0.10095216244120683,
+      "eval_loss": 1.0417358875274658,
+      "eval_runtime": 13.0303,
+      "eval_samples_per_second": 505.284,
+      "eval_steps_per_second": 7.905,
+      "step": 880
+    },
+    {
+      "epoch": 0.10209934610531146,
+      "eval_loss": 1.0353840589523315,
+      "eval_runtime": 12.9765,
+      "eval_samples_per_second": 507.381,
+      "eval_steps_per_second": 7.937,
+      "step": 890
+    },
+    {
+      "epoch": 0.10324652976941609,
+      "grad_norm": 3.705247402191162,
+      "learning_rate": 3.4403669724770644e-06,
+      "loss": 1.8763,
+      "step": 900
+    },
+    {
+      "epoch": 0.10324652976941609,
+      "eval_loss": 1.0295778512954712,
+      "eval_runtime": 13.0396,
+      "eval_samples_per_second": 504.924,
+      "eval_steps_per_second": 7.899,
+      "step": 900
+    },
+    {
+      "epoch": 0.1043937134335207,
+      "eval_loss": 1.0238651037216187,
+      "eval_runtime": 12.973,
+      "eval_samples_per_second": 507.515,
+      "eval_steps_per_second": 7.94,
+      "step": 910
+    },
+    {
+      "epoch": 0.10554089709762533,
+      "eval_loss": 1.0180109739303589,
+      "eval_runtime": 13.0196,
+      "eval_samples_per_second": 505.7,
+      "eval_steps_per_second": 7.911,
+      "step": 920
+    },
+    {
+      "epoch": 0.10668808076172995,
+      "eval_loss": 1.0122654438018799,
+      "eval_runtime": 12.9877,
+      "eval_samples_per_second": 506.942,
+      "eval_steps_per_second": 7.931,
+      "step": 930
+    },
+    {
+      "epoch": 0.10783526442583458,
+      "eval_loss": 1.0065423250198364,
+      "eval_runtime": 13.0039,
+      "eval_samples_per_second": 506.312,
+      "eval_steps_per_second": 7.921,
+      "step": 940
+    },
+    {
+      "epoch": 0.1089824480899392,
+      "eval_loss": 1.0008338689804077,
+      "eval_runtime": 13.1129,
+      "eval_samples_per_second": 502.101,
+      "eval_steps_per_second": 7.855,
+      "step": 950
+    },
+    {
+      "epoch": 0.11012963175404382,
+      "eval_loss": 0.9950375556945801,
+      "eval_runtime": 13.0191,
+      "eval_samples_per_second": 505.718,
+      "eval_steps_per_second": 7.911,
+      "step": 960
+    },
+    {
+      "epoch": 0.11127681541814845,
+      "eval_loss": 0.9893631935119629,
+      "eval_runtime": 12.9794,
+      "eval_samples_per_second": 507.267,
+      "eval_steps_per_second": 7.936,
+      "step": 970
+    },
+    {
+      "epoch": 0.11242399908225306,
+      "eval_loss": 0.9840025305747986,
+      "eval_runtime": 12.9799,
+      "eval_samples_per_second": 507.247,
+      "eval_steps_per_second": 7.935,
+      "step": 980
+    },
+    {
+      "epoch": 0.11357118274635769,
+      "eval_loss": 0.9792994856834412,
+      "eval_runtime": 12.9866,
+      "eval_samples_per_second": 506.983,
+      "eval_steps_per_second": 7.931,
+      "step": 990
+    },
+    {
+      "epoch": 0.11471836641046232,
+      "grad_norm": 4.240755558013916,
+      "learning_rate": 3.8226299694189605e-06,
+      "loss": 1.7287,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11471836641046232,
+      "eval_loss": 0.9751574993133545,
+      "eval_runtime": 13.0083,
+      "eval_samples_per_second": 506.136,
+      "eval_steps_per_second": 7.918,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11586555007456693,
+      "eval_loss": 0.9705988764762878,
+      "eval_runtime": 12.9835,
+      "eval_samples_per_second": 507.107,
+      "eval_steps_per_second": 7.933,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11701273373867156,
+      "eval_loss": 0.9658572673797607,
+      "eval_runtime": 13.0627,
+      "eval_samples_per_second": 504.031,
+      "eval_steps_per_second": 7.885,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11815991740277619,
+      "eval_loss": 0.9614543318748474,
+      "eval_runtime": 13.0169,
+      "eval_samples_per_second": 505.803,
+      "eval_steps_per_second": 7.913,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1193071010668808,
+      "eval_loss": 0.9571945667266846,
+      "eval_runtime": 12.9762,
+      "eval_samples_per_second": 507.389,
+      "eval_steps_per_second": 7.938,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12045428473098543,
+      "eval_loss": 0.9531480073928833,
+      "eval_runtime": 13.0036,
+      "eval_samples_per_second": 506.323,
+      "eval_steps_per_second": 7.921,
+      "step": 1050
+    },
+    {
+      "epoch": 0.12160146839509005,
+      "eval_loss": 0.9493557810783386,
+      "eval_runtime": 13.0046,
+      "eval_samples_per_second": 506.282,
+      "eval_steps_per_second": 7.92,
+      "step": 1060
+    },
+    {
+      "epoch": 0.12274865205919468,
+      "eval_loss": 0.9455849528312683,
+      "eval_runtime": 13.0676,
+      "eval_samples_per_second": 503.843,
+      "eval_steps_per_second": 7.882,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1238958357232993,
+      "eval_loss": 0.9415374398231506,
+      "eval_runtime": 13.0698,
+      "eval_samples_per_second": 503.755,
+      "eval_steps_per_second": 7.881,
+      "step": 1080
+    },
+    {
+      "epoch": 0.12504301938740392,
+      "eval_loss": 0.9376588463783264,
+      "eval_runtime": 12.9843,
+      "eval_samples_per_second": 507.072,
+      "eval_steps_per_second": 7.933,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12619020305150855,
+      "grad_norm": 3.3081679344177246,
+      "learning_rate": 4.204892966360857e-06,
+      "loss": 1.6312,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12619020305150855,
+      "eval_loss": 0.9338813424110413,
+      "eval_runtime": 13.0365,
+      "eval_samples_per_second": 505.043,
+      "eval_steps_per_second": 7.901,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12733738671561318,
+      "eval_loss": 0.9302825927734375,
+      "eval_runtime": 13.0692,
+      "eval_samples_per_second": 503.779,
+      "eval_steps_per_second": 7.881,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1284845703797178,
+      "eval_loss": 0.9267016053199768,
+      "eval_runtime": 13.0455,
+      "eval_samples_per_second": 504.694,
+      "eval_steps_per_second": 7.895,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1296317540438224,
+      "eval_loss": 0.9232119917869568,
+      "eval_runtime": 13.0051,
+      "eval_samples_per_second": 506.262,
+      "eval_steps_per_second": 7.92,
+      "step": 1130
+    },
+    {
+      "epoch": 0.13077893770792703,
+      "eval_loss": 0.9197220206260681,
+      "eval_runtime": 13.0069,
+      "eval_samples_per_second": 506.192,
+      "eval_steps_per_second": 7.919,
+      "step": 1140
+    },
+    {
+      "epoch": 0.13192612137203166,
+      "eval_loss": 0.9162309765815735,
+      "eval_runtime": 12.9851,
+      "eval_samples_per_second": 507.044,
+      "eval_steps_per_second": 7.932,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1330733050361363,
+      "eval_loss": 0.9127652049064636,
+      "eval_runtime": 12.9768,
+      "eval_samples_per_second": 507.366,
+      "eval_steps_per_second": 7.937,
+      "step": 1160
+    },
+    {
+      "epoch": 0.13422048870024092,
+      "eval_loss": 0.9096914529800415,
+      "eval_runtime": 12.9953,
+      "eval_samples_per_second": 506.645,
+      "eval_steps_per_second": 7.926,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13536767236434552,
+      "eval_loss": 0.9069137573242188,
+      "eval_runtime": 13.0109,
+      "eval_samples_per_second": 506.038,
+      "eval_steps_per_second": 7.916,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13651485602845015,
+      "eval_loss": 0.9039744734764099,
+      "eval_runtime": 12.9772,
+      "eval_samples_per_second": 507.352,
+      "eval_steps_per_second": 7.937,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13766203969255478,
+      "grad_norm": 3.420077085494995,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 1.5316,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13766203969255478,
+      "eval_loss": 0.9009912610054016,
+      "eval_runtime": 12.9848,
+      "eval_samples_per_second": 507.056,
+      "eval_steps_per_second": 7.932,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1388092233566594,
+      "eval_loss": 0.8979319930076599,
+      "eval_runtime": 13.0295,
+      "eval_samples_per_second": 505.313,
+      "eval_steps_per_second": 7.905,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13995640702076403,
+      "eval_loss": 0.8946565985679626,
+      "eval_runtime": 13.0119,
+      "eval_samples_per_second": 505.997,
+      "eval_steps_per_second": 7.916,
+      "step": 1220
+    },
+    {
+      "epoch": 0.14110359068486863,
+      "eval_loss": 0.8915460705757141,
+      "eval_runtime": 12.9979,
+      "eval_samples_per_second": 506.545,
+      "eval_steps_per_second": 7.924,
+      "step": 1230
+    },
+    {
+      "epoch": 0.14225077434897326,
+      "eval_loss": 0.8887993097305298,
+      "eval_runtime": 12.9633,
+      "eval_samples_per_second": 507.895,
+      "eval_steps_per_second": 7.945,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1433979580130779,
+      "eval_loss": 0.886054277420044,
+      "eval_runtime": 12.9996,
+      "eval_samples_per_second": 506.477,
+      "eval_steps_per_second": 7.923,
+      "step": 1250
+    },
+    {
+      "epoch": 0.14454514167718252,
+      "eval_loss": 0.8833284974098206,
+      "eval_runtime": 13.1133,
+      "eval_samples_per_second": 502.086,
+      "eval_steps_per_second": 7.855,
+      "step": 1260
+    },
+    {
+      "epoch": 0.14569232534128715,
+      "eval_loss": 0.8806452751159668,
+      "eval_runtime": 13.0208,
+      "eval_samples_per_second": 505.654,
+      "eval_steps_per_second": 7.91,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14683950900539175,
+      "eval_loss": 0.8778645992279053,
+      "eval_runtime": 13.1071,
+      "eval_samples_per_second": 502.322,
+      "eval_steps_per_second": 7.858,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14798669266949638,
+      "eval_loss": 0.8747658133506775,
+      "eval_runtime": 13.0362,
+      "eval_samples_per_second": 505.055,
+      "eval_steps_per_second": 7.901,
+      "step": 1290
+    },
+    {
+      "epoch": 0.149133876333601,
+      "grad_norm": 3.3882789611816406,
+      "learning_rate": 4.969418960244649e-06,
+      "loss": 1.4961,
+      "step": 1300
+    },
+    {
+      "epoch": 0.149133876333601,
+      "eval_loss": 0.8717615008354187,
+      "eval_runtime": 13.0104,
+      "eval_samples_per_second": 506.058,
+      "eval_steps_per_second": 7.917,
+      "step": 1300
+    },
+    {
+      "epoch": 0.15028105999770563,
+      "eval_loss": 0.8690453767776489,
+      "eval_runtime": 13.0189,
+      "eval_samples_per_second": 505.727,
+      "eval_steps_per_second": 7.912,
+      "step": 1310
+    },
+    {
+      "epoch": 0.15142824366181026,
+      "eval_loss": 0.8664180040359497,
+      "eval_runtime": 12.988,
+      "eval_samples_per_second": 506.931,
+      "eval_steps_per_second": 7.93,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1525754273259149,
+      "eval_loss": 0.8634527921676636,
+      "eval_runtime": 12.9897,
+      "eval_samples_per_second": 506.861,
+      "eval_steps_per_second": 7.929,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1537226109900195,
+      "eval_loss": 0.8603318929672241,
+      "eval_runtime": 13.0269,
+      "eval_samples_per_second": 505.417,
+      "eval_steps_per_second": 7.907,
+      "step": 1340
+    },
+    {
+      "epoch": 0.15486979465412412,
+      "eval_loss": 0.8573653697967529,
+      "eval_runtime": 12.9945,
+      "eval_samples_per_second": 506.677,
+      "eval_steps_per_second": 7.926,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15601697831822875,
+      "eval_loss": 0.8545491695404053,
+      "eval_runtime": 12.9981,
+      "eval_samples_per_second": 506.536,
+      "eval_steps_per_second": 7.924,
+      "step": 1360
+    },
+    {
+      "epoch": 0.15716416198233338,
+      "eval_loss": 0.85209721326828,
+      "eval_runtime": 13.0301,
+      "eval_samples_per_second": 505.293,
+      "eval_steps_per_second": 7.905,
+      "step": 1370
+    },
+    {
+      "epoch": 0.158311345646438,
+      "eval_loss": 0.8497452139854431,
+      "eval_runtime": 13.1229,
+      "eval_samples_per_second": 501.717,
+      "eval_steps_per_second": 7.849,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1594585293105426,
+      "eval_loss": 0.8474416136741638,
+      "eval_runtime": 13.0624,
+      "eval_samples_per_second": 504.043,
+      "eval_steps_per_second": 7.885,
+      "step": 1390
+    },
+    {
+      "epoch": 0.16060571297464724,
+      "grad_norm": 3.7449283599853516,
+      "learning_rate": 5.351681957186545e-06,
+      "loss": 1.451,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16060571297464724,
+      "eval_loss": 0.8452581763267517,
+      "eval_runtime": 13.0191,
+      "eval_samples_per_second": 505.72,
+      "eval_steps_per_second": 7.911,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16175289663875186,
+      "eval_loss": 0.8428558111190796,
+      "eval_runtime": 13.0038,
+      "eval_samples_per_second": 506.314,
+      "eval_steps_per_second": 7.921,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1629000803028565,
+      "eval_loss": 0.8404139876365662,
+      "eval_runtime": 12.9904,
+      "eval_samples_per_second": 506.837,
+      "eval_steps_per_second": 7.929,
+      "step": 1420
+    },
+    {
+      "epoch": 0.16404726396696112,
+      "eval_loss": 0.8380302786827087,
+      "eval_runtime": 12.9812,
+      "eval_samples_per_second": 507.195,
+      "eval_steps_per_second": 7.935,
+      "step": 1430
+    },
+    {
+      "epoch": 0.16519444763106572,
+      "eval_loss": 0.8357470631599426,
+      "eval_runtime": 13.0241,
+      "eval_samples_per_second": 505.526,
+      "eval_steps_per_second": 7.908,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16634163129517035,
+      "eval_loss": 0.8336036205291748,
+      "eval_runtime": 13.142,
+      "eval_samples_per_second": 500.99,
+      "eval_steps_per_second": 7.837,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16748881495927498,
+      "eval_loss": 0.8312162756919861,
+      "eval_runtime": 12.9935,
+      "eval_samples_per_second": 506.714,
+      "eval_steps_per_second": 7.927,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1686359986233796,
+      "eval_loss": 0.8288681507110596,
+      "eval_runtime": 13.1018,
+      "eval_samples_per_second": 502.527,
+      "eval_steps_per_second": 7.862,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16978318228748424,
+      "eval_loss": 0.8261700868606567,
+      "eval_runtime": 13.0327,
+      "eval_samples_per_second": 505.19,
+      "eval_steps_per_second": 7.903,
+      "step": 1480
+    },
+    {
+      "epoch": 0.17093036595158884,
+      "eval_loss": 0.8235771656036377,
+      "eval_runtime": 12.9895,
+      "eval_samples_per_second": 506.873,
+      "eval_steps_per_second": 7.93,
+      "step": 1490
+    },
+    {
+      "epoch": 0.17207754961569346,
+      "grad_norm": 2.5695531368255615,
+      "learning_rate": 5.733944954128441e-06,
+      "loss": 1.4177,
+      "step": 1500
+    },
+    {
+      "epoch": 0.17207754961569346,
+      "eval_loss": 0.8213275074958801,
+      "eval_runtime": 13.0468,
+      "eval_samples_per_second": 504.647,
+      "eval_steps_per_second": 7.895,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1732247332797981,
+      "eval_loss": 0.8189388513565063,
+      "eval_runtime": 12.9877,
+      "eval_samples_per_second": 506.943,
+      "eval_steps_per_second": 7.931,
+      "step": 1510
+    },
+    {
+      "epoch": 0.17437191694390272,
+      "eval_loss": 0.8168380856513977,
+      "eval_runtime": 12.9915,
+      "eval_samples_per_second": 506.793,
+      "eval_steps_per_second": 7.928,
+      "step": 1520
+    },
+    {
+      "epoch": 0.17551910060800735,
+      "eval_loss": 0.814733624458313,
+      "eval_runtime": 13.1726,
+      "eval_samples_per_second": 499.824,
+      "eval_steps_per_second": 7.819,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17666628427211198,
+      "eval_loss": 0.8127268552780151,
+      "eval_runtime": 13.0207,
+      "eval_samples_per_second": 505.657,
+      "eval_steps_per_second": 7.91,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17781346793621658,
+      "eval_loss": 0.8106749057769775,
+      "eval_runtime": 13.0973,
+      "eval_samples_per_second": 502.697,
+      "eval_steps_per_second": 7.864,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1789606516003212,
+      "eval_loss": 0.808172345161438,
+      "eval_runtime": 13.2116,
+      "eval_samples_per_second": 498.349,
+      "eval_steps_per_second": 7.796,
+      "step": 1560
+    },
+    {
+      "epoch": 0.18010783526442584,
+      "eval_loss": 0.8059112429618835,
+      "eval_runtime": 12.9871,
+      "eval_samples_per_second": 506.963,
+      "eval_steps_per_second": 7.931,
+      "step": 1570
+    },
+    {
+      "epoch": 0.18125501892853046,
+      "eval_loss": 0.8035888671875,
+      "eval_runtime": 13.0963,
+      "eval_samples_per_second": 502.737,
+      "eval_steps_per_second": 7.865,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1824022025926351,
+      "eval_loss": 0.8014644384384155,
+      "eval_runtime": 13.0028,
+      "eval_samples_per_second": 506.351,
+      "eval_steps_per_second": 7.921,
+      "step": 1590
+    },
+    {
+      "epoch": 0.1835493862567397,
+      "grad_norm": 2.681454658508301,
+      "learning_rate": 6.1162079510703365e-06,
+      "loss": 1.3734,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1835493862567397,
+      "eval_loss": 0.7993160486221313,
+      "eval_runtime": 12.9893,
+      "eval_samples_per_second": 506.879,
+      "eval_steps_per_second": 7.93,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18469656992084432,
+      "eval_loss": 0.7969831228256226,
+      "eval_runtime": 13.0137,
+      "eval_samples_per_second": 505.929,
+      "eval_steps_per_second": 7.915,
+      "step": 1610
+    },
+    {
+      "epoch": 0.18584375358494895,
+      "eval_loss": 0.7947937250137329,
+      "eval_runtime": 13.0467,
+      "eval_samples_per_second": 504.648,
+      "eval_steps_per_second": 7.895,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18699093724905358,
+      "eval_loss": 0.7922271490097046,
+      "eval_runtime": 13.1133,
+      "eval_samples_per_second": 502.086,
+      "eval_steps_per_second": 7.855,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1881381209131582,
+      "eval_loss": 0.790046751499176,
+      "eval_runtime": 13.0687,
+      "eval_samples_per_second": 503.801,
+      "eval_steps_per_second": 7.881,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1892853045772628,
+      "eval_loss": 0.7877430319786072,
+      "eval_runtime": 13.1712,
+      "eval_samples_per_second": 499.877,
+      "eval_steps_per_second": 7.82,
+      "step": 1650
+    },
+    {
+      "epoch": 0.19043248824136744,
+      "eval_loss": 0.7852274179458618,
+      "eval_runtime": 13.0721,
+      "eval_samples_per_second": 503.669,
+      "eval_steps_per_second": 7.879,
+      "step": 1660
+    },
+    {
+      "epoch": 0.19157967190547207,
+      "eval_loss": 0.782863438129425,
+      "eval_runtime": 12.9741,
+      "eval_samples_per_second": 507.471,
+      "eval_steps_per_second": 7.939,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1927268555695767,
+      "eval_loss": 0.7804363965988159,
+      "eval_runtime": 12.9826,
+      "eval_samples_per_second": 507.142,
+      "eval_steps_per_second": 7.934,
+      "step": 1680
+    },
+    {
+      "epoch": 0.19387403923368132,
+      "eval_loss": 0.7779432535171509,
+      "eval_runtime": 12.9731,
+      "eval_samples_per_second": 507.514,
+      "eval_steps_per_second": 7.94,
+      "step": 1690
+    },
+    {
+      "epoch": 0.19502122289778592,
+      "grad_norm": 3.1423346996307373,
+      "learning_rate": 6.4984709480122335e-06,
+      "loss": 1.3327,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19502122289778592,
+      "eval_loss": 0.7756665945053101,
+      "eval_runtime": 12.9767,
+      "eval_samples_per_second": 507.371,
+      "eval_steps_per_second": 7.937,
+      "step": 1700
+    },
+    {
+      "epoch": 0.19616840656189055,
+      "eval_loss": 0.7737661004066467,
+      "eval_runtime": 12.9633,
+      "eval_samples_per_second": 507.896,
+      "eval_steps_per_second": 7.946,
+      "step": 1710
+    },
+    {
+      "epoch": 0.19731559022599518,
+      "eval_loss": 0.7719039916992188,
+      "eval_runtime": 12.9798,
+      "eval_samples_per_second": 507.248,
+      "eval_steps_per_second": 7.935,
+      "step": 1720
+    },
+    {
+      "epoch": 0.1984627738900998,
+      "eval_loss": 0.7699739933013916,
+      "eval_runtime": 12.9817,
+      "eval_samples_per_second": 507.177,
+      "eval_steps_per_second": 7.934,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19960995755420444,
+      "eval_loss": 0.7679038643836975,
+      "eval_runtime": 13.0593,
+      "eval_samples_per_second": 504.164,
+      "eval_steps_per_second": 7.887,
+      "step": 1740
+    },
+    {
+      "epoch": 0.20075714121830904,
+      "eval_loss": 0.765801191329956,
+      "eval_runtime": 12.9827,
+      "eval_samples_per_second": 507.136,
+      "eval_steps_per_second": 7.934,
+      "step": 1750
+    },
+    {
+      "epoch": 0.20190432488241367,
+      "eval_loss": 0.7640795707702637,
+      "eval_runtime": 12.9885,
+      "eval_samples_per_second": 506.908,
+      "eval_steps_per_second": 7.93,
+      "step": 1760
+    },
+    {
+      "epoch": 0.2030515085465183,
+      "eval_loss": 0.7621497511863708,
+      "eval_runtime": 12.9872,
+      "eval_samples_per_second": 506.96,
+      "eval_steps_per_second": 7.931,
+      "step": 1770
+    },
+    {
+      "epoch": 0.20419869221062292,
+      "eval_loss": 0.7601312398910522,
+      "eval_runtime": 12.9805,
+      "eval_samples_per_second": 507.223,
+      "eval_steps_per_second": 7.935,
+      "step": 1780
+    },
+    {
+      "epoch": 0.20534587587472755,
+      "eval_loss": 0.7580214738845825,
+      "eval_runtime": 12.9849,
+      "eval_samples_per_second": 507.05,
+      "eval_steps_per_second": 7.932,
+      "step": 1790
+    },
+    {
+      "epoch": 0.20649305953883218,
+      "grad_norm": 3.4953625202178955,
+      "learning_rate": 6.880733944954129e-06,
+      "loss": 1.2804,
+      "step": 1800
+    },
+    {
+      "epoch": 0.20649305953883218,
+      "eval_loss": 0.7558028697967529,
+      "eval_runtime": 12.989,
+      "eval_samples_per_second": 506.889,
+      "eval_steps_per_second": 7.93,
+      "step": 1800
+    },
+    {
+      "epoch": 0.20764024320293678,
+      "eval_loss": 0.7535884380340576,
+      "eval_runtime": 12.9795,
+      "eval_samples_per_second": 507.262,
+      "eval_steps_per_second": 7.936,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2087874268670414,
+      "eval_loss": 0.7514472603797913,
+      "eval_runtime": 12.9901,
+      "eval_samples_per_second": 506.846,
+      "eval_steps_per_second": 7.929,
+      "step": 1820
+    },
+    {
+      "epoch": 0.20993461053114604,
+      "eval_loss": 0.7493338584899902,
+      "eval_runtime": 13.0011,
+      "eval_samples_per_second": 506.418,
+      "eval_steps_per_second": 7.922,
+      "step": 1830
+    },
+    {
+      "epoch": 0.21108179419525067,
+      "eval_loss": 0.7473368644714355,
+      "eval_runtime": 13.0424,
+      "eval_samples_per_second": 504.815,
+      "eval_steps_per_second": 7.897,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2122289778593553,
+      "eval_loss": 0.7450571060180664,
+      "eval_runtime": 13.1033,
+      "eval_samples_per_second": 502.468,
+      "eval_steps_per_second": 7.861,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2133761615234599,
+      "eval_loss": 0.7429091930389404,
+      "eval_runtime": 12.9903,
+      "eval_samples_per_second": 506.841,
+      "eval_steps_per_second": 7.929,
+      "step": 1860
+    },
+    {
+      "epoch": 0.21452334518756452,
+      "eval_loss": 0.7408153414726257,
+      "eval_runtime": 12.983,
+      "eval_samples_per_second": 507.126,
+      "eval_steps_per_second": 7.933,
+      "step": 1870
+    },
+    {
+      "epoch": 0.21567052885166915,
+      "eval_loss": 0.7389461994171143,
+      "eval_runtime": 12.9922,
+      "eval_samples_per_second": 506.767,
+      "eval_steps_per_second": 7.928,
+      "step": 1880
+    },
+    {
+      "epoch": 0.21681771251577378,
+      "eval_loss": 0.7368388772010803,
+      "eval_runtime": 13.0283,
+      "eval_samples_per_second": 505.36,
+      "eval_steps_per_second": 7.906,
+      "step": 1890
+    },
+    {
+      "epoch": 0.2179648961798784,
+      "grad_norm": 2.869717597961426,
+      "learning_rate": 7.262996941896026e-06,
+      "loss": 1.2255,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2179648961798784,
+      "eval_loss": 0.73488450050354,
+      "eval_runtime": 12.9924,
+      "eval_samples_per_second": 506.759,
+      "eval_steps_per_second": 7.928,
+      "step": 1900
+    },
+    {
+      "epoch": 0.219112079843983,
+      "eval_loss": 0.7328305244445801,
+      "eval_runtime": 12.9892,
+      "eval_samples_per_second": 506.881,
+      "eval_steps_per_second": 7.93,
+      "step": 1910
+    },
+    {
+      "epoch": 0.22025926350808764,
+      "eval_loss": 0.7309767007827759,
+      "eval_runtime": 13.1016,
+      "eval_samples_per_second": 502.534,
+      "eval_steps_per_second": 7.862,
+      "step": 1920
+    },
+    {
+      "epoch": 0.22140644717219227,
+      "eval_loss": 0.7293325662612915,
+      "eval_runtime": 12.9943,
+      "eval_samples_per_second": 506.684,
+      "eval_steps_per_second": 7.927,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2225536308362969,
+      "eval_loss": 0.7277292013168335,
+      "eval_runtime": 13.0122,
+      "eval_samples_per_second": 505.986,
+      "eval_steps_per_second": 7.916,
+      "step": 1940
+    },
+    {
+      "epoch": 0.22370081450040152,
+      "eval_loss": 0.7258683443069458,
+      "eval_runtime": 13.0042,
+      "eval_samples_per_second": 506.298,
+      "eval_steps_per_second": 7.921,
+      "step": 1950
+    },
+    {
+      "epoch": 0.22484799816450612,
+      "eval_loss": 0.7239750027656555,
+      "eval_runtime": 13.0619,
+      "eval_samples_per_second": 504.061,
+      "eval_steps_per_second": 7.886,
+      "step": 1960
+    },
+    {
+      "epoch": 0.22599518182861075,
+      "eval_loss": 0.7221319079399109,
+      "eval_runtime": 12.9944,
+      "eval_samples_per_second": 506.679,
+      "eval_steps_per_second": 7.926,
+      "step": 1970
+    },
+    {
+      "epoch": 0.22714236549271538,
+      "eval_loss": 0.720324695110321,
+      "eval_runtime": 12.982,
+      "eval_samples_per_second": 507.165,
+      "eval_steps_per_second": 7.934,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22828954915682,
+      "eval_loss": 0.7184233069419861,
+      "eval_runtime": 12.9988,
+      "eval_samples_per_second": 506.507,
+      "eval_steps_per_second": 7.924,
+      "step": 1990
+    },
+    {
+      "epoch": 0.22943673282092464,
+      "grad_norm": 9.275369644165039,
+      "learning_rate": 7.645259938837921e-06,
+      "loss": 1.2635,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22943673282092464,
+      "eval_loss": 0.7165008187294006,
+      "eval_runtime": 12.9966,
+      "eval_samples_per_second": 506.594,
+      "eval_steps_per_second": 7.925,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23058391648502927,
+      "eval_loss": 0.7150009870529175,
+      "eval_runtime": 12.9901,
+      "eval_samples_per_second": 506.849,
+      "eval_steps_per_second": 7.929,
+      "step": 2010
+    },
+    {
+      "epoch": 0.23173110014913387,
+      "eval_loss": 0.7134894728660583,
+      "eval_runtime": 13.0392,
+      "eval_samples_per_second": 504.938,
+      "eval_steps_per_second": 7.899,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2328782838132385,
+      "eval_loss": 0.7117303013801575,
+      "eval_runtime": 13.0384,
+      "eval_samples_per_second": 504.971,
+      "eval_steps_per_second": 7.9,
+      "step": 2030
+    },
+    {
+      "epoch": 0.23402546747734312,
+      "eval_loss": 0.7099109292030334,
+      "eval_runtime": 12.9879,
+      "eval_samples_per_second": 506.932,
+      "eval_steps_per_second": 7.93,
+      "step": 2040
+    },
+    {
+      "epoch": 0.23517265114144775,
+      "eval_loss": 0.7084246873855591,
+      "eval_runtime": 12.9995,
+      "eval_samples_per_second": 506.481,
+      "eval_steps_per_second": 7.923,
+      "step": 2050
+    },
+    {
+      "epoch": 0.23631983480555238,
+      "eval_loss": 0.7067718505859375,
+      "eval_runtime": 12.9881,
+      "eval_samples_per_second": 506.925,
+      "eval_steps_per_second": 7.93,
+      "step": 2060
+    },
+    {
+      "epoch": 0.23746701846965698,
+      "eval_loss": 0.7053564786911011,
+      "eval_runtime": 13.1243,
+      "eval_samples_per_second": 501.665,
+      "eval_steps_per_second": 7.848,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2386142021337616,
+      "eval_loss": 0.7037129402160645,
+      "eval_runtime": 12.9868,
+      "eval_samples_per_second": 506.975,
+      "eval_steps_per_second": 7.931,
+      "step": 2080
+    },
+    {
+      "epoch": 0.23976138579786624,
+      "eval_loss": 0.7023361921310425,
+      "eval_runtime": 12.9909,
+      "eval_samples_per_second": 506.818,
+      "eval_steps_per_second": 7.929,
+      "step": 2090
+    },
+    {
+      "epoch": 0.24090856946197087,
+      "grad_norm": 2.9726195335388184,
+      "learning_rate": 8.027522935779817e-06,
+      "loss": 1.1912,
+      "step": 2100
+    },
+    {
+      "epoch": 0.24090856946197087,
+      "eval_loss": 0.7009308338165283,
+      "eval_runtime": 13.1789,
+      "eval_samples_per_second": 499.587,
+      "eval_steps_per_second": 7.816,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2420557531260755,
+      "eval_loss": 0.6990575790405273,
+      "eval_runtime": 13.0982,
+      "eval_samples_per_second": 502.665,
+      "eval_steps_per_second": 7.864,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2432029367901801,
+      "eval_loss": 0.6973636746406555,
+      "eval_runtime": 13.0024,
+      "eval_samples_per_second": 506.369,
+      "eval_steps_per_second": 7.922,
+      "step": 2120
+    },
+    {
+      "epoch": 0.24435012045428472,
+      "eval_loss": 0.6962077617645264,
+      "eval_runtime": 12.9266,
+      "eval_samples_per_second": 509.337,
+      "eval_steps_per_second": 7.968,
+      "step": 2130
+    },
+    {
+      "epoch": 0.24549730411838935,
+      "eval_loss": 0.6949887275695801,
+      "eval_runtime": 12.9495,
+      "eval_samples_per_second": 508.438,
+      "eval_steps_per_second": 7.954,
+      "step": 2140
+    },
+    {
+      "epoch": 0.24664448778249398,
+      "eval_loss": 0.6937941312789917,
+      "eval_runtime": 12.9625,
+      "eval_samples_per_second": 507.926,
+      "eval_steps_per_second": 7.946,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2477916714465986,
+      "eval_loss": 0.6922134160995483,
+      "eval_runtime": 13.0384,
+      "eval_samples_per_second": 504.971,
+      "eval_steps_per_second": 7.9,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2489388551107032,
+      "eval_loss": 0.6908959150314331,
+      "eval_runtime": 13.0162,
+      "eval_samples_per_second": 505.831,
+      "eval_steps_per_second": 7.913,
+      "step": 2170
+    },
+    {
+      "epoch": 0.25008603877480784,
+      "eval_loss": 0.6896905899047852,
+      "eval_runtime": 13.0281,
+      "eval_samples_per_second": 505.368,
+      "eval_steps_per_second": 7.906,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2512332224389125,
+      "eval_loss": 0.6883853673934937,
+      "eval_runtime": 13.059,
+      "eval_samples_per_second": 504.173,
+      "eval_steps_per_second": 7.887,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2523804061030171,
+      "grad_norm": 9.038525581359863,
+      "learning_rate": 8.409785932721713e-06,
+      "loss": 1.2144,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2523804061030171,
+      "eval_loss": 0.68684321641922,
+      "eval_runtime": 12.9342,
+      "eval_samples_per_second": 509.039,
+      "eval_steps_per_second": 7.963,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2535275897671217,
+      "eval_loss": 0.6855539679527283,
+      "eval_runtime": 12.947,
+      "eval_samples_per_second": 508.536,
+      "eval_steps_per_second": 7.956,
+      "step": 2210
+    },
+    {
+      "epoch": 0.25467477343122635,
+      "eval_loss": 0.684305727481842,
+      "eval_runtime": 13.0482,
+      "eval_samples_per_second": 504.59,
+      "eval_steps_per_second": 7.894,
+      "step": 2220
+    },
+    {
+      "epoch": 0.25582195709533095,
+      "eval_loss": 0.6828807592391968,
+      "eval_runtime": 13.0317,
+      "eval_samples_per_second": 505.23,
+      "eval_steps_per_second": 7.904,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2569691407594356,
+      "eval_loss": 0.6817071437835693,
+      "eval_runtime": 13.0489,
+      "eval_samples_per_second": 504.564,
+      "eval_steps_per_second": 7.893,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2581163244235402,
+      "eval_loss": 0.6803813576698303,
+      "eval_runtime": 13.0166,
+      "eval_samples_per_second": 505.817,
+      "eval_steps_per_second": 7.913,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2592635080876448,
+      "eval_loss": 0.6788561344146729,
+      "eval_runtime": 13.018,
+      "eval_samples_per_second": 505.76,
+      "eval_steps_per_second": 7.912,
+      "step": 2260
+    },
+    {
+      "epoch": 0.26041069175174947,
+      "eval_loss": 0.6775069832801819,
+      "eval_runtime": 12.9958,
+      "eval_samples_per_second": 506.623,
+      "eval_steps_per_second": 7.926,
+      "step": 2270
+    },
+    {
+      "epoch": 0.26155787541585407,
+      "eval_loss": 0.6762964725494385,
+      "eval_runtime": 12.9456,
+      "eval_samples_per_second": 508.59,
+      "eval_steps_per_second": 7.956,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2627050590799587,
+      "eval_loss": 0.6750586032867432,
+      "eval_runtime": 12.9511,
+      "eval_samples_per_second": 508.372,
+      "eval_steps_per_second": 7.953,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2638522427440633,
+      "grad_norm": 3.329761266708374,
+      "learning_rate": 8.79204892966361e-06,
+      "loss": 1.1498,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2638522427440633,
+      "eval_loss": 0.6738935112953186,
+      "eval_runtime": 13.0495,
+      "eval_samples_per_second": 504.542,
+      "eval_steps_per_second": 7.893,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2649994264081679,
+      "eval_loss": 0.6725099682807922,
+      "eval_runtime": 13.0328,
+      "eval_samples_per_second": 505.188,
+      "eval_steps_per_second": 7.903,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2661466100722726,
+      "eval_loss": 0.6710599660873413,
+      "eval_runtime": 13.0505,
+      "eval_samples_per_second": 504.502,
+      "eval_steps_per_second": 7.892,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2672937937363772,
+      "eval_loss": 0.6697894930839539,
+      "eval_runtime": 13.0616,
+      "eval_samples_per_second": 504.074,
+      "eval_steps_per_second": 7.886,
+      "step": 2330
+    },
+    {
+      "epoch": 0.26844097740048184,
+      "eval_loss": 0.668381929397583,
+      "eval_runtime": 12.9197,
+      "eval_samples_per_second": 509.608,
+      "eval_steps_per_second": 7.972,
+      "step": 2340
+    },
+    {
+      "epoch": 0.26958816106458644,
+      "eval_loss": 0.6666426062583923,
+      "eval_runtime": 13.0254,
+      "eval_samples_per_second": 505.475,
+      "eval_steps_per_second": 7.908,
+      "step": 2350
+    },
+    {
+      "epoch": 0.27073534472869104,
+      "eval_loss": 0.6652824878692627,
+      "eval_runtime": 13.0301,
+      "eval_samples_per_second": 505.293,
+      "eval_steps_per_second": 7.905,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2718825283927957,
+      "eval_loss": 0.6638170480728149,
+      "eval_runtime": 12.917,
+      "eval_samples_per_second": 509.715,
+      "eval_steps_per_second": 7.974,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2730297120569003,
+      "eval_loss": 0.6620729565620422,
+      "eval_runtime": 13.0292,
+      "eval_samples_per_second": 505.327,
+      "eval_steps_per_second": 7.905,
+      "step": 2380
+    },
+    {
+      "epoch": 0.27417689572100495,
+      "eval_loss": 0.660891056060791,
+      "eval_runtime": 13.022,
+      "eval_samples_per_second": 505.604,
+      "eval_steps_per_second": 7.91,
+      "step": 2390
+    },
+    {
+      "epoch": 0.27532407938510955,
+      "grad_norm": 2.91483211517334,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 1.1446,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27532407938510955,
+      "eval_loss": 0.6595732569694519,
+      "eval_runtime": 13.036,
+      "eval_samples_per_second": 505.063,
+      "eval_steps_per_second": 7.901,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27647126304921416,
+      "eval_loss": 0.6582459807395935,
+      "eval_runtime": 12.9309,
+      "eval_samples_per_second": 509.168,
+      "eval_steps_per_second": 7.965,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2776184467133188,
+      "eval_loss": 0.6568032503128052,
+      "eval_runtime": 13.0039,
+      "eval_samples_per_second": 506.309,
+      "eval_steps_per_second": 7.921,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2787656303774234,
+      "eval_loss": 0.6553359031677246,
+      "eval_runtime": 12.9157,
+      "eval_samples_per_second": 509.766,
+      "eval_steps_per_second": 7.975,
+      "step": 2430
+    },
+    {
+      "epoch": 0.27991281404152807,
+      "eval_loss": 0.6541372537612915,
+      "eval_runtime": 12.9901,
+      "eval_samples_per_second": 506.846,
+      "eval_steps_per_second": 7.929,
+      "step": 2440
+    },
+    {
+      "epoch": 0.28105999770563267,
+      "eval_loss": 0.6527132391929626,
+      "eval_runtime": 12.9209,
+      "eval_samples_per_second": 509.562,
+      "eval_steps_per_second": 7.972,
+      "step": 2450
+    },
+    {
+      "epoch": 0.28220718136973727,
+      "eval_loss": 0.6512923836708069,
+      "eval_runtime": 12.9637,
+      "eval_samples_per_second": 507.879,
+      "eval_steps_per_second": 7.945,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2833543650338419,
+      "eval_loss": 0.6496260762214661,
+      "eval_runtime": 13.0604,
+      "eval_samples_per_second": 504.119,
+      "eval_steps_per_second": 7.886,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2845015486979465,
+      "eval_loss": 0.6483332514762878,
+      "eval_runtime": 13.0233,
+      "eval_samples_per_second": 505.557,
+      "eval_steps_per_second": 7.909,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2856487323620512,
+      "eval_loss": 0.647476077079773,
+      "eval_runtime": 13.0344,
+      "eval_samples_per_second": 505.124,
+      "eval_steps_per_second": 7.902,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2867959160261558,
+      "grad_norm": 3.6412861347198486,
+      "learning_rate": 9.556574923547402e-06,
+      "loss": 1.1309,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2867959160261558,
+      "eval_loss": 0.6465412378311157,
+      "eval_runtime": 12.9143,
+      "eval_samples_per_second": 509.824,
+      "eval_steps_per_second": 7.976,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2879430996902604,
+      "eval_loss": 0.6455466747283936,
+      "eval_runtime": 13.0269,
+      "eval_samples_per_second": 505.417,
+      "eval_steps_per_second": 7.907,
+      "step": 2510
+    },
+    {
+      "epoch": 0.28909028335436504,
+      "eval_loss": 0.6446804404258728,
+      "eval_runtime": 13.0271,
+      "eval_samples_per_second": 505.407,
+      "eval_steps_per_second": 7.907,
+      "step": 2520
+    },
+    {
+      "epoch": 0.29023746701846964,
+      "eval_loss": 0.6436929702758789,
+      "eval_runtime": 13.0149,
+      "eval_samples_per_second": 505.882,
+      "eval_steps_per_second": 7.914,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2913846506825743,
+      "eval_loss": 0.6427727341651917,
+      "eval_runtime": 13.0187,
+      "eval_samples_per_second": 505.734,
+      "eval_steps_per_second": 7.912,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2925318343466789,
+      "eval_loss": 0.6415053009986877,
+      "eval_runtime": 13.0192,
+      "eval_samples_per_second": 505.716,
+      "eval_steps_per_second": 7.911,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2936790180107835,
+      "eval_loss": 0.6403249502182007,
+      "eval_runtime": 13.0143,
+      "eval_samples_per_second": 505.905,
+      "eval_steps_per_second": 7.914,
+      "step": 2560
+    },
+    {
+      "epoch": 0.29482620167488816,
+      "eval_loss": 0.639183759689331,
+      "eval_runtime": 13.0127,
+      "eval_samples_per_second": 505.966,
+      "eval_steps_per_second": 7.915,
+      "step": 2570
+    },
+    {
+      "epoch": 0.29597338533899276,
+      "eval_loss": 0.6380952000617981,
+      "eval_runtime": 13.0253,
+      "eval_samples_per_second": 505.478,
+      "eval_steps_per_second": 7.908,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2971205690030974,
+      "eval_loss": 0.6371238231658936,
+      "eval_runtime": 12.9266,
+      "eval_samples_per_second": 509.339,
+      "eval_steps_per_second": 7.968,
+      "step": 2590
+    },
+    {
+      "epoch": 0.298267752667202,
+      "grad_norm": 3.0183558464050293,
+      "learning_rate": 9.938837920489298e-06,
+      "loss": 1.1006,
+      "step": 2600
+    },
+    {
+      "epoch": 0.298267752667202,
+      "eval_loss": 0.6357853412628174,
+      "eval_runtime": 12.9136,
+      "eval_samples_per_second": 509.848,
+      "eval_steps_per_second": 7.976,
+      "step": 2600
+    },
+    {
+      "epoch": 0.29941493633130667,
+      "eval_loss": 0.6347528696060181,
+      "eval_runtime": 13.031,
+      "eval_samples_per_second": 505.258,
+      "eval_steps_per_second": 7.904,
+      "step": 2610
+    },
+    {
+      "epoch": 0.30056211999541127,
+      "eval_loss": 0.634018063545227,
+      "eval_runtime": 13.0415,
+      "eval_samples_per_second": 504.85,
+      "eval_steps_per_second": 7.898,
+      "step": 2620
+    },
+    {
+      "epoch": 0.30170930365951587,
+      "eval_loss": 0.6330114006996155,
+      "eval_runtime": 12.9387,
+      "eval_samples_per_second": 508.863,
+      "eval_steps_per_second": 7.961,
+      "step": 2630
+    },
+    {
+      "epoch": 0.3028564873236205,
+      "eval_loss": 0.6319145560264587,
+      "eval_runtime": 12.919,
+      "eval_samples_per_second": 509.636,
+      "eval_steps_per_second": 7.973,
+      "step": 2640
+    },
+    {
+      "epoch": 0.3040036709877251,
+      "eval_loss": 0.6307594180107117,
+      "eval_runtime": 12.9268,
+      "eval_samples_per_second": 509.328,
+      "eval_steps_per_second": 7.968,
+      "step": 2650
+    },
+    {
+      "epoch": 0.3051508546518298,
+      "eval_loss": 0.6300274133682251,
+      "eval_runtime": 13.0,
+      "eval_samples_per_second": 506.461,
+      "eval_steps_per_second": 7.923,
+      "step": 2660
+    },
+    {
+      "epoch": 0.3062980383159344,
+      "eval_loss": 0.6291391849517822,
+      "eval_runtime": 13.023,
+      "eval_samples_per_second": 505.566,
+      "eval_steps_per_second": 7.909,
+      "step": 2670
+    },
+    {
+      "epoch": 0.307445221980039,
+      "eval_loss": 0.6279686093330383,
+      "eval_runtime": 12.9099,
+      "eval_samples_per_second": 509.995,
+      "eval_steps_per_second": 7.978,
+      "step": 2680
+    },
+    {
+      "epoch": 0.30859240564414364,
+      "eval_loss": 0.6267827153205872,
+      "eval_runtime": 12.9942,
+      "eval_samples_per_second": 506.687,
+      "eval_steps_per_second": 7.927,
+      "step": 2690
+    },
+    {
+      "epoch": 0.30973958930824824,
+      "grad_norm": 2.923506021499634,
+      "learning_rate": 9.964308476736776e-06,
+      "loss": 1.0772,
+      "step": 2700
+    },
+    {
+      "epoch": 0.30973958930824824,
+      "eval_loss": 0.6254075765609741,
+      "eval_runtime": 13.0366,
+      "eval_samples_per_second": 505.041,
+      "eval_steps_per_second": 7.901,
+      "step": 2700
+    },
+    {
+      "epoch": 0.3108867729723529,
+      "eval_loss": 0.6242946982383728,
+      "eval_runtime": 13.0323,
+      "eval_samples_per_second": 505.207,
+      "eval_steps_per_second": 7.903,
+      "step": 2710
+    },
+    {
+      "epoch": 0.3120339566364575,
+      "eval_loss": 0.6231787204742432,
+      "eval_runtime": 12.9273,
+      "eval_samples_per_second": 509.309,
+      "eval_steps_per_second": 7.968,
+      "step": 2720
+    },
+    {
+      "epoch": 0.3131811403005621,
+      "eval_loss": 0.62239009141922,
+      "eval_runtime": 12.9153,
+      "eval_samples_per_second": 509.781,
+      "eval_steps_per_second": 7.975,
+      "step": 2730
+    },
+    {
+      "epoch": 0.31432832396466676,
+      "eval_loss": 0.6214902400970459,
+      "eval_runtime": 12.9178,
+      "eval_samples_per_second": 509.684,
+      "eval_steps_per_second": 7.973,
+      "step": 2740
+    },
+    {
+      "epoch": 0.31547550762877136,
+      "eval_loss": 0.6204728484153748,
+      "eval_runtime": 12.9141,
+      "eval_samples_per_second": 509.831,
+      "eval_steps_per_second": 7.976,
+      "step": 2750
+    },
+    {
+      "epoch": 0.316622691292876,
+      "eval_loss": 0.6194381713867188,
+      "eval_runtime": 12.9151,
+      "eval_samples_per_second": 509.791,
+      "eval_steps_per_second": 7.975,
+      "step": 2760
+    },
+    {
+      "epoch": 0.3177698749569806,
+      "eval_loss": 0.6183201670646667,
+      "eval_runtime": 12.9127,
+      "eval_samples_per_second": 509.887,
+      "eval_steps_per_second": 7.977,
+      "step": 2770
+    },
+    {
+      "epoch": 0.3189170586210852,
+      "eval_loss": 0.6171374917030334,
+      "eval_runtime": 12.9112,
+      "eval_samples_per_second": 509.944,
+      "eval_steps_per_second": 7.978,
+      "step": 2780
+    },
+    {
+      "epoch": 0.32006424228518987,
+      "eval_loss": 0.6160247921943665,
+      "eval_runtime": 13.0361,
+      "eval_samples_per_second": 505.06,
+      "eval_steps_per_second": 7.901,
+      "step": 2790
+    },
+    {
+      "epoch": 0.32121142594929447,
+      "grad_norm": 2.8031253814697266,
+      "learning_rate": 9.92181856809008e-06,
+      "loss": 1.0648,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32121142594929447,
+      "eval_loss": 0.6153244376182556,
+      "eval_runtime": 13.0215,
+      "eval_samples_per_second": 505.624,
+      "eval_steps_per_second": 7.91,
+      "step": 2800
+    },
+    {
+      "epoch": 0.3223586096133991,
+      "eval_loss": 0.6140704154968262,
+      "eval_runtime": 13.0201,
+      "eval_samples_per_second": 505.679,
+      "eval_steps_per_second": 7.911,
+      "step": 2810
+    },
+    {
+      "epoch": 0.32350579327750373,
+      "eval_loss": 0.6129491329193115,
+      "eval_runtime": 13.0287,
+      "eval_samples_per_second": 505.347,
+      "eval_steps_per_second": 7.906,
+      "step": 2820
+    },
+    {
+      "epoch": 0.32465297694160833,
+      "eval_loss": 0.6119452118873596,
+      "eval_runtime": 13.045,
+      "eval_samples_per_second": 504.716,
+      "eval_steps_per_second": 7.896,
+      "step": 2830
+    },
+    {
+      "epoch": 0.325800160605713,
+      "eval_loss": 0.6109396815299988,
+      "eval_runtime": 13.0301,
+      "eval_samples_per_second": 505.29,
+      "eval_steps_per_second": 7.905,
+      "step": 2840
+    },
+    {
+      "epoch": 0.3269473442698176,
+      "eval_loss": 0.6099222898483276,
+      "eval_runtime": 13.0261,
+      "eval_samples_per_second": 505.446,
+      "eval_steps_per_second": 7.907,
+      "step": 2850
+    },
+    {
+      "epoch": 0.32809452793392224,
+      "eval_loss": 0.6087589859962463,
+      "eval_runtime": 12.9193,
+      "eval_samples_per_second": 509.625,
+      "eval_steps_per_second": 7.973,
+      "step": 2860
+    },
+    {
+      "epoch": 0.32924171159802684,
+      "eval_loss": 0.6078950762748718,
+      "eval_runtime": 12.9854,
+      "eval_samples_per_second": 507.031,
+      "eval_steps_per_second": 7.932,
+      "step": 2870
+    },
+    {
+      "epoch": 0.33038889526213144,
+      "eval_loss": 0.60728919506073,
+      "eval_runtime": 13.0282,
+      "eval_samples_per_second": 505.366,
+      "eval_steps_per_second": 7.906,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3315360789262361,
+      "eval_loss": 0.606285810470581,
+      "eval_runtime": 13.0385,
+      "eval_samples_per_second": 504.964,
+      "eval_steps_per_second": 7.9,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3326832625903407,
+      "grad_norm": 3.0763423442840576,
+      "learning_rate": 9.879328659443383e-06,
+      "loss": 1.0398,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3326832625903407,
+      "eval_loss": 0.6053714752197266,
+      "eval_runtime": 13.0412,
+      "eval_samples_per_second": 504.863,
+      "eval_steps_per_second": 7.898,
+      "step": 2900
+    },
+    {
+      "epoch": 0.33383044625444536,
+      "eval_loss": 0.6043825149536133,
+      "eval_runtime": 13.0358,
+      "eval_samples_per_second": 505.072,
+      "eval_steps_per_second": 7.901,
+      "step": 2910
+    },
+    {
+      "epoch": 0.33497762991854996,
+      "eval_loss": 0.6032926440238953,
+      "eval_runtime": 13.0011,
+      "eval_samples_per_second": 506.418,
+      "eval_steps_per_second": 7.922,
+      "step": 2920
+    },
+    {
+      "epoch": 0.33612481358265456,
+      "eval_loss": 0.6022117137908936,
+      "eval_runtime": 12.922,
+      "eval_samples_per_second": 509.517,
+      "eval_steps_per_second": 7.971,
+      "step": 2930
+    },
+    {
+      "epoch": 0.3372719972467592,
+      "eval_loss": 0.6011677980422974,
+      "eval_runtime": 12.9117,
+      "eval_samples_per_second": 509.924,
+      "eval_steps_per_second": 7.977,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3384191809108638,
+      "eval_loss": 0.6003371477127075,
+      "eval_runtime": 13.0239,
+      "eval_samples_per_second": 505.531,
+      "eval_steps_per_second": 7.909,
+      "step": 2950
+    },
+    {
+      "epoch": 0.33956636457496847,
+      "eval_loss": 0.5993051528930664,
+      "eval_runtime": 12.9359,
+      "eval_samples_per_second": 508.971,
+      "eval_steps_per_second": 7.962,
+      "step": 2960
+    },
+    {
+      "epoch": 0.34071354823907307,
+      "eval_loss": 0.5986045598983765,
+      "eval_runtime": 12.9502,
+      "eval_samples_per_second": 508.409,
+      "eval_steps_per_second": 7.954,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3418607319031777,
+      "eval_loss": 0.5977791547775269,
+      "eval_runtime": 13.0367,
+      "eval_samples_per_second": 505.037,
+      "eval_steps_per_second": 7.901,
+      "step": 2980
+    },
+    {
+      "epoch": 0.34300791556728233,
+      "eval_loss": 0.596732497215271,
+      "eval_runtime": 13.0228,
+      "eval_samples_per_second": 505.576,
+      "eval_steps_per_second": 7.909,
+      "step": 2990
+    },
+    {
+      "epoch": 0.34415509923138693,
+      "grad_norm": 2.941866636276245,
+      "learning_rate": 9.836838750796687e-06,
+      "loss": 1.0256,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34415509923138693,
+      "eval_loss": 0.5959278345108032,
+      "eval_runtime": 12.9239,
+      "eval_samples_per_second": 509.443,
+      "eval_steps_per_second": 7.97,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3453022828954916,
+      "eval_loss": 0.5946770906448364,
+      "eval_runtime": 12.9123,
+      "eval_samples_per_second": 509.903,
+      "eval_steps_per_second": 7.977,
+      "step": 3010
+    },
+    {
+      "epoch": 0.3464494665595962,
+      "eval_loss": 0.593720555305481,
+      "eval_runtime": 12.92,
+      "eval_samples_per_second": 509.596,
+      "eval_steps_per_second": 7.972,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3475966502237008,
+      "eval_loss": 0.5929400324821472,
+      "eval_runtime": 12.9184,
+      "eval_samples_per_second": 509.661,
+      "eval_steps_per_second": 7.973,
+      "step": 3030
+    },
+    {
+      "epoch": 0.34874383388780544,
+      "eval_loss": 0.5920357704162598,
+      "eval_runtime": 12.9232,
+      "eval_samples_per_second": 509.47,
+      "eval_steps_per_second": 7.97,
+      "step": 3040
+    },
+    {
+      "epoch": 0.34989101755191004,
+      "eval_loss": 0.5908279418945312,
+      "eval_runtime": 12.9197,
+      "eval_samples_per_second": 509.609,
+      "eval_steps_per_second": 7.972,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3510382012160147,
+      "eval_loss": 0.5896555185317993,
+      "eval_runtime": 13.0254,
+      "eval_samples_per_second": 505.473,
+      "eval_steps_per_second": 7.908,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3521853848801193,
+      "eval_loss": 0.5887726545333862,
+      "eval_runtime": 13.0242,
+      "eval_samples_per_second": 505.522,
+      "eval_steps_per_second": 7.908,
+      "step": 3070
+    },
+    {
+      "epoch": 0.35333256854422396,
+      "eval_loss": 0.5881541967391968,
+      "eval_runtime": 13.0358,
+      "eval_samples_per_second": 505.069,
+      "eval_steps_per_second": 7.901,
+      "step": 3080
+    },
+    {
+      "epoch": 0.35447975220832856,
+      "eval_loss": 0.5873861312866211,
+      "eval_runtime": 12.9582,
+      "eval_samples_per_second": 508.095,
+      "eval_steps_per_second": 7.949,
+      "step": 3090
+    },
+    {
+      "epoch": 0.35562693587243316,
+      "grad_norm": 3.1805124282836914,
+      "learning_rate": 9.79434884214999e-06,
+      "loss": 1.0489,
+      "step": 3100
+    },
+    {
+      "epoch": 0.35562693587243316,
+      "eval_loss": 0.5868309736251831,
+      "eval_runtime": 13.0375,
+      "eval_samples_per_second": 505.003,
+      "eval_steps_per_second": 7.9,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3567741195365378,
+      "eval_loss": 0.5860410928726196,
+      "eval_runtime": 12.9224,
+      "eval_samples_per_second": 509.501,
+      "eval_steps_per_second": 7.971,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3579213032006424,
+      "eval_loss": 0.5853646993637085,
+      "eval_runtime": 12.9171,
+      "eval_samples_per_second": 509.711,
+      "eval_steps_per_second": 7.974,
+      "step": 3120
+    },
+    {
+      "epoch": 0.35906848686474707,
+      "eval_loss": 0.5839424729347229,
+      "eval_runtime": 12.9136,
+      "eval_samples_per_second": 509.848,
+      "eval_steps_per_second": 7.976,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3602156705288517,
+      "eval_loss": 0.5830027461051941,
+      "eval_runtime": 12.9177,
+      "eval_samples_per_second": 509.686,
+      "eval_steps_per_second": 7.974,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3613628541929563,
+      "eval_loss": 0.5822171568870544,
+      "eval_runtime": 13.0378,
+      "eval_samples_per_second": 504.995,
+      "eval_steps_per_second": 7.9,
+      "step": 3150
+    },
+    {
+      "epoch": 0.36251003785706093,
+      "eval_loss": 0.5814335942268372,
+      "eval_runtime": 12.9196,
+      "eval_samples_per_second": 509.612,
+      "eval_steps_per_second": 7.972,
+      "step": 3160
+    },
+    {
+      "epoch": 0.36365722152116553,
+      "eval_loss": 0.5807969570159912,
+      "eval_runtime": 12.9184,
+      "eval_samples_per_second": 509.659,
+      "eval_steps_per_second": 7.973,
+      "step": 3170
+    },
+    {
+      "epoch": 0.3648044051852702,
+      "eval_loss": 0.5801600813865662,
+      "eval_runtime": 12.9016,
+      "eval_samples_per_second": 510.323,
+      "eval_steps_per_second": 7.983,
+      "step": 3180
+    },
+    {
+      "epoch": 0.3659515888493748,
+      "eval_loss": 0.5793916583061218,
+      "eval_runtime": 12.9241,
+      "eval_samples_per_second": 509.434,
+      "eval_steps_per_second": 7.97,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3670987725134794,
+      "grad_norm": 3.2195188999176025,
+      "learning_rate": 9.751858933503294e-06,
+      "loss": 1.038,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3670987725134794,
+      "eval_loss": 0.578770637512207,
+      "eval_runtime": 13.0163,
+      "eval_samples_per_second": 505.828,
+      "eval_steps_per_second": 7.913,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36824595617758404,
+      "eval_loss": 0.5778489112854004,
+      "eval_runtime": 13.0447,
+      "eval_samples_per_second": 504.726,
+      "eval_steps_per_second": 7.896,
+      "step": 3210
+    },
+    {
+      "epoch": 0.36939313984168864,
+      "eval_loss": 0.577012300491333,
+      "eval_runtime": 13.038,
+      "eval_samples_per_second": 504.984,
+      "eval_steps_per_second": 7.9,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3705403235057933,
+      "eval_loss": 0.5763078331947327,
+      "eval_runtime": 12.9775,
+      "eval_samples_per_second": 507.339,
+      "eval_steps_per_second": 7.937,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3716875071698979,
+      "eval_loss": 0.5751996040344238,
+      "eval_runtime": 12.9227,
+      "eval_samples_per_second": 509.492,
+      "eval_steps_per_second": 7.97,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3728346908340025,
+      "eval_loss": 0.5744589567184448,
+      "eval_runtime": 12.9251,
+      "eval_samples_per_second": 509.398,
+      "eval_steps_per_second": 7.969,
+      "step": 3250
+    },
+    {
+      "epoch": 0.37398187449810716,
+      "eval_loss": 0.5737271308898926,
+      "eval_runtime": 12.9481,
+      "eval_samples_per_second": 508.493,
+      "eval_steps_per_second": 7.955,
+      "step": 3260
+    },
+    {
+      "epoch": 0.37512905816221176,
+      "eval_loss": 0.5727923512458801,
+      "eval_runtime": 13.0412,
+      "eval_samples_per_second": 504.861,
+      "eval_steps_per_second": 7.898,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3762762418263164,
+      "eval_loss": 0.5719892382621765,
+      "eval_runtime": 12.9946,
+      "eval_samples_per_second": 506.674,
+      "eval_steps_per_second": 7.926,
+      "step": 3280
+    },
+    {
+      "epoch": 0.377423425490421,
+      "eval_loss": 0.5713133811950684,
+      "eval_runtime": 12.9204,
+      "eval_samples_per_second": 509.58,
+      "eval_steps_per_second": 7.972,
+      "step": 3290
+    },
+    {
+      "epoch": 0.3785706091545256,
+      "grad_norm": 2.951904058456421,
+      "learning_rate": 9.709369024856598e-06,
+      "loss": 1.0058,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3785706091545256,
+      "eval_loss": 0.5707195401191711,
+      "eval_runtime": 12.9678,
+      "eval_samples_per_second": 507.718,
+      "eval_steps_per_second": 7.943,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3797177928186303,
+      "eval_loss": 0.5699546933174133,
+      "eval_runtime": 12.9635,
+      "eval_samples_per_second": 507.889,
+      "eval_steps_per_second": 7.945,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3808649764827349,
+      "eval_loss": 0.5689657926559448,
+      "eval_runtime": 12.9065,
+      "eval_samples_per_second": 510.131,
+      "eval_steps_per_second": 7.98,
+      "step": 3320
+    },
+    {
+      "epoch": 0.38201216014683953,
+      "eval_loss": 0.5680745244026184,
+      "eval_runtime": 12.9206,
+      "eval_samples_per_second": 509.574,
+      "eval_steps_per_second": 7.972,
+      "step": 3330
+    },
+    {
+      "epoch": 0.38315934381094413,
+      "eval_loss": 0.5672716498374939,
+      "eval_runtime": 12.9202,
+      "eval_samples_per_second": 509.591,
+      "eval_steps_per_second": 7.972,
+      "step": 3340
+    },
+    {
+      "epoch": 0.38430652747504873,
+      "eval_loss": 0.5669021606445312,
+      "eval_runtime": 12.9282,
+      "eval_samples_per_second": 509.275,
+      "eval_steps_per_second": 7.967,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3854537111391534,
+      "eval_loss": 0.5667062997817993,
+      "eval_runtime": 12.9165,
+      "eval_samples_per_second": 509.737,
+      "eval_steps_per_second": 7.974,
+      "step": 3360
+    },
+    {
+      "epoch": 0.386600894803258,
+      "eval_loss": 0.5664732456207275,
+      "eval_runtime": 12.9163,
+      "eval_samples_per_second": 509.742,
+      "eval_steps_per_second": 7.974,
+      "step": 3370
+    },
+    {
+      "epoch": 0.38774807846736264,
+      "eval_loss": 0.5658523440361023,
+      "eval_runtime": 12.9214,
+      "eval_samples_per_second": 509.542,
+      "eval_steps_per_second": 7.971,
+      "step": 3380
+    },
+    {
+      "epoch": 0.38889526213146725,
+      "eval_loss": 0.5650349855422974,
+      "eval_runtime": 12.9209,
+      "eval_samples_per_second": 509.562,
+      "eval_steps_per_second": 7.972,
+      "step": 3390
+    },
+    {
+      "epoch": 0.39004244579557185,
+      "grad_norm": 2.8982646465301514,
+      "learning_rate": 9.666879116209901e-06,
+      "loss": 1.0413,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39004244579557185,
+      "eval_loss": 0.5644938349723816,
+      "eval_runtime": 12.9147,
+      "eval_samples_per_second": 509.808,
+      "eval_steps_per_second": 7.975,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3911896294596765,
+      "eval_loss": 0.5641273260116577,
+      "eval_runtime": 12.9416,
+      "eval_samples_per_second": 508.745,
+      "eval_steps_per_second": 7.959,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3923368131237811,
+      "eval_loss": 0.5635139346122742,
+      "eval_runtime": 12.9481,
+      "eval_samples_per_second": 508.491,
+      "eval_steps_per_second": 7.955,
+      "step": 3420
+    },
+    {
+      "epoch": 0.39348399678788576,
+      "eval_loss": 0.5628678202629089,
+      "eval_runtime": 12.9965,
+      "eval_samples_per_second": 506.596,
+      "eval_steps_per_second": 7.925,
+      "step": 3430
+    },
+    {
+      "epoch": 0.39463118045199036,
+      "eval_loss": 0.5622407793998718,
+      "eval_runtime": 12.9926,
+      "eval_samples_per_second": 506.748,
+      "eval_steps_per_second": 7.928,
+      "step": 3440
+    },
+    {
+      "epoch": 0.39577836411609496,
+      "eval_loss": 0.5617032051086426,
+      "eval_runtime": 12.92,
+      "eval_samples_per_second": 509.598,
+      "eval_steps_per_second": 7.972,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3969255477801996,
+      "eval_loss": 0.5613731741905212,
+      "eval_runtime": 12.9152,
+      "eval_samples_per_second": 509.786,
+      "eval_steps_per_second": 7.975,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3980727314443042,
+      "eval_loss": 0.5606963038444519,
+      "eval_runtime": 12.9194,
+      "eval_samples_per_second": 509.62,
+      "eval_steps_per_second": 7.972,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3992199151084089,
+      "eval_loss": 0.5602635145187378,
+      "eval_runtime": 12.9242,
+      "eval_samples_per_second": 509.432,
+      "eval_steps_per_second": 7.97,
+      "step": 3480
+    },
+    {
+      "epoch": 0.4003670987725135,
+      "eval_loss": 0.5598347783088684,
+      "eval_runtime": 12.9579,
+      "eval_samples_per_second": 508.107,
+      "eval_steps_per_second": 7.949,
+      "step": 3490
+    },
+    {
+      "epoch": 0.4015142824366181,
+      "grad_norm": 3.217618465423584,
+      "learning_rate": 9.624389207563205e-06,
+      "loss": 0.938,
+      "step": 3500
+    },
+    {
+      "epoch": 0.4015142824366181,
+      "eval_loss": 0.5596053600311279,
+      "eval_runtime": 13.0466,
+      "eval_samples_per_second": 504.652,
+      "eval_steps_per_second": 7.895,
+      "step": 3500
+    },
+    {
+      "epoch": 0.40266146610072273,
+      "eval_loss": 0.5589408874511719,
+      "eval_runtime": 13.0715,
+      "eval_samples_per_second": 503.69,
+      "eval_steps_per_second": 7.88,
+      "step": 3510
+    },
+    {
+      "epoch": 0.40380864976482733,
+      "eval_loss": 0.558090329170227,
+      "eval_runtime": 12.9472,
+      "eval_samples_per_second": 508.528,
+      "eval_steps_per_second": 7.955,
+      "step": 3520
+    },
+    {
+      "epoch": 0.404955833428932,
+      "eval_loss": 0.5570879578590393,
+      "eval_runtime": 12.9169,
+      "eval_samples_per_second": 509.72,
+      "eval_steps_per_second": 7.974,
+      "step": 3530
+    },
+    {
+      "epoch": 0.4061030170930366,
+      "eval_loss": 0.5562835931777954,
+      "eval_runtime": 12.9892,
+      "eval_samples_per_second": 506.881,
+      "eval_steps_per_second": 7.93,
+      "step": 3540
+    },
+    {
+      "epoch": 0.40725020075714125,
+      "eval_loss": 0.5557334423065186,
+      "eval_runtime": 12.9142,
+      "eval_samples_per_second": 509.826,
+      "eval_steps_per_second": 7.976,
+      "step": 3550
+    },
+    {
+      "epoch": 0.40839738442124585,
+      "eval_loss": 0.5551453828811646,
+      "eval_runtime": 12.9967,
+      "eval_samples_per_second": 506.59,
+      "eval_steps_per_second": 7.925,
+      "step": 3560
+    },
+    {
+      "epoch": 0.40954456808535045,
+      "eval_loss": 0.5546151995658875,
+      "eval_runtime": 13.0283,
+      "eval_samples_per_second": 505.362,
+      "eval_steps_per_second": 7.906,
+      "step": 3570
+    },
+    {
+      "epoch": 0.4106917517494551,
+      "eval_loss": 0.5540978312492371,
+      "eval_runtime": 13.0306,
+      "eval_samples_per_second": 505.273,
+      "eval_steps_per_second": 7.904,
+      "step": 3580
+    },
+    {
+      "epoch": 0.4118389354135597,
+      "eval_loss": 0.5535085797309875,
+      "eval_runtime": 13.0453,
+      "eval_samples_per_second": 504.704,
+      "eval_steps_per_second": 7.896,
+      "step": 3590
+    },
+    {
+      "epoch": 0.41298611907766436,
+      "grad_norm": 3.073883056640625,
+      "learning_rate": 9.581899298916509e-06,
+      "loss": 0.955,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41298611907766436,
+      "eval_loss": 0.5527840256690979,
+      "eval_runtime": 13.0322,
+      "eval_samples_per_second": 505.21,
+      "eval_steps_per_second": 7.903,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41413330274176896,
+      "eval_loss": 0.5521849989891052,
+      "eval_runtime": 13.0324,
+      "eval_samples_per_second": 505.202,
+      "eval_steps_per_second": 7.903,
+      "step": 3610
+    },
+    {
+      "epoch": 0.41528048640587356,
+      "eval_loss": 0.5515934824943542,
+      "eval_runtime": 13.0428,
+      "eval_samples_per_second": 504.799,
+      "eval_steps_per_second": 7.897,
+      "step": 3620
+    },
+    {
+      "epoch": 0.4164276700699782,
+      "eval_loss": 0.5508650541305542,
+      "eval_runtime": 13.0451,
+      "eval_samples_per_second": 504.709,
+      "eval_steps_per_second": 7.896,
+      "step": 3630
+    },
+    {
+      "epoch": 0.4175748537340828,
+      "eval_loss": 0.5502671003341675,
+      "eval_runtime": 13.0251,
+      "eval_samples_per_second": 505.484,
+      "eval_steps_per_second": 7.908,
+      "step": 3640
+    },
+    {
+      "epoch": 0.4187220373981875,
+      "eval_loss": 0.5495434403419495,
+      "eval_runtime": 13.0242,
+      "eval_samples_per_second": 505.519,
+      "eval_steps_per_second": 7.908,
+      "step": 3650
+    },
+    {
+      "epoch": 0.4198692210622921,
+      "eval_loss": 0.5489790439605713,
+      "eval_runtime": 13.0389,
+      "eval_samples_per_second": 504.951,
+      "eval_steps_per_second": 7.899,
+      "step": 3660
+    },
+    {
+      "epoch": 0.4210164047263967,
+      "eval_loss": 0.5481202006340027,
+      "eval_runtime": 12.9193,
+      "eval_samples_per_second": 509.627,
+      "eval_steps_per_second": 7.973,
+      "step": 3670
+    },
+    {
+      "epoch": 0.42216358839050133,
+      "eval_loss": 0.5474987030029297,
+      "eval_runtime": 13.0455,
+      "eval_samples_per_second": 504.697,
+      "eval_steps_per_second": 7.895,
+      "step": 3680
+    },
+    {
+      "epoch": 0.42331077205460593,
+      "eval_loss": 0.5467217564582825,
+      "eval_runtime": 12.9516,
+      "eval_samples_per_second": 508.356,
+      "eval_steps_per_second": 7.953,
+      "step": 3690
+    },
+    {
+      "epoch": 0.4244579557187106,
+      "grad_norm": 4.54811954498291,
+      "learning_rate": 9.539409390269812e-06,
+      "loss": 0.9387,
+      "step": 3700
+    },
+    {
+      "epoch": 0.4244579557187106,
+      "eval_loss": 0.5463354587554932,
+      "eval_runtime": 12.949,
+      "eval_samples_per_second": 508.456,
+      "eval_steps_per_second": 7.954,
+      "step": 3700
+    },
+    {
+      "epoch": 0.4256051393828152,
+      "eval_loss": 0.5458938479423523,
+      "eval_runtime": 12.9107,
+      "eval_samples_per_second": 509.966,
+      "eval_steps_per_second": 7.978,
+      "step": 3710
+    },
+    {
+      "epoch": 0.4267523230469198,
+      "eval_loss": 0.5451832413673401,
+      "eval_runtime": 13.0127,
+      "eval_samples_per_second": 505.968,
+      "eval_steps_per_second": 7.915,
+      "step": 3720
+    },
+    {
+      "epoch": 0.42789950671102445,
+      "eval_loss": 0.5448161363601685,
+      "eval_runtime": 13.033,
+      "eval_samples_per_second": 505.18,
+      "eval_steps_per_second": 7.903,
+      "step": 3730
+    },
+    {
+      "epoch": 0.42904669037512905,
+      "eval_loss": 0.5443439483642578,
+      "eval_runtime": 13.0146,
+      "eval_samples_per_second": 505.895,
+      "eval_steps_per_second": 7.914,
+      "step": 3740
+    },
+    {
+      "epoch": 0.4301938740392337,
+      "eval_loss": 0.544032633304596,
+      "eval_runtime": 12.9589,
+      "eval_samples_per_second": 508.069,
+      "eval_steps_per_second": 7.948,
+      "step": 3750
+    },
+    {
+      "epoch": 0.4313410577033383,
+      "eval_loss": 0.5435395836830139,
+      "eval_runtime": 12.9098,
+      "eval_samples_per_second": 510.001,
+      "eval_steps_per_second": 7.978,
+      "step": 3760
+    },
+    {
+      "epoch": 0.4324882413674429,
+      "eval_loss": 0.5429771542549133,
+      "eval_runtime": 12.9366,
+      "eval_samples_per_second": 508.944,
+      "eval_steps_per_second": 7.962,
+      "step": 3770
+    },
+    {
+      "epoch": 0.43363542503154756,
+      "eval_loss": 0.5423203706741333,
+      "eval_runtime": 13.1085,
+      "eval_samples_per_second": 502.268,
+      "eval_steps_per_second": 7.857,
+      "step": 3780
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "eval_loss": 0.5417954921722412,
+      "eval_runtime": 13.0184,
+      "eval_samples_per_second": 505.747,
+      "eval_steps_per_second": 7.912,
+      "step": 3790
+    },
+    {
+      "epoch": 0.4359297923597568,
+      "grad_norm": 6.273746967315674,
+      "learning_rate": 9.496919481623116e-06,
+      "loss": 0.9672,
+      "step": 3800
+    },
+    {
+      "epoch": 0.4359297923597568,
+      "eval_loss": 0.5414963364601135,
+      "eval_runtime": 13.0358,
+      "eval_samples_per_second": 505.07,
+      "eval_steps_per_second": 7.901,
+      "step": 3800
+    },
+    {
+      "epoch": 0.4370769760238614,
+      "eval_loss": 0.541257381439209,
+      "eval_runtime": 12.9726,
+      "eval_samples_per_second": 507.53,
+      "eval_steps_per_second": 7.94,
+      "step": 3810
+    },
+    {
+      "epoch": 0.438224159687966,
+      "eval_loss": 0.5409945845603943,
+      "eval_runtime": 13.0274,
+      "eval_samples_per_second": 505.398,
+      "eval_steps_per_second": 7.906,
+      "step": 3820
+    },
+    {
+      "epoch": 0.4393713433520707,
+      "eval_loss": 0.5406415462493896,
+      "eval_runtime": 13.0472,
+      "eval_samples_per_second": 504.629,
+      "eval_steps_per_second": 7.894,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4405185270161753,
+      "eval_loss": 0.5403361916542053,
+      "eval_runtime": 12.9682,
+      "eval_samples_per_second": 507.704,
+      "eval_steps_per_second": 7.943,
+      "step": 3840
+    },
+    {
+      "epoch": 0.44166571068027993,
+      "eval_loss": 0.5397402048110962,
+      "eval_runtime": 13.0503,
+      "eval_samples_per_second": 504.509,
+      "eval_steps_per_second": 7.893,
+      "step": 3850
+    },
+    {
+      "epoch": 0.44281289434438453,
+      "eval_loss": 0.5394212007522583,
+      "eval_runtime": 12.9962,
+      "eval_samples_per_second": 506.611,
+      "eval_steps_per_second": 7.925,
+      "step": 3860
+    },
+    {
+      "epoch": 0.44396007800848913,
+      "eval_loss": 0.538590669631958,
+      "eval_runtime": 13.0433,
+      "eval_samples_per_second": 504.78,
+      "eval_steps_per_second": 7.897,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4451072616725938,
+      "eval_loss": 0.537807822227478,
+      "eval_runtime": 12.9414,
+      "eval_samples_per_second": 508.756,
+      "eval_steps_per_second": 7.959,
+      "step": 3880
+    },
+    {
+      "epoch": 0.4462544453366984,
+      "eval_loss": 0.536992073059082,
+      "eval_runtime": 12.9902,
+      "eval_samples_per_second": 506.844,
+      "eval_steps_per_second": 7.929,
+      "step": 3890
+    },
+    {
+      "epoch": 0.44740162900080305,
+      "grad_norm": 2.6543989181518555,
+      "learning_rate": 9.45442957297642e-06,
+      "loss": 0.926,
+      "step": 3900
+    },
+    {
+      "epoch": 0.44740162900080305,
+      "eval_loss": 0.5359557867050171,
+      "eval_runtime": 13.0495,
+      "eval_samples_per_second": 504.542,
+      "eval_steps_per_second": 7.893,
+      "step": 3900
+    },
+    {
+      "epoch": 0.44854881266490765,
+      "eval_loss": 0.5351366400718689,
+      "eval_runtime": 12.923,
+      "eval_samples_per_second": 509.48,
+      "eval_steps_per_second": 7.97,
+      "step": 3910
+    },
+    {
+      "epoch": 0.44969599632901225,
+      "eval_loss": 0.5345980525016785,
+      "eval_runtime": 12.9402,
+      "eval_samples_per_second": 508.803,
+      "eval_steps_per_second": 7.96,
+      "step": 3920
+    },
+    {
+      "epoch": 0.4508431799931169,
+      "eval_loss": 0.5343027710914612,
+      "eval_runtime": 13.0184,
+      "eval_samples_per_second": 505.747,
+      "eval_steps_per_second": 7.912,
+      "step": 3930
+    },
+    {
+      "epoch": 0.4519903636572215,
+      "eval_loss": 0.5339150428771973,
+      "eval_runtime": 13.0383,
+      "eval_samples_per_second": 504.974,
+      "eval_steps_per_second": 7.9,
+      "step": 3940
+    },
+    {
+      "epoch": 0.45313754732132616,
+      "eval_loss": 0.533658504486084,
+      "eval_runtime": 13.0155,
+      "eval_samples_per_second": 505.857,
+      "eval_steps_per_second": 7.914,
+      "step": 3950
+    },
+    {
+      "epoch": 0.45428473098543076,
+      "eval_loss": 0.5334100127220154,
+      "eval_runtime": 12.9238,
+      "eval_samples_per_second": 509.45,
+      "eval_steps_per_second": 7.97,
+      "step": 3960
+    },
+    {
+      "epoch": 0.45543191464953536,
+      "eval_loss": 0.5330451726913452,
+      "eval_runtime": 12.9194,
+      "eval_samples_per_second": 509.619,
+      "eval_steps_per_second": 7.972,
+      "step": 3970
+    },
+    {
+      "epoch": 0.45657909831364,
+      "eval_loss": 0.5327152609825134,
+      "eval_runtime": 13.015,
+      "eval_samples_per_second": 505.879,
+      "eval_steps_per_second": 7.914,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4577262819777446,
+      "eval_loss": 0.5323871374130249,
+      "eval_runtime": 13.021,
+      "eval_samples_per_second": 505.646,
+      "eval_steps_per_second": 7.91,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4588734656418493,
+      "grad_norm": 2.330146074295044,
+      "learning_rate": 9.411939664329721e-06,
+      "loss": 0.867,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4588734656418493,
+      "eval_loss": 0.5319190621376038,
+      "eval_runtime": 13.0222,
+      "eval_samples_per_second": 505.6,
+      "eval_steps_per_second": 7.91,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4600206493059539,
+      "eval_loss": 0.5313332676887512,
+      "eval_runtime": 13.0478,
+      "eval_samples_per_second": 504.605,
+      "eval_steps_per_second": 7.894,
+      "step": 4010
+    },
+    {
+      "epoch": 0.46116783297005853,
+      "eval_loss": 0.5308486223220825,
+      "eval_runtime": 12.964,
+      "eval_samples_per_second": 507.868,
+      "eval_steps_per_second": 7.945,
+      "step": 4020
+    },
+    {
+      "epoch": 0.46231501663416313,
+      "eval_loss": 0.5300149917602539,
+      "eval_runtime": 12.914,
+      "eval_samples_per_second": 509.835,
+      "eval_steps_per_second": 7.976,
+      "step": 4030
+    },
+    {
+      "epoch": 0.46346220029826773,
+      "eval_loss": 0.5292573571205139,
+      "eval_runtime": 12.9769,
+      "eval_samples_per_second": 507.361,
+      "eval_steps_per_second": 7.937,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4646093839623724,
+      "eval_loss": 0.5286610722541809,
+      "eval_runtime": 13.0368,
+      "eval_samples_per_second": 505.031,
+      "eval_steps_per_second": 7.901,
+      "step": 4050
+    },
+    {
+      "epoch": 0.465756567626477,
+      "eval_loss": 0.5283547043800354,
+      "eval_runtime": 13.0356,
+      "eval_samples_per_second": 505.077,
+      "eval_steps_per_second": 7.901,
+      "step": 4060
+    },
+    {
+      "epoch": 0.46690375129058165,
+      "eval_loss": 0.5280731916427612,
+      "eval_runtime": 13.0048,
+      "eval_samples_per_second": 506.275,
+      "eval_steps_per_second": 7.92,
+      "step": 4070
+    },
+    {
+      "epoch": 0.46805093495468625,
+      "eval_loss": 0.5276577472686768,
+      "eval_runtime": 12.9174,
+      "eval_samples_per_second": 509.699,
+      "eval_steps_per_second": 7.974,
+      "step": 4080
+    },
+    {
+      "epoch": 0.46919811861879085,
+      "eval_loss": 0.5272203087806702,
+      "eval_runtime": 12.9134,
+      "eval_samples_per_second": 509.857,
+      "eval_steps_per_second": 7.976,
+      "step": 4090
+    },
+    {
+      "epoch": 0.4703453022828955,
+      "grad_norm": 2.8037149906158447,
+      "learning_rate": 9.369449755683025e-06,
+      "loss": 0.916,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4703453022828955,
+      "eval_loss": 0.5267237424850464,
+      "eval_runtime": 12.9151,
+      "eval_samples_per_second": 509.79,
+      "eval_steps_per_second": 7.975,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4714924859470001,
+      "eval_loss": 0.5260052680969238,
+      "eval_runtime": 12.9148,
+      "eval_samples_per_second": 509.801,
+      "eval_steps_per_second": 7.975,
+      "step": 4110
+    },
+    {
+      "epoch": 0.47263966961110476,
+      "eval_loss": 0.5252425074577332,
+      "eval_runtime": 13.014,
+      "eval_samples_per_second": 505.915,
+      "eval_steps_per_second": 7.915,
+      "step": 4120
+    },
+    {
+      "epoch": 0.47378685327520936,
+      "eval_loss": 0.5246235132217407,
+      "eval_runtime": 13.0291,
+      "eval_samples_per_second": 505.332,
+      "eval_steps_per_second": 7.905,
+      "step": 4130
+    },
+    {
+      "epoch": 0.47493403693931396,
+      "eval_loss": 0.5238820314407349,
+      "eval_runtime": 12.9264,
+      "eval_samples_per_second": 509.345,
+      "eval_steps_per_second": 7.968,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4760812206034186,
+      "eval_loss": 0.5231978297233582,
+      "eval_runtime": 12.9421,
+      "eval_samples_per_second": 508.729,
+      "eval_steps_per_second": 7.959,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4772284042675232,
+      "eval_loss": 0.5225175023078918,
+      "eval_runtime": 13.0292,
+      "eval_samples_per_second": 505.328,
+      "eval_steps_per_second": 7.905,
+      "step": 4160
+    },
+    {
+      "epoch": 0.4783755879316279,
+      "eval_loss": 0.522089958190918,
+      "eval_runtime": 13.0462,
+      "eval_samples_per_second": 504.666,
+      "eval_steps_per_second": 7.895,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4795227715957325,
+      "eval_loss": 0.5216270685195923,
+      "eval_runtime": 13.0261,
+      "eval_samples_per_second": 505.448,
+      "eval_steps_per_second": 7.907,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4806699552598371,
+      "eval_loss": 0.5211179256439209,
+      "eval_runtime": 13.0229,
+      "eval_samples_per_second": 505.569,
+      "eval_steps_per_second": 7.909,
+      "step": 4190
+    },
+    {
+      "epoch": 0.48181713892394173,
+      "grad_norm": 4.981523036956787,
+      "learning_rate": 9.32695984703633e-06,
+      "loss": 0.9667,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48181713892394173,
+      "eval_loss": 0.5206459760665894,
+      "eval_runtime": 13.0189,
+      "eval_samples_per_second": 505.728,
+      "eval_steps_per_second": 7.912,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48296432258804634,
+      "eval_loss": 0.5203832983970642,
+      "eval_runtime": 13.0372,
+      "eval_samples_per_second": 505.016,
+      "eval_steps_per_second": 7.9,
+      "step": 4210
+    },
+    {
+      "epoch": 0.484111506252151,
+      "eval_loss": 0.5200340151786804,
+      "eval_runtime": 12.921,
+      "eval_samples_per_second": 509.558,
+      "eval_steps_per_second": 7.972,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4852586899162556,
+      "eval_loss": 0.5191519856452942,
+      "eval_runtime": 12.9412,
+      "eval_samples_per_second": 508.761,
+      "eval_steps_per_second": 7.959,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4864058735803602,
+      "eval_loss": 0.5187414884567261,
+      "eval_runtime": 13.0244,
+      "eval_samples_per_second": 505.513,
+      "eval_steps_per_second": 7.908,
+      "step": 4240
+    },
+    {
+      "epoch": 0.48755305724446485,
+      "eval_loss": 0.5184528827667236,
+      "eval_runtime": 13.0462,
+      "eval_samples_per_second": 504.669,
+      "eval_steps_per_second": 7.895,
+      "step": 4250
+    },
+    {
+      "epoch": 0.48870024090856945,
+      "eval_loss": 0.5179461240768433,
+      "eval_runtime": 12.9404,
+      "eval_samples_per_second": 508.793,
+      "eval_steps_per_second": 7.96,
+      "step": 4260
+    },
+    {
+      "epoch": 0.4898474245726741,
+      "eval_loss": 0.5173225402832031,
+      "eval_runtime": 12.9886,
+      "eval_samples_per_second": 506.904,
+      "eval_steps_per_second": 7.93,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4909946082367787,
+      "eval_loss": 0.5169667601585388,
+      "eval_runtime": 12.918,
+      "eval_samples_per_second": 509.678,
+      "eval_steps_per_second": 7.973,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4921417919008833,
+      "eval_loss": 0.5165331363677979,
+      "eval_runtime": 12.9162,
+      "eval_samples_per_second": 509.747,
+      "eval_steps_per_second": 7.974,
+      "step": 4290
+    },
+    {
+      "epoch": 0.49328897556498796,
+      "grad_norm": 2.352576732635498,
+      "learning_rate": 9.284469938389632e-06,
+      "loss": 0.9276,
+      "step": 4300
+    },
+    {
+      "epoch": 0.49328897556498796,
+      "eval_loss": 0.5160226821899414,
+      "eval_runtime": 12.9928,
+      "eval_samples_per_second": 506.741,
+      "eval_steps_per_second": 7.927,
+      "step": 4300
+    },
+    {
+      "epoch": 0.49443615922909256,
+      "eval_loss": 0.5154353976249695,
+      "eval_runtime": 13.0317,
+      "eval_samples_per_second": 505.228,
+      "eval_steps_per_second": 7.904,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4955833428931972,
+      "eval_loss": 0.5149873495101929,
+      "eval_runtime": 12.9748,
+      "eval_samples_per_second": 507.445,
+      "eval_steps_per_second": 7.938,
+      "step": 4320
+    },
+    {
+      "epoch": 0.4967305265573018,
+      "eval_loss": 0.514439582824707,
+      "eval_runtime": 12.9649,
+      "eval_samples_per_second": 507.833,
+      "eval_steps_per_second": 7.945,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4978777102214064,
+      "eval_loss": 0.5140998363494873,
+      "eval_runtime": 13.0221,
+      "eval_samples_per_second": 505.603,
+      "eval_steps_per_second": 7.91,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4990248938855111,
+      "eval_loss": 0.5139394402503967,
+      "eval_runtime": 13.0098,
+      "eval_samples_per_second": 506.081,
+      "eval_steps_per_second": 7.917,
+      "step": 4350
+    },
+    {
+      "epoch": 0.5001720775496157,
+      "eval_loss": 0.5137957334518433,
+      "eval_runtime": 13.0381,
+      "eval_samples_per_second": 504.983,
+      "eval_steps_per_second": 7.9,
+      "step": 4360
+    },
+    {
+      "epoch": 0.5013192612137203,
+      "eval_loss": 0.5136282444000244,
+      "eval_runtime": 13.0808,
+      "eval_samples_per_second": 503.332,
+      "eval_steps_per_second": 7.874,
+      "step": 4370
+    },
+    {
+      "epoch": 0.502466444877825,
+      "eval_loss": 0.5132753849029541,
+      "eval_runtime": 13.0402,
+      "eval_samples_per_second": 504.902,
+      "eval_steps_per_second": 7.899,
+      "step": 4380
+    },
+    {
+      "epoch": 0.5036136285419296,
+      "eval_loss": 0.5129442811012268,
+      "eval_runtime": 12.9179,
+      "eval_samples_per_second": 509.682,
+      "eval_steps_per_second": 7.973,
+      "step": 4390
+    },
+    {
+      "epoch": 0.5047608122060342,
+      "grad_norm": 2.744755983352661,
+      "learning_rate": 9.241980029742936e-06,
+      "loss": 0.9331,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5047608122060342,
+      "eval_loss": 0.5125741362571716,
+      "eval_runtime": 12.9316,
+      "eval_samples_per_second": 509.142,
+      "eval_steps_per_second": 7.965,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5059079958701388,
+      "eval_loss": 0.5123194456100464,
+      "eval_runtime": 13.0104,
+      "eval_samples_per_second": 506.056,
+      "eval_steps_per_second": 7.917,
+      "step": 4410
+    },
+    {
+      "epoch": 0.5070551795342434,
+      "eval_loss": 0.5116878747940063,
+      "eval_runtime": 13.0458,
+      "eval_samples_per_second": 504.684,
+      "eval_steps_per_second": 7.895,
+      "step": 4420
+    },
+    {
+      "epoch": 0.5082023631983481,
+      "eval_loss": 0.5112791061401367,
+      "eval_runtime": 13.0399,
+      "eval_samples_per_second": 504.911,
+      "eval_steps_per_second": 7.899,
+      "step": 4430
+    },
+    {
+      "epoch": 0.5093495468624527,
+      "eval_loss": 0.5108391642570496,
+      "eval_runtime": 13.0444,
+      "eval_samples_per_second": 504.738,
+      "eval_steps_per_second": 7.896,
+      "step": 4440
+    },
+    {
+      "epoch": 0.5104967305265573,
+      "eval_loss": 0.5106413960456848,
+      "eval_runtime": 13.0495,
+      "eval_samples_per_second": 504.542,
+      "eval_steps_per_second": 7.893,
+      "step": 4450
+    },
+    {
+      "epoch": 0.5116439141906619,
+      "eval_loss": 0.5105773210525513,
+      "eval_runtime": 13.026,
+      "eval_samples_per_second": 505.453,
+      "eval_steps_per_second": 7.907,
+      "step": 4460
+    },
+    {
+      "epoch": 0.5127910978547665,
+      "eval_loss": 0.5106124877929688,
+      "eval_runtime": 13.0887,
+      "eval_samples_per_second": 503.028,
+      "eval_steps_per_second": 7.869,
+      "step": 4470
+    },
+    {
+      "epoch": 0.5139382815188712,
+      "eval_loss": 0.5104395151138306,
+      "eval_runtime": 13.0488,
+      "eval_samples_per_second": 504.567,
+      "eval_steps_per_second": 7.893,
+      "step": 4480
+    },
+    {
+      "epoch": 0.5150854651829758,
+      "eval_loss": 0.510217547416687,
+      "eval_runtime": 13.0309,
+      "eval_samples_per_second": 505.261,
+      "eval_steps_per_second": 7.904,
+      "step": 4490
+    },
+    {
+      "epoch": 0.5162326488470804,
+      "grad_norm": 3.2935314178466797,
+      "learning_rate": 9.199490121096241e-06,
+      "loss": 0.907,
+      "step": 4500
+    },
+    {
+      "epoch": 0.5162326488470804,
+      "eval_loss": 0.5096931457519531,
+      "eval_runtime": 13.0419,
+      "eval_samples_per_second": 504.834,
+      "eval_steps_per_second": 7.898,
+      "step": 4500
+    },
+    {
+      "epoch": 0.517379832511185,
+      "eval_loss": 0.5092429518699646,
+      "eval_runtime": 13.0377,
+      "eval_samples_per_second": 504.999,
+      "eval_steps_per_second": 7.9,
+      "step": 4510
+    },
+    {
+      "epoch": 0.5185270161752896,
+      "eval_loss": 0.5086391568183899,
+      "eval_runtime": 12.9418,
+      "eval_samples_per_second": 508.739,
+      "eval_steps_per_second": 7.959,
+      "step": 4520
+    },
+    {
+      "epoch": 0.5196741998393943,
+      "eval_loss": 0.508176326751709,
+      "eval_runtime": 13.154,
+      "eval_samples_per_second": 500.533,
+      "eval_steps_per_second": 7.83,
+      "step": 4530
+    },
+    {
+      "epoch": 0.5208213835034989,
+      "eval_loss": 0.5079353451728821,
+      "eval_runtime": 13.0269,
+      "eval_samples_per_second": 505.417,
+      "eval_steps_per_second": 7.907,
+      "step": 4540
+    },
+    {
+      "epoch": 0.5219685671676035,
+      "eval_loss": 0.5075249671936035,
+      "eval_runtime": 13.0283,
+      "eval_samples_per_second": 505.36,
+      "eval_steps_per_second": 7.906,
+      "step": 4550
+    },
+    {
+      "epoch": 0.5231157508317081,
+      "eval_loss": 0.5070711970329285,
+      "eval_runtime": 13.0867,
+      "eval_samples_per_second": 503.106,
+      "eval_steps_per_second": 7.871,
+      "step": 4560
+    },
+    {
+      "epoch": 0.5242629344958127,
+      "eval_loss": 0.5066924095153809,
+      "eval_runtime": 13.0158,
+      "eval_samples_per_second": 505.847,
+      "eval_steps_per_second": 7.913,
+      "step": 4570
+    },
+    {
+      "epoch": 0.5254101181599174,
+      "eval_loss": 0.5066002607345581,
+      "eval_runtime": 13.0247,
+      "eval_samples_per_second": 505.499,
+      "eval_steps_per_second": 7.908,
+      "step": 4580
+    },
+    {
+      "epoch": 0.526557301824022,
+      "eval_loss": 0.5061988830566406,
+      "eval_runtime": 12.9452,
+      "eval_samples_per_second": 508.604,
+      "eval_steps_per_second": 7.957,
+      "step": 4590
+    },
+    {
+      "epoch": 0.5277044854881267,
+      "grad_norm": 2.8888604640960693,
+      "learning_rate": 9.157000212449543e-06,
+      "loss": 0.913,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5277044854881267,
+      "eval_loss": 0.5059130787849426,
+      "eval_runtime": 13.0522,
+      "eval_samples_per_second": 504.435,
+      "eval_steps_per_second": 7.891,
+      "step": 4600
+    },
+    {
+      "epoch": 0.5288516691522313,
+      "eval_loss": 0.5056325793266296,
+      "eval_runtime": 13.0474,
+      "eval_samples_per_second": 504.62,
+      "eval_steps_per_second": 7.894,
+      "step": 4610
+    },
+    {
+      "epoch": 0.5299988528163359,
+      "eval_loss": 0.5051873922348022,
+      "eval_runtime": 13.0297,
+      "eval_samples_per_second": 505.306,
+      "eval_steps_per_second": 7.905,
+      "step": 4620
+    },
+    {
+      "epoch": 0.5311460364804406,
+      "eval_loss": 0.5046107769012451,
+      "eval_runtime": 13.0544,
+      "eval_samples_per_second": 504.351,
+      "eval_steps_per_second": 7.89,
+      "step": 4630
+    },
+    {
+      "epoch": 0.5322932201445452,
+      "eval_loss": 0.5039039254188538,
+      "eval_runtime": 13.0434,
+      "eval_samples_per_second": 504.778,
+      "eval_steps_per_second": 7.897,
+      "step": 4640
+    },
+    {
+      "epoch": 0.5334404038086498,
+      "eval_loss": 0.5032686591148376,
+      "eval_runtime": 13.0225,
+      "eval_samples_per_second": 505.586,
+      "eval_steps_per_second": 7.909,
+      "step": 4650
+    },
+    {
+      "epoch": 0.5345875874727544,
+      "eval_loss": 0.5030205249786377,
+      "eval_runtime": 13.044,
+      "eval_samples_per_second": 504.753,
+      "eval_steps_per_second": 7.896,
+      "step": 4660
+    },
+    {
+      "epoch": 0.535734771136859,
+      "eval_loss": 0.5027660727500916,
+      "eval_runtime": 13.0891,
+      "eval_samples_per_second": 503.013,
+      "eval_steps_per_second": 7.869,
+      "step": 4670
+    },
+    {
+      "epoch": 0.5368819548009637,
+      "eval_loss": 0.5026541352272034,
+      "eval_runtime": 13.0208,
+      "eval_samples_per_second": 505.653,
+      "eval_steps_per_second": 7.91,
+      "step": 4680
+    },
+    {
+      "epoch": 0.5380291384650683,
+      "eval_loss": 0.5023172497749329,
+      "eval_runtime": 13.0369,
+      "eval_samples_per_second": 505.028,
+      "eval_steps_per_second": 7.901,
+      "step": 4690
+    },
+    {
+      "epoch": 0.5391763221291729,
+      "grad_norm": 2.457648992538452,
+      "learning_rate": 9.114510303802847e-06,
+      "loss": 0.9047,
+      "step": 4700
+    },
+    {
+      "epoch": 0.5391763221291729,
+      "eval_loss": 0.5020408034324646,
+      "eval_runtime": 13.0428,
+      "eval_samples_per_second": 504.8,
+      "eval_steps_per_second": 7.897,
+      "step": 4700
+    },
+    {
+      "epoch": 0.5403235057932775,
+      "eval_loss": 0.5017600059509277,
+      "eval_runtime": 13.0177,
+      "eval_samples_per_second": 505.772,
+      "eval_steps_per_second": 7.912,
+      "step": 4710
+    },
+    {
+      "epoch": 0.5414706894573821,
+      "eval_loss": 0.5015026926994324,
+      "eval_runtime": 13.0186,
+      "eval_samples_per_second": 505.738,
+      "eval_steps_per_second": 7.912,
+      "step": 4720
+    },
+    {
+      "epoch": 0.5426178731214868,
+      "eval_loss": 0.5008870959281921,
+      "eval_runtime": 12.9903,
+      "eval_samples_per_second": 506.841,
+      "eval_steps_per_second": 7.929,
+      "step": 4730
+    },
+    {
+      "epoch": 0.5437650567855914,
+      "eval_loss": 0.500299334526062,
+      "eval_runtime": 13.032,
+      "eval_samples_per_second": 505.218,
+      "eval_steps_per_second": 7.904,
+      "step": 4740
+    },
+    {
+      "epoch": 0.544912240449696,
+      "eval_loss": 0.49969592690467834,
+      "eval_runtime": 13.0199,
+      "eval_samples_per_second": 505.686,
+      "eval_steps_per_second": 7.911,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5460594241138006,
+      "eval_loss": 0.4990648329257965,
+      "eval_runtime": 13.0229,
+      "eval_samples_per_second": 505.573,
+      "eval_steps_per_second": 7.909,
+      "step": 4760
+    },
+    {
+      "epoch": 0.5472066077779052,
+      "eval_loss": 0.4983595013618469,
+      "eval_runtime": 13.0558,
+      "eval_samples_per_second": 504.298,
+      "eval_steps_per_second": 7.889,
+      "step": 4770
+    },
+    {
+      "epoch": 0.5483537914420099,
+      "eval_loss": 0.4980465769767761,
+      "eval_runtime": 13.0075,
+      "eval_samples_per_second": 506.171,
+      "eval_steps_per_second": 7.919,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5495009751061145,
+      "eval_loss": 0.4980180859565735,
+      "eval_runtime": 13.0189,
+      "eval_samples_per_second": 505.727,
+      "eval_steps_per_second": 7.912,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5506481587702191,
+      "grad_norm": 2.9509494304656982,
+      "learning_rate": 9.07202039515615e-06,
+      "loss": 0.887,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5506481587702191,
+      "eval_loss": 0.4978945553302765,
+      "eval_runtime": 13.073,
+      "eval_samples_per_second": 503.632,
+      "eval_steps_per_second": 7.879,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5517953424343237,
+      "eval_loss": 0.49748191237449646,
+      "eval_runtime": 13.0456,
+      "eval_samples_per_second": 504.69,
+      "eval_steps_per_second": 7.895,
+      "step": 4810
+    },
+    {
+      "epoch": 0.5529425260984283,
+      "eval_loss": 0.49734070897102356,
+      "eval_runtime": 12.917,
+      "eval_samples_per_second": 509.716,
+      "eval_steps_per_second": 7.974,
+      "step": 4820
+    },
+    {
+      "epoch": 0.554089709762533,
+      "eval_loss": 0.4968598783016205,
+      "eval_runtime": 12.9242,
+      "eval_samples_per_second": 509.434,
+      "eval_steps_per_second": 7.97,
+      "step": 4830
+    },
+    {
+      "epoch": 0.5552368934266376,
+      "eval_loss": 0.49663135409355164,
+      "eval_runtime": 12.917,
+      "eval_samples_per_second": 509.714,
+      "eval_steps_per_second": 7.974,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5563840770907422,
+      "eval_loss": 0.4964074492454529,
+      "eval_runtime": 12.9307,
+      "eval_samples_per_second": 509.176,
+      "eval_steps_per_second": 7.966,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5575312607548468,
+      "eval_loss": 0.49635863304138184,
+      "eval_runtime": 13.0256,
+      "eval_samples_per_second": 505.465,
+      "eval_steps_per_second": 7.907,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5586784444189514,
+      "eval_loss": 0.4959893226623535,
+      "eval_runtime": 13.019,
+      "eval_samples_per_second": 505.724,
+      "eval_steps_per_second": 7.912,
+      "step": 4870
+    },
+    {
+      "epoch": 0.5598256280830561,
+      "eval_loss": 0.495743066072464,
+      "eval_runtime": 12.9529,
+      "eval_samples_per_second": 508.302,
+      "eval_steps_per_second": 7.952,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5609728117471607,
+      "eval_loss": 0.49548840522766113,
+      "eval_runtime": 12.9165,
+      "eval_samples_per_second": 509.734,
+      "eval_steps_per_second": 7.974,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5621199954112653,
+      "grad_norm": 2.523698329925537,
+      "learning_rate": 9.029530486509454e-06,
+      "loss": 0.8645,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5621199954112653,
+      "eval_loss": 0.4952070116996765,
+      "eval_runtime": 12.9252,
+      "eval_samples_per_second": 509.394,
+      "eval_steps_per_second": 7.969,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5632671790753699,
+      "eval_loss": 0.49501940608024597,
+      "eval_runtime": 12.9174,
+      "eval_samples_per_second": 509.699,
+      "eval_steps_per_second": 7.974,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5644143627394745,
+      "eval_loss": 0.49515894055366516,
+      "eval_runtime": 12.9248,
+      "eval_samples_per_second": 509.409,
+      "eval_steps_per_second": 7.969,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5655615464035793,
+      "eval_loss": 0.4948519468307495,
+      "eval_runtime": 13.0296,
+      "eval_samples_per_second": 505.311,
+      "eval_steps_per_second": 7.905,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5667087300676839,
+      "eval_loss": 0.4942854344844818,
+      "eval_runtime": 13.0211,
+      "eval_samples_per_second": 505.639,
+      "eval_steps_per_second": 7.91,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5678559137317885,
+      "eval_loss": 0.49377119541168213,
+      "eval_runtime": 13.0383,
+      "eval_samples_per_second": 504.973,
+      "eval_steps_per_second": 7.9,
+      "step": 4950
+    },
+    {
+      "epoch": 0.569003097395893,
+      "eval_loss": 0.4936191439628601,
+      "eval_runtime": 13.02,
+      "eval_samples_per_second": 505.683,
+      "eval_steps_per_second": 7.911,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5701502810599977,
+      "eval_loss": 0.49330177903175354,
+      "eval_runtime": 13.023,
+      "eval_samples_per_second": 505.566,
+      "eval_steps_per_second": 7.909,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5712974647241024,
+      "eval_loss": 0.4930713176727295,
+      "eval_runtime": 13.0236,
+      "eval_samples_per_second": 505.545,
+      "eval_steps_per_second": 7.909,
+      "step": 4980
+    },
+    {
+      "epoch": 0.572444648388207,
+      "eval_loss": 0.49287787079811096,
+      "eval_runtime": 12.9359,
+      "eval_samples_per_second": 508.972,
+      "eval_steps_per_second": 7.962,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5735918320523116,
+      "grad_norm": 8.273838996887207,
+      "learning_rate": 8.987040577862758e-06,
+      "loss": 0.8348,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5735918320523116,
+      "eval_loss": 0.49241214990615845,
+      "eval_runtime": 12.9315,
+      "eval_samples_per_second": 509.146,
+      "eval_steps_per_second": 7.965,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 100,
+  "max_steps": 26151,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}