diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/last-checkpoint/trainer_state.json"
@@ -0,0 +1,7120 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.14742739200943536,
+  "eval_steps": 500,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 2.948547840188707e-05,
+      "grad_norm": 4288.0,
+      "learning_rate": 5.89622641509434e-10,
+      "loss": 5.8964,
+      "step": 1
+    },
+    {
+      "epoch": 0.00014742739200943535,
+      "grad_norm": 5152.0,
+      "learning_rate": 2.9481132075471697e-09,
+      "loss": 7.2418,
+      "step": 5
+    },
+    {
+      "epoch": 0.0002948547840188707,
+      "grad_norm": 3872.0,
+      "learning_rate": 5.896226415094339e-09,
+      "loss": 6.7926,
+      "step": 10
+    },
+    {
+      "epoch": 0.0004422821760283061,
+      "grad_norm": 2704.0,
+      "learning_rate": 8.844339622641509e-09,
+      "loss": 7.0144,
+      "step": 15
+    },
+    {
+      "epoch": 0.0005897095680377414,
+      "grad_norm": 5824.0,
+      "learning_rate": 1.1792452830188679e-08,
+      "loss": 7.431,
+      "step": 20
+    },
+    {
+      "epoch": 0.0007371369600471767,
+      "grad_norm": 3488.0,
+      "learning_rate": 1.4740566037735849e-08,
+      "loss": 7.3504,
+      "step": 25
+    },
+    {
+      "epoch": 0.0008845643520566122,
+      "grad_norm": 4448.0,
+      "learning_rate": 1.7688679245283017e-08,
+      "loss": 6.8362,
+      "step": 30
+    },
+    {
+      "epoch": 0.0010319917440660474,
+      "grad_norm": 3984.0,
+      "learning_rate": 2.0636792452830187e-08,
+      "loss": 7.2579,
+      "step": 35
+    },
+    {
+      "epoch": 0.0011794191360754828,
+      "grad_norm": 6112.0,
+      "learning_rate": 2.3584905660377358e-08,
+      "loss": 7.0144,
+      "step": 40
+    },
+    {
+      "epoch": 0.0013268465280849183,
+      "grad_norm": 3536.0,
+      "learning_rate": 2.6533018867924528e-08,
+      "loss": 6.6585,
+      "step": 45
+    },
+    {
+      "epoch": 0.0014742739200943535,
+      "grad_norm": 5184.0,
+      "learning_rate": 2.9481132075471698e-08,
+      "loss": 7.1249,
+      "step": 50
+    },
+    {
+      "epoch": 0.001621701312103789,
+      "grad_norm": 4736.0,
+      "learning_rate": 3.242924528301887e-08,
+      "loss": 7.0672,
+      "step": 55
+    },
+    {
+      "epoch": 0.0017691287041132243,
+      "grad_norm": 3888.0,
+      "learning_rate": 3.5377358490566035e-08,
+      "loss": 6.9418,
+      "step": 60
+    },
+    {
+      "epoch": 0.0019165560961226596,
+      "grad_norm": 4512.0,
+      "learning_rate": 3.83254716981132e-08,
+      "loss": 6.7915,
+      "step": 65
+    },
+    {
+      "epoch": 0.0020639834881320948,
+      "grad_norm": 4544.0,
+      "learning_rate": 4.1273584905660375e-08,
+      "loss": 6.9867,
+      "step": 70
+    },
+    {
+      "epoch": 0.0022114108801415304,
+      "grad_norm": 4160.0,
+      "learning_rate": 4.422169811320754e-08,
+      "loss": 6.9295,
+      "step": 75
+    },
+    {
+      "epoch": 0.0023588382721509656,
+      "grad_norm": 5024.0,
+      "learning_rate": 4.7169811320754715e-08,
+      "loss": 7.1398,
+      "step": 80
+    },
+    {
+      "epoch": 0.002506265664160401,
+      "grad_norm": 6688.0,
+      "learning_rate": 5.011792452830189e-08,
+      "loss": 6.8533,
+      "step": 85
+    },
+    {
+      "epoch": 0.0026536930561698365,
+      "grad_norm": 4128.0,
+      "learning_rate": 5.3066037735849055e-08,
+      "loss": 6.8373,
+      "step": 90
+    },
+    {
+      "epoch": 0.0028011204481792717,
+      "grad_norm": 4512.0,
+      "learning_rate": 5.601415094339622e-08,
+      "loss": 6.9857,
+      "step": 95
+    },
+    {
+      "epoch": 0.002948547840188707,
+      "grad_norm": 4704.0,
+      "learning_rate": 5.8962264150943396e-08,
+      "loss": 7.3043,
+      "step": 100
+    },
+    {
+      "epoch": 0.0030959752321981426,
+      "grad_norm": 5824.0,
+      "learning_rate": 6.191037735849057e-08,
+      "loss": 7.4446,
+      "step": 105
+    },
+    {
+      "epoch": 0.003243402624207578,
+      "grad_norm": 5600.0,
+      "learning_rate": 6.485849056603774e-08,
+      "loss": 6.4237,
+      "step": 110
+    },
+    {
+      "epoch": 0.003390830016217013,
+      "grad_norm": 3360.0,
+      "learning_rate": 6.78066037735849e-08,
+      "loss": 7.1134,
+      "step": 115
+    },
+    {
+      "epoch": 0.0035382574082264487,
+      "grad_norm": 5120.0,
+      "learning_rate": 7.075471698113207e-08,
+      "loss": 6.541,
+      "step": 120
+    },
+    {
+      "epoch": 0.003685684800235884,
+      "grad_norm": 4016.0,
+      "learning_rate": 7.370283018867925e-08,
+      "loss": 7.1745,
+      "step": 125
+    },
+    {
+      "epoch": 0.003833112192245319,
+      "grad_norm": 4096.0,
+      "learning_rate": 7.66509433962264e-08,
+      "loss": 6.6107,
+      "step": 130
+    },
+    {
+      "epoch": 0.003980539584254755,
+      "grad_norm": 6784.0,
+      "learning_rate": 7.959905660377358e-08,
+      "loss": 7.1709,
+      "step": 135
+    },
+    {
+      "epoch": 0.0041279669762641896,
+      "grad_norm": 4800.0,
+      "learning_rate": 8.254716981132075e-08,
+      "loss": 6.7303,
+      "step": 140
+    },
+    {
+      "epoch": 0.004275394368273625,
+      "grad_norm": 5472.0,
+      "learning_rate": 8.549528301886792e-08,
+      "loss": 6.6023,
+      "step": 145
+    },
+    {
+      "epoch": 0.004422821760283061,
+      "grad_norm": 5952.0,
+      "learning_rate": 8.844339622641508e-08,
+      "loss": 7.3574,
+      "step": 150
+    },
+    {
+      "epoch": 0.004570249152292496,
+      "grad_norm": 4480.0,
+      "learning_rate": 9.139150943396226e-08,
+      "loss": 6.4739,
+      "step": 155
+    },
+    {
+      "epoch": 0.004717676544301931,
+      "grad_norm": 2928.0,
+      "learning_rate": 9.433962264150943e-08,
+      "loss": 6.5303,
+      "step": 160
+    },
+    {
+      "epoch": 0.004865103936311367,
+      "grad_norm": 4064.0,
+      "learning_rate": 9.72877358490566e-08,
+      "loss": 6.7455,
+      "step": 165
+    },
+    {
+      "epoch": 0.005012531328320802,
+      "grad_norm": 2928.0,
+      "learning_rate": 1.0023584905660378e-07,
+      "loss": 6.2466,
+      "step": 170
+    },
+    {
+      "epoch": 0.005159958720330237,
+      "grad_norm": 3536.0,
+      "learning_rate": 1.0318396226415093e-07,
+      "loss": 6.493,
+      "step": 175
+    },
+    {
+      "epoch": 0.005307386112339673,
+      "grad_norm": 3792.0,
+      "learning_rate": 1.0613207547169811e-07,
+      "loss": 6.1627,
+      "step": 180
+    },
+    {
+      "epoch": 0.005454813504349108,
+      "grad_norm": 5568.0,
+      "learning_rate": 1.0908018867924528e-07,
+      "loss": 6.7054,
+      "step": 185
+    },
+    {
+      "epoch": 0.0056022408963585435,
+      "grad_norm": 5312.0,
+      "learning_rate": 1.1202830188679244e-07,
+      "loss": 6.3761,
+      "step": 190
+    },
+    {
+      "epoch": 0.005749668288367979,
+      "grad_norm": 3888.0,
+      "learning_rate": 1.1497641509433961e-07,
+      "loss": 6.4957,
+      "step": 195
+    },
+    {
+      "epoch": 0.005897095680377414,
+      "grad_norm": 4016.0,
+      "learning_rate": 1.1792452830188679e-07,
+      "loss": 6.3387,
+      "step": 200
+    },
+    {
+      "epoch": 0.0060445230723868495,
+      "grad_norm": 3712.0,
+      "learning_rate": 1.2087264150943396e-07,
+      "loss": 6.8239,
+      "step": 205
+    },
+    {
+      "epoch": 0.006191950464396285,
+      "grad_norm": 2224.0,
+      "learning_rate": 1.2382075471698114e-07,
+      "loss": 6.4956,
+      "step": 210
+    },
+    {
+      "epoch": 0.00633937785640572,
+      "grad_norm": 4736.0,
+      "learning_rate": 1.267688679245283e-07,
+      "loss": 6.2678,
+      "step": 215
+    },
+    {
+      "epoch": 0.006486805248415156,
+      "grad_norm": 4320.0,
+      "learning_rate": 1.2971698113207547e-07,
+      "loss": 6.4021,
+      "step": 220
+    },
+    {
+      "epoch": 0.006634232640424591,
+      "grad_norm": 4032.0,
+      "learning_rate": 1.3266509433962265e-07,
+      "loss": 5.8306,
+      "step": 225
+    },
+    {
+      "epoch": 0.006781660032434026,
+      "grad_norm": 3664.0,
+      "learning_rate": 1.356132075471698e-07,
+      "loss": 6.7351,
+      "step": 230
+    },
+    {
+      "epoch": 0.006929087424443462,
+      "grad_norm": 2032.0,
+      "learning_rate": 1.3856132075471696e-07,
+      "loss": 6.3457,
+      "step": 235
+    },
+    {
+      "epoch": 0.007076514816452897,
+      "grad_norm": 2784.0,
+      "learning_rate": 1.4150943396226414e-07,
+      "loss": 6.1769,
+      "step": 240
+    },
+    {
+      "epoch": 0.007223942208462332,
+      "grad_norm": 1464.0,
+      "learning_rate": 1.4445754716981132e-07,
+      "loss": 5.8652,
+      "step": 245
+    },
+    {
+      "epoch": 0.007371369600471768,
+      "grad_norm": 2008.0,
+      "learning_rate": 1.474056603773585e-07,
+      "loss": 6.0493,
+      "step": 250
+    },
+    {
+      "epoch": 0.007518796992481203,
+      "grad_norm": 1040.0,
+      "learning_rate": 1.5035377358490565e-07,
+      "loss": 5.9201,
+      "step": 255
+    },
+    {
+      "epoch": 0.007666224384490638,
+      "grad_norm": 1816.0,
+      "learning_rate": 1.533018867924528e-07,
+      "loss": 6.2471,
+      "step": 260
+    },
+    {
+      "epoch": 0.007813651776500074,
+      "grad_norm": 1432.0,
+      "learning_rate": 1.5624999999999999e-07,
+      "loss": 5.8522,
+      "step": 265
+    },
+    {
+      "epoch": 0.00796107916850951,
+      "grad_norm": 1816.0,
+      "learning_rate": 1.5919811320754717e-07,
+      "loss": 5.9513,
+      "step": 270
+    },
+    {
+      "epoch": 0.008108506560518945,
+      "grad_norm": 520.0,
+      "learning_rate": 1.6214622641509435e-07,
+      "loss": 5.784,
+      "step": 275
+    },
+    {
+      "epoch": 0.008255933952528379,
+      "grad_norm": 1272.0,
+      "learning_rate": 1.650943396226415e-07,
+      "loss": 5.6907,
+      "step": 280
+    },
+    {
+      "epoch": 0.008403361344537815,
+      "grad_norm": 852.0,
+      "learning_rate": 1.6804245283018868e-07,
+      "loss": 5.6086,
+      "step": 285
+    },
+    {
+      "epoch": 0.00855078873654725,
+      "grad_norm": 266.0,
+      "learning_rate": 1.7099056603773583e-07,
+      "loss": 5.3121,
+      "step": 290
+    },
+    {
+      "epoch": 0.008698216128556686,
+      "grad_norm": 756.0,
+      "learning_rate": 1.7393867924528301e-07,
+      "loss": 5.6198,
+      "step": 295
+    },
+    {
+      "epoch": 0.008845643520566122,
+      "grad_norm": 568.0,
+      "learning_rate": 1.7688679245283017e-07,
+      "loss": 5.5308,
+      "step": 300
+    },
+    {
+      "epoch": 0.008993070912575557,
+      "grad_norm": 644.0,
+      "learning_rate": 1.7983490566037735e-07,
+      "loss": 5.522,
+      "step": 305
+    },
+    {
+      "epoch": 0.009140498304584991,
+      "grad_norm": 596.0,
+      "learning_rate": 1.8278301886792453e-07,
+      "loss": 5.6054,
+      "step": 310
+    },
+    {
+      "epoch": 0.009287925696594427,
+      "grad_norm": 384.0,
+      "learning_rate": 1.857311320754717e-07,
+      "loss": 5.5085,
+      "step": 315
+    },
+    {
+      "epoch": 0.009435353088603863,
+      "grad_norm": 504.0,
+      "learning_rate": 1.8867924528301886e-07,
+      "loss": 5.6657,
+      "step": 320
+    },
+    {
+      "epoch": 0.009582780480613298,
+      "grad_norm": 460.0,
+      "learning_rate": 1.9162735849056601e-07,
+      "loss": 5.7185,
+      "step": 325
+    },
+    {
+      "epoch": 0.009730207872622734,
+      "grad_norm": 141.0,
+      "learning_rate": 1.945754716981132e-07,
+      "loss": 5.5482,
+      "step": 330
+    },
+    {
+      "epoch": 0.009877635264632168,
+      "grad_norm": 560.0,
+      "learning_rate": 1.9752358490566037e-07,
+      "loss": 5.378,
+      "step": 335
+    },
+    {
+      "epoch": 0.010025062656641603,
+      "grad_norm": 512.0,
+      "learning_rate": 2.0047169811320755e-07,
+      "loss": 5.6247,
+      "step": 340
+    },
+    {
+      "epoch": 0.010172490048651039,
+      "grad_norm": 528.0,
+      "learning_rate": 2.034198113207547e-07,
+      "loss": 5.1692,
+      "step": 345
+    },
+    {
+      "epoch": 0.010319917440660475,
+      "grad_norm": 624.0,
+      "learning_rate": 2.0636792452830186e-07,
+      "loss": 5.43,
+      "step": 350
+    },
+    {
+      "epoch": 0.01046734483266991,
+      "grad_norm": 592.0,
+      "learning_rate": 2.0931603773584904e-07,
+      "loss": 5.2031,
+      "step": 355
+    },
+    {
+      "epoch": 0.010614772224679346,
+      "grad_norm": 396.0,
+      "learning_rate": 2.1226415094339622e-07,
+      "loss": 5.1502,
+      "step": 360
+    },
+    {
+      "epoch": 0.01076219961668878,
+      "grad_norm": 516.0,
+      "learning_rate": 2.1521226415094338e-07,
+      "loss": 5.826,
+      "step": 365
+    },
+    {
+      "epoch": 0.010909627008698216,
+      "grad_norm": 760.0,
+      "learning_rate": 2.1816037735849056e-07,
+      "loss": 5.5087,
+      "step": 370
+    },
+    {
+      "epoch": 0.011057054400707651,
+      "grad_norm": 420.0,
+      "learning_rate": 2.2110849056603774e-07,
+      "loss": 5.146,
+      "step": 375
+    },
+    {
+      "epoch": 0.011204481792717087,
+      "grad_norm": 192.0,
+      "learning_rate": 2.240566037735849e-07,
+      "loss": 5.3402,
+      "step": 380
+    },
+    {
+      "epoch": 0.011351909184726523,
+      "grad_norm": 370.0,
+      "learning_rate": 2.2700471698113207e-07,
+      "loss": 5.3036,
+      "step": 385
+    },
+    {
+      "epoch": 0.011499336576735958,
+      "grad_norm": 352.0,
+      "learning_rate": 2.2995283018867922e-07,
+      "loss": 5.1917,
+      "step": 390
+    },
+    {
+      "epoch": 0.011646763968745392,
+      "grad_norm": 414.0,
+      "learning_rate": 2.329009433962264e-07,
+      "loss": 5.0797,
+      "step": 395
+    },
+    {
+      "epoch": 0.011794191360754828,
+      "grad_norm": 484.0,
+      "learning_rate": 2.3584905660377358e-07,
+      "loss": 5.296,
+      "step": 400
+    },
+    {
+      "epoch": 0.011941618752764263,
+      "grad_norm": 508.0,
+      "learning_rate": 2.3879716981132076e-07,
+      "loss": 5.026,
+      "step": 405
+    },
+    {
+      "epoch": 0.012089046144773699,
+      "grad_norm": 450.0,
+      "learning_rate": 2.417452830188679e-07,
+      "loss": 4.9075,
+      "step": 410
+    },
+    {
+      "epoch": 0.012236473536783135,
+      "grad_norm": 382.0,
+      "learning_rate": 2.4469339622641507e-07,
+      "loss": 5.0301,
+      "step": 415
+    },
+    {
+      "epoch": 0.01238390092879257,
+      "grad_norm": 384.0,
+      "learning_rate": 2.476415094339623e-07,
+      "loss": 4.9027,
+      "step": 420
+    },
+    {
+      "epoch": 0.012531328320802004,
+      "grad_norm": 188.0,
+      "learning_rate": 2.505896226415094e-07,
+      "loss": 5.0141,
+      "step": 425
+    },
+    {
+      "epoch": 0.01267875571281144,
+      "grad_norm": 157.0,
+      "learning_rate": 2.535377358490566e-07,
+      "loss": 4.5994,
+      "step": 430
+    },
+    {
+      "epoch": 0.012826183104820876,
+      "grad_norm": 386.0,
+      "learning_rate": 2.5648584905660374e-07,
+      "loss": 4.8006,
+      "step": 435
+    },
+    {
+      "epoch": 0.012973610496830311,
+      "grad_norm": 270.0,
+      "learning_rate": 2.5943396226415094e-07,
+      "loss": 4.4348,
+      "step": 440
+    },
+    {
+      "epoch": 0.013121037888839747,
+      "grad_norm": 161.0,
+      "learning_rate": 2.623820754716981e-07,
+      "loss": 4.7565,
+      "step": 445
+    },
+    {
+      "epoch": 0.013268465280849183,
+      "grad_norm": 186.0,
+      "learning_rate": 2.653301886792453e-07,
+      "loss": 4.7097,
+      "step": 450
+    },
+    {
+      "epoch": 0.013415892672858616,
+      "grad_norm": 115.0,
+      "learning_rate": 2.6827830188679246e-07,
+      "loss": 4.3228,
+      "step": 455
+    },
+    {
+      "epoch": 0.013563320064868052,
+      "grad_norm": 83.0,
+      "learning_rate": 2.712264150943396e-07,
+      "loss": 4.52,
+      "step": 460
+    },
+    {
+      "epoch": 0.013710747456877488,
+      "grad_norm": 142.0,
+      "learning_rate": 2.7417452830188676e-07,
+      "loss": 4.4627,
+      "step": 465
+    },
+    {
+      "epoch": 0.013858174848886923,
+      "grad_norm": 247.0,
+      "learning_rate": 2.771226415094339e-07,
+      "loss": 4.2847,
+      "step": 470
+    },
+    {
+      "epoch": 0.014005602240896359,
+      "grad_norm": 59.0,
+      "learning_rate": 2.800707547169811e-07,
+      "loss": 4.5636,
+      "step": 475
+    },
+    {
+      "epoch": 0.014153029632905795,
+      "grad_norm": 168.0,
+      "learning_rate": 2.830188679245283e-07,
+      "loss": 4.3123,
+      "step": 480
+    },
+    {
+      "epoch": 0.014300457024915229,
+      "grad_norm": 106.0,
+      "learning_rate": 2.8596698113207543e-07,
+      "loss": 4.3122,
+      "step": 485
+    },
+    {
+      "epoch": 0.014447884416924664,
+      "grad_norm": 97.0,
+      "learning_rate": 2.8891509433962264e-07,
+      "loss": 4.2267,
+      "step": 490
+    },
+    {
+      "epoch": 0.0145953118089341,
+      "grad_norm": 73.0,
+      "learning_rate": 2.918632075471698e-07,
+      "loss": 4.1379,
+      "step": 495
+    },
+    {
+      "epoch": 0.014742739200943536,
+      "grad_norm": 60.25,
+      "learning_rate": 2.94811320754717e-07,
+      "loss": 4.0958,
+      "step": 500
+    },
+    {
+      "epoch": 0.014742739200943536,
+      "eval_loss": 5.273393630981445,
+      "eval_runtime": 4.7076,
+      "eval_samples_per_second": 84.119,
+      "eval_steps_per_second": 2.761,
+      "step": 500
+    },
+    {
+      "epoch": 0.014890166592952971,
+      "grad_norm": 77.0,
+      "learning_rate": 2.9775943396226415e-07,
+      "loss": 3.849,
+      "step": 505
+    },
+    {
+      "epoch": 0.015037593984962405,
+      "grad_norm": 83.0,
+      "learning_rate": 3.007075471698113e-07,
+      "loss": 4.2059,
+      "step": 510
+    },
+    {
+      "epoch": 0.01518502137697184,
+      "grad_norm": 120.0,
+      "learning_rate": 3.0365566037735846e-07,
+      "loss": 4.1134,
+      "step": 515
+    },
+    {
+      "epoch": 0.015332448768981276,
+      "grad_norm": 58.25,
+      "learning_rate": 3.066037735849056e-07,
+      "loss": 3.9189,
+      "step": 520
+    },
+    {
+      "epoch": 0.015479876160990712,
+      "grad_norm": 77.0,
+      "learning_rate": 3.095518867924528e-07,
+      "loss": 4.1662,
+      "step": 525
+    },
+    {
+      "epoch": 0.015627303553000148,
+      "grad_norm": 120.5,
+      "learning_rate": 3.1249999999999997e-07,
+      "loss": 4.0004,
+      "step": 530
+    },
+    {
+      "epoch": 0.01577473094500958,
+      "grad_norm": 51.75,
+      "learning_rate": 3.154481132075472e-07,
+      "loss": 4.1284,
+      "step": 535
+    },
+    {
+      "epoch": 0.01592215833701902,
+      "grad_norm": 56.0,
+      "learning_rate": 3.1839622641509433e-07,
+      "loss": 4.1131,
+      "step": 540
+    },
+    {
+      "epoch": 0.016069585729028453,
+      "grad_norm": 87.0,
+      "learning_rate": 3.2134433962264154e-07,
+      "loss": 4.0497,
+      "step": 545
+    },
+    {
+      "epoch": 0.01621701312103789,
+      "grad_norm": 53.25,
+      "learning_rate": 3.242924528301887e-07,
+      "loss": 3.8763,
+      "step": 550
+    },
+    {
+      "epoch": 0.016364440513047324,
+      "grad_norm": 80.5,
+      "learning_rate": 3.272405660377358e-07,
+      "loss": 4.1176,
+      "step": 555
+    },
+    {
+      "epoch": 0.016511867905056758,
+      "grad_norm": 72.0,
+      "learning_rate": 3.30188679245283e-07,
+      "loss": 4.0426,
+      "step": 560
+    },
+    {
+      "epoch": 0.016659295297066196,
+      "grad_norm": 52.25,
+      "learning_rate": 3.3313679245283015e-07,
+      "loss": 3.7246,
+      "step": 565
+    },
+    {
+      "epoch": 0.01680672268907563,
+      "grad_norm": 50.0,
+      "learning_rate": 3.3608490566037736e-07,
+      "loss": 3.914,
+      "step": 570
+    },
+    {
+      "epoch": 0.016954150081085067,
+      "grad_norm": 48.25,
+      "learning_rate": 3.390330188679245e-07,
+      "loss": 4.1672,
+      "step": 575
+    },
+    {
+      "epoch": 0.0171015774730945,
+      "grad_norm": 56.5,
+      "learning_rate": 3.4198113207547167e-07,
+      "loss": 3.7885,
+      "step": 580
+    },
+    {
+      "epoch": 0.017249004865103935,
+      "grad_norm": 130.0,
+      "learning_rate": 3.449292452830189e-07,
+      "loss": 3.9511,
+      "step": 585
+    },
+    {
+      "epoch": 0.017396432257113372,
+      "grad_norm": 69.5,
+      "learning_rate": 3.4787735849056603e-07,
+      "loss": 4.0633,
+      "step": 590
+    },
+    {
+      "epoch": 0.017543859649122806,
+      "grad_norm": 49.5,
+      "learning_rate": 3.5082547169811323e-07,
+      "loss": 3.801,
+      "step": 595
+    },
+    {
+      "epoch": 0.017691287041132243,
+      "grad_norm": 38.0,
+      "learning_rate": 3.5377358490566033e-07,
+      "loss": 3.6217,
+      "step": 600
+    },
+    {
+      "epoch": 0.017838714433141677,
+      "grad_norm": 35.75,
+      "learning_rate": 3.567216981132075e-07,
+      "loss": 3.9652,
+      "step": 605
+    },
+    {
+      "epoch": 0.017986141825151115,
+      "grad_norm": 52.0,
+      "learning_rate": 3.596698113207547e-07,
+      "loss": 3.7182,
+      "step": 610
+    },
+    {
+      "epoch": 0.01813356921716055,
+      "grad_norm": 58.5,
+      "learning_rate": 3.6261792452830185e-07,
+      "loss": 3.6423,
+      "step": 615
+    },
+    {
+      "epoch": 0.018280996609169983,
+      "grad_norm": 46.25,
+      "learning_rate": 3.6556603773584905e-07,
+      "loss": 3.7992,
+      "step": 620
+    },
+    {
+      "epoch": 0.01842842400117942,
+      "grad_norm": 42.25,
+      "learning_rate": 3.685141509433962e-07,
+      "loss": 3.8397,
+      "step": 625
+    },
+    {
+      "epoch": 0.018575851393188854,
+      "grad_norm": 36.0,
+      "learning_rate": 3.714622641509434e-07,
+      "loss": 3.6173,
+      "step": 630
+    },
+    {
+      "epoch": 0.01872327878519829,
+      "grad_norm": 49.75,
+      "learning_rate": 3.7441037735849057e-07,
+      "loss": 4.0317,
+      "step": 635
+    },
+    {
+      "epoch": 0.018870706177207725,
+      "grad_norm": 38.5,
+      "learning_rate": 3.773584905660377e-07,
+      "loss": 3.7633,
+      "step": 640
+    },
+    {
+      "epoch": 0.01901813356921716,
+      "grad_norm": 41.75,
+      "learning_rate": 3.803066037735849e-07,
+      "loss": 3.8188,
+      "step": 645
+    },
+    {
+      "epoch": 0.019165560961226596,
+      "grad_norm": 63.5,
+      "learning_rate": 3.8325471698113203e-07,
+      "loss": 3.7537,
+      "step": 650
+    },
+    {
+      "epoch": 0.01931298835323603,
+      "grad_norm": 49.25,
+      "learning_rate": 3.8620283018867924e-07,
+      "loss": 3.6774,
+      "step": 655
+    },
+    {
+      "epoch": 0.019460415745245468,
+      "grad_norm": 39.0,
+      "learning_rate": 3.891509433962264e-07,
+      "loss": 3.7567,
+      "step": 660
+    },
+    {
+      "epoch": 0.0196078431372549,
+      "grad_norm": 39.0,
+      "learning_rate": 3.920990566037736e-07,
+      "loss": 3.756,
+      "step": 665
+    },
+    {
+      "epoch": 0.019755270529264336,
+      "grad_norm": 69.0,
+      "learning_rate": 3.9504716981132075e-07,
+      "loss": 3.6765,
+      "step": 670
+    },
+    {
+      "epoch": 0.019902697921273773,
+      "grad_norm": 60.75,
+      "learning_rate": 3.979952830188679e-07,
+      "loss": 3.949,
+      "step": 675
+    },
+    {
+      "epoch": 0.020050125313283207,
+      "grad_norm": 47.25,
+      "learning_rate": 4.009433962264151e-07,
+      "loss": 3.7218,
+      "step": 680
+    },
+    {
+      "epoch": 0.020197552705292644,
+      "grad_norm": 93.5,
+      "learning_rate": 4.038915094339622e-07,
+      "loss": 3.6829,
+      "step": 685
+    },
+    {
+      "epoch": 0.020344980097302078,
+      "grad_norm": 39.0,
+      "learning_rate": 4.068396226415094e-07,
+      "loss": 3.5724,
+      "step": 690
+    },
+    {
+      "epoch": 0.020492407489311516,
+      "grad_norm": 62.0,
+      "learning_rate": 4.0978773584905657e-07,
+      "loss": 3.6344,
+      "step": 695
+    },
+    {
+      "epoch": 0.02063983488132095,
+      "grad_norm": 196.0,
+      "learning_rate": 4.127358490566037e-07,
+      "loss": 3.5317,
+      "step": 700
+    },
+    {
+      "epoch": 0.020787262273330383,
+      "grad_norm": 41.0,
+      "learning_rate": 4.1568396226415093e-07,
+      "loss": 3.9269,
+      "step": 705
+    },
+    {
+      "epoch": 0.02093468966533982,
+      "grad_norm": 49.0,
+      "learning_rate": 4.186320754716981e-07,
+      "loss": 3.7463,
+      "step": 710
+    },
+    {
+      "epoch": 0.021082117057349255,
+      "grad_norm": 61.75,
+      "learning_rate": 4.215801886792453e-07,
+      "loss": 3.426,
+      "step": 715
+    },
+    {
+      "epoch": 0.021229544449358692,
+      "grad_norm": 40.25,
+      "learning_rate": 4.2452830188679244e-07,
+      "loss": 3.5511,
+      "step": 720
+    },
+    {
+      "epoch": 0.021376971841368126,
+      "grad_norm": 60.25,
+      "learning_rate": 4.2747641509433965e-07,
+      "loss": 3.8714,
+      "step": 725
+    },
+    {
+      "epoch": 0.02152439923337756,
+      "grad_norm": 67.0,
+      "learning_rate": 4.3042452830188675e-07,
+      "loss": 3.756,
+      "step": 730
+    },
+    {
+      "epoch": 0.021671826625386997,
+      "grad_norm": 47.0,
+      "learning_rate": 4.333726415094339e-07,
+      "loss": 3.548,
+      "step": 735
+    },
+    {
+      "epoch": 0.02181925401739643,
+      "grad_norm": 32.25,
+      "learning_rate": 4.363207547169811e-07,
+      "loss": 3.6436,
+      "step": 740
+    },
+    {
+      "epoch": 0.02196668140940587,
+      "grad_norm": 54.75,
+      "learning_rate": 4.3926886792452826e-07,
+      "loss": 3.4437,
+      "step": 745
+    },
+    {
+      "epoch": 0.022114108801415303,
+      "grad_norm": 55.75,
+      "learning_rate": 4.4221698113207547e-07,
+      "loss": 3.6474,
+      "step": 750
+    },
+    {
+      "epoch": 0.02226153619342474,
+      "grad_norm": 49.0,
+      "learning_rate": 4.451650943396226e-07,
+      "loss": 3.7664,
+      "step": 755
+    },
+    {
+      "epoch": 0.022408963585434174,
+      "grad_norm": 37.0,
+      "learning_rate": 4.481132075471698e-07,
+      "loss": 3.4262,
+      "step": 760
+    },
+    {
+      "epoch": 0.022556390977443608,
+      "grad_norm": 72.0,
+      "learning_rate": 4.51061320754717e-07,
+      "loss": 3.5507,
+      "step": 765
+    },
+    {
+      "epoch": 0.022703818369453045,
+      "grad_norm": 34.75,
+      "learning_rate": 4.5400943396226414e-07,
+      "loss": 3.6294,
+      "step": 770
+    },
+    {
+      "epoch": 0.02285124576146248,
+      "grad_norm": 71.0,
+      "learning_rate": 4.569575471698113e-07,
+      "loss": 3.3822,
+      "step": 775
+    },
+    {
+      "epoch": 0.022998673153471916,
+      "grad_norm": 57.5,
+      "learning_rate": 4.5990566037735845e-07,
+      "loss": 3.6301,
+      "step": 780
+    },
+    {
+      "epoch": 0.02314610054548135,
+      "grad_norm": 44.5,
+      "learning_rate": 4.6285377358490565e-07,
+      "loss": 3.5542,
+      "step": 785
+    },
+    {
+      "epoch": 0.023293527937490784,
+      "grad_norm": 30.25,
+      "learning_rate": 4.658018867924528e-07,
+      "loss": 3.4738,
+      "step": 790
+    },
+    {
+      "epoch": 0.02344095532950022,
+      "grad_norm": 36.25,
+      "learning_rate": 4.6874999999999996e-07,
+      "loss": 3.6581,
+      "step": 795
+    },
+    {
+      "epoch": 0.023588382721509656,
+      "grad_norm": 27.5,
+      "learning_rate": 4.7169811320754717e-07,
+      "loss": 3.4342,
+      "step": 800
+    },
+    {
+      "epoch": 0.023735810113519093,
+      "grad_norm": 36.5,
+      "learning_rate": 4.746462264150943e-07,
+      "loss": 3.6826,
+      "step": 805
+    },
+    {
+      "epoch": 0.023883237505528527,
+      "grad_norm": 41.75,
+      "learning_rate": 4.775943396226415e-07,
+      "loss": 3.485,
+      "step": 810
+    },
+    {
+      "epoch": 0.024030664897537964,
+      "grad_norm": 171.0,
+      "learning_rate": 4.805424528301887e-07,
+      "loss": 3.5775,
+      "step": 815
+    },
+    {
+      "epoch": 0.024178092289547398,
+      "grad_norm": 50.25,
+      "learning_rate": 4.834905660377358e-07,
+      "loss": 3.6154,
+      "step": 820
+    },
+    {
+      "epoch": 0.024325519681556832,
+      "grad_norm": 28.875,
+      "learning_rate": 4.86438679245283e-07,
+      "loss": 3.6943,
+      "step": 825
+    },
+    {
+      "epoch": 0.02447294707356627,
+      "grad_norm": 35.75,
+      "learning_rate": 4.893867924528301e-07,
+      "loss": 3.7264,
+      "step": 830
+    },
+    {
+      "epoch": 0.024620374465575703,
+      "grad_norm": 46.5,
+      "learning_rate": 4.923349056603773e-07,
+      "loss": 3.4934,
+      "step": 835
+    },
+    {
+      "epoch": 0.02476780185758514,
+      "grad_norm": 42.0,
+      "learning_rate": 4.952830188679246e-07,
+      "loss": 3.5727,
+      "step": 840
+    },
+    {
+      "epoch": 0.024915229249594575,
+      "grad_norm": 27.5,
+      "learning_rate": 4.982311320754717e-07,
+      "loss": 3.3648,
+      "step": 845
+    },
+    {
+      "epoch": 0.02506265664160401,
+      "grad_norm": 22.75,
+      "learning_rate": 5.011792452830188e-07,
+      "loss": 3.5408,
+      "step": 850
+    },
+    {
+      "epoch": 0.025210084033613446,
+      "grad_norm": 44.75,
+      "learning_rate": 5.04127358490566e-07,
+      "loss": 3.7565,
+      "step": 855
+    },
+    {
+      "epoch": 0.02535751142562288,
+      "grad_norm": 33.75,
+      "learning_rate": 5.070754716981132e-07,
+      "loss": 3.5236,
+      "step": 860
+    },
+    {
+      "epoch": 0.025504938817632317,
+      "grad_norm": 79.0,
+      "learning_rate": 5.100235849056603e-07,
+      "loss": 3.4544,
+      "step": 865
+    },
+    {
+      "epoch": 0.02565236620964175,
+      "grad_norm": 54.75,
+      "learning_rate": 5.129716981132075e-07,
+      "loss": 3.3027,
+      "step": 870
+    },
+    {
+      "epoch": 0.025799793601651185,
+      "grad_norm": 37.0,
+      "learning_rate": 5.159198113207547e-07,
+      "loss": 3.2541,
+      "step": 875
+    },
+    {
+      "epoch": 0.025947220993660623,
+      "grad_norm": 33.0,
+      "learning_rate": 5.188679245283019e-07,
+      "loss": 3.5423,
+      "step": 880
+    },
+    {
+      "epoch": 0.026094648385670056,
+      "grad_norm": 31.125,
+      "learning_rate": 5.21816037735849e-07,
+      "loss": 3.5534,
+      "step": 885
+    },
+    {
+      "epoch": 0.026242075777679494,
+      "grad_norm": 30.25,
+      "learning_rate": 5.247641509433962e-07,
+      "loss": 3.6136,
+      "step": 890
+    },
+    {
+      "epoch": 0.026389503169688928,
+      "grad_norm": 29.125,
+      "learning_rate": 5.277122641509433e-07,
+      "loss": 3.3698,
+      "step": 895
+    },
+    {
+      "epoch": 0.026536930561698365,
+      "grad_norm": 30.5,
+      "learning_rate": 5.306603773584906e-07,
+      "loss": 3.4885,
+      "step": 900
+    },
+    {
+      "epoch": 0.0266843579537078,
+      "grad_norm": 24.375,
+      "learning_rate": 5.336084905660378e-07,
+      "loss": 3.3895,
+      "step": 905
+    },
+    {
+      "epoch": 0.026831785345717233,
+      "grad_norm": 34.5,
+      "learning_rate": 5.365566037735849e-07,
+      "loss": 3.5425,
+      "step": 910
+    },
+    {
+      "epoch": 0.02697921273772667,
+      "grad_norm": 53.5,
+      "learning_rate": 5.395047169811321e-07,
+      "loss": 3.5077,
+      "step": 915
+    },
+    {
+      "epoch": 0.027126640129736104,
+      "grad_norm": 25.375,
+      "learning_rate": 5.424528301886792e-07,
+      "loss": 3.3913,
+      "step": 920
+    },
+    {
+      "epoch": 0.02727406752174554,
+      "grad_norm": 27.75,
+      "learning_rate": 5.454009433962265e-07,
+      "loss": 3.467,
+      "step": 925
+    },
+    {
+      "epoch": 0.027421494913754976,
+      "grad_norm": 53.0,
+      "learning_rate": 5.483490566037735e-07,
+      "loss": 3.7339,
+      "step": 930
+    },
+    {
+      "epoch": 0.02756892230576441,
+      "grad_norm": 34.0,
+      "learning_rate": 5.512971698113207e-07,
+      "loss": 3.4162,
+      "step": 935
+    },
+    {
+      "epoch": 0.027716349697773847,
+      "grad_norm": 26.125,
+      "learning_rate": 5.542452830188678e-07,
+      "loss": 3.4891,
+      "step": 940
+    },
+    {
+      "epoch": 0.02786377708978328,
+      "grad_norm": 30.625,
+      "learning_rate": 5.57193396226415e-07,
+      "loss": 3.6314,
+      "step": 945
+    },
+    {
+      "epoch": 0.028011204481792718,
+      "grad_norm": 26.5,
+      "learning_rate": 5.601415094339622e-07,
+      "loss": 3.5523,
+      "step": 950
+    },
+    {
+      "epoch": 0.028158631873802152,
+      "grad_norm": 33.75,
+      "learning_rate": 5.630896226415094e-07,
+      "loss": 3.3893,
+      "step": 955
+    },
+    {
+      "epoch": 0.02830605926581159,
+      "grad_norm": 32.0,
+      "learning_rate": 5.660377358490566e-07,
+      "loss": 3.3843,
+      "step": 960
+    },
+    {
+      "epoch": 0.028453486657821023,
+      "grad_norm": 29.5,
+      "learning_rate": 5.689858490566037e-07,
+      "loss": 3.5737,
+      "step": 965
+    },
+    {
+      "epoch": 0.028600914049830457,
+      "grad_norm": 71.0,
+      "learning_rate": 5.719339622641509e-07,
+      "loss": 3.2828,
+      "step": 970
+    },
+    {
+      "epoch": 0.028748341441839895,
+      "grad_norm": 42.25,
+      "learning_rate": 5.748820754716981e-07,
+      "loss": 3.414,
+      "step": 975
+    },
+    {
+      "epoch": 0.02889576883384933,
+      "grad_norm": 30.0,
+      "learning_rate": 5.778301886792453e-07,
+      "loss": 3.3214,
+      "step": 980
+    },
+    {
+      "epoch": 0.029043196225858766,
+      "grad_norm": 24.0,
+      "learning_rate": 5.807783018867924e-07,
+      "loss": 3.3956,
+      "step": 985
+    },
+    {
+      "epoch": 0.0291906236178682,
+      "grad_norm": 29.0,
+      "learning_rate": 5.837264150943396e-07,
+      "loss": 3.4536,
+      "step": 990
+    },
+    {
+      "epoch": 0.029338051009877634,
+      "grad_norm": 73.0,
+      "learning_rate": 5.866745283018868e-07,
+      "loss": 3.8505,
+      "step": 995
+    },
+    {
+      "epoch": 0.02948547840188707,
+      "grad_norm": 28.875,
+      "learning_rate": 5.89622641509434e-07,
+      "loss": 3.4702,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02948547840188707,
+      "eval_loss": 4.123366832733154,
+      "eval_runtime": 4.7173,
+      "eval_samples_per_second": 83.946,
+      "eval_steps_per_second": 2.756,
+      "step": 1000
+    },
+    {
+      "epoch": 0.029632905793896505,
+      "grad_norm": 40.5,
+      "learning_rate": 5.925707547169812e-07,
+      "loss": 3.4618,
+      "step": 1005
+    },
+    {
+      "epoch": 0.029780333185905943,
+      "grad_norm": 31.0,
+      "learning_rate": 5.955188679245283e-07,
+      "loss": 3.3067,
+      "step": 1010
+    },
+    {
+      "epoch": 0.029927760577915376,
+      "grad_norm": 25.5,
+      "learning_rate": 5.984669811320755e-07,
+      "loss": 3.2988,
+      "step": 1015
+    },
+    {
+      "epoch": 0.03007518796992481,
+      "grad_norm": 24.875,
+      "learning_rate": 6.014150943396226e-07,
+      "loss": 3.4574,
+      "step": 1020
+    },
+    {
+      "epoch": 0.030222615361934248,
+      "grad_norm": 34.0,
+      "learning_rate": 6.043632075471698e-07,
+      "loss": 3.4777,
+      "step": 1025
+    },
+    {
+      "epoch": 0.03037004275394368,
+      "grad_norm": 21.375,
+      "learning_rate": 6.073113207547169e-07,
+      "loss": 3.2491,
+      "step": 1030
+    },
+    {
+      "epoch": 0.03051747014595312,
+      "grad_norm": 32.75,
+      "learning_rate": 6.102594339622641e-07,
+      "loss": 3.4012,
+      "step": 1035
+    },
+    {
+      "epoch": 0.030664897537962553,
+      "grad_norm": 20.625,
+      "learning_rate": 6.132075471698112e-07,
+      "loss": 3.2219,
+      "step": 1040
+    },
+    {
+      "epoch": 0.03081232492997199,
+      "grad_norm": 28.625,
+      "learning_rate": 6.161556603773585e-07,
+      "loss": 3.4702,
+      "step": 1045
+    },
+    {
+      "epoch": 0.030959752321981424,
+      "grad_norm": 32.5,
+      "learning_rate": 6.191037735849056e-07,
+      "loss": 3.2885,
+      "step": 1050
+    },
+    {
+      "epoch": 0.031107179713990858,
+      "grad_norm": 28.0,
+      "learning_rate": 6.220518867924528e-07,
+      "loss": 3.414,
+      "step": 1055
+    },
+    {
+      "epoch": 0.031254607106000296,
+      "grad_norm": 24.125,
+      "learning_rate": 6.249999999999999e-07,
+      "loss": 3.35,
+      "step": 1060
+    },
+    {
+      "epoch": 0.03140203449800973,
+      "grad_norm": 26.25,
+      "learning_rate": 6.279481132075471e-07,
+      "loss": 3.4552,
+      "step": 1065
+    },
+    {
+      "epoch": 0.03154946189001916,
+      "grad_norm": 78.0,
+      "learning_rate": 6.308962264150944e-07,
+      "loss": 3.2664,
+      "step": 1070
+    },
+    {
+      "epoch": 0.0316968892820286,
+      "grad_norm": 27.375,
+      "learning_rate": 6.338443396226415e-07,
+      "loss": 3.4228,
+      "step": 1075
+    },
+    {
+      "epoch": 0.03184431667403804,
+      "grad_norm": 32.25,
+      "learning_rate": 6.367924528301887e-07,
+      "loss": 3.2762,
+      "step": 1080
+    },
+    {
+      "epoch": 0.03199174406604747,
+      "grad_norm": 25.875,
+      "learning_rate": 6.397405660377358e-07,
+      "loss": 3.2858,
+      "step": 1085
+    },
+    {
+      "epoch": 0.032139171458056906,
+      "grad_norm": 27.875,
+      "learning_rate": 6.426886792452831e-07,
+      "loss": 3.3251,
+      "step": 1090
+    },
+    {
+      "epoch": 0.03228659885006634,
+      "grad_norm": 21.5,
+      "learning_rate": 6.456367924528302e-07,
+      "loss": 3.3962,
+      "step": 1095
+    },
+    {
+      "epoch": 0.03243402624207578,
+      "grad_norm": 20.5,
+      "learning_rate": 6.485849056603774e-07,
+      "loss": 3.4683,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03258145363408521,
+      "grad_norm": 23.75,
+      "learning_rate": 6.515330188679244e-07,
+      "loss": 3.2019,
+      "step": 1105
+    },
+    {
+      "epoch": 0.03272888102609465,
+      "grad_norm": 55.5,
+      "learning_rate": 6.544811320754716e-07,
+      "loss": 3.4679,
+      "step": 1110
+    },
+    {
+      "epoch": 0.032876308418104086,
+      "grad_norm": 36.75,
+      "learning_rate": 6.574292452830188e-07,
+      "loss": 3.3551,
+      "step": 1115
+    },
+    {
+      "epoch": 0.033023735810113516,
+      "grad_norm": 23.5,
+      "learning_rate": 6.60377358490566e-07,
+      "loss": 3.3352,
+      "step": 1120
+    },
+    {
+      "epoch": 0.033171163202122954,
+      "grad_norm": 31.5,
+      "learning_rate": 6.633254716981132e-07,
+      "loss": 3.4598,
+      "step": 1125
+    },
+    {
+      "epoch": 0.03331859059413239,
+      "grad_norm": 36.0,
+      "learning_rate": 6.662735849056603e-07,
+      "loss": 3.5186,
+      "step": 1130
+    },
+    {
+      "epoch": 0.03346601798614183,
+      "grad_norm": 34.75,
+      "learning_rate": 6.692216981132075e-07,
+      "loss": 3.3776,
+      "step": 1135
+    },
+    {
+      "epoch": 0.03361344537815126,
+      "grad_norm": 45.0,
+      "learning_rate": 6.721698113207547e-07,
+      "loss": 3.4942,
+      "step": 1140
+    },
+    {
+      "epoch": 0.033760872770160696,
+      "grad_norm": 21.0,
+      "learning_rate": 6.751179245283019e-07,
+      "loss": 3.4764,
+      "step": 1145
+    },
+    {
+      "epoch": 0.033908300162170134,
+      "grad_norm": 21.875,
+      "learning_rate": 6.78066037735849e-07,
+      "loss": 3.3914,
+      "step": 1150
+    },
+    {
+      "epoch": 0.034055727554179564,
+      "grad_norm": 22.375,
+      "learning_rate": 6.810141509433962e-07,
+      "loss": 3.2916,
+      "step": 1155
+    },
+    {
+      "epoch": 0.034203154946189,
+      "grad_norm": 24.875,
+      "learning_rate": 6.839622641509433e-07,
+      "loss": 3.3966,
+      "step": 1160
+    },
+    {
+      "epoch": 0.03435058233819844,
+      "grad_norm": 32.25,
+      "learning_rate": 6.869103773584906e-07,
+      "loss": 3.2619,
+      "step": 1165
+    },
+    {
+      "epoch": 0.03449800973020787,
+      "grad_norm": 24.875,
+      "learning_rate": 6.898584905660377e-07,
+      "loss": 3.2777,
+      "step": 1170
+    },
+    {
+      "epoch": 0.03464543712221731,
+      "grad_norm": 20.875,
+      "learning_rate": 6.928066037735849e-07,
+      "loss": 3.228,
+      "step": 1175
+    },
+    {
+      "epoch": 0.034792864514226744,
+      "grad_norm": 24.0,
+      "learning_rate": 6.957547169811321e-07,
+      "loss": 3.2484,
+      "step": 1180
+    },
+    {
+      "epoch": 0.03494029190623618,
+      "grad_norm": 21.5,
+      "learning_rate": 6.987028301886792e-07,
+      "loss": 3.2644,
+      "step": 1185
+    },
+    {
+      "epoch": 0.03508771929824561,
+      "grad_norm": 21.0,
+      "learning_rate": 7.016509433962265e-07,
+      "loss": 3.0695,
+      "step": 1190
+    },
+    {
+      "epoch": 0.03523514669025505,
+      "grad_norm": 35.25,
+      "learning_rate": 7.045990566037735e-07,
+      "loss": 3.3176,
+      "step": 1195
+    },
+    {
+      "epoch": 0.03538257408226449,
+      "grad_norm": 19.25,
+      "learning_rate": 7.075471698113207e-07,
+      "loss": 3.2097,
+      "step": 1200
+    },
+    {
+      "epoch": 0.03553000147427392,
+      "grad_norm": 25.875,
+      "learning_rate": 7.104952830188678e-07,
+      "loss": 3.3202,
+      "step": 1205
+    },
+    {
+      "epoch": 0.035677428866283355,
+      "grad_norm": 27.25,
+      "learning_rate": 7.13443396226415e-07,
+      "loss": 3.1499,
+      "step": 1210
+    },
+    {
+      "epoch": 0.03582485625829279,
+      "grad_norm": 34.25,
+      "learning_rate": 7.163915094339622e-07,
+      "loss": 3.308,
+      "step": 1215
+    },
+    {
+      "epoch": 0.03597228365030223,
+      "grad_norm": 23.875,
+      "learning_rate": 7.193396226415094e-07,
+      "loss": 3.0438,
+      "step": 1220
+    },
+    {
+      "epoch": 0.03611971104231166,
+      "grad_norm": 26.5,
+      "learning_rate": 7.222877358490565e-07,
+      "loss": 3.2286,
+      "step": 1225
+    },
+    {
+      "epoch": 0.0362671384343211,
+      "grad_norm": 25.0,
+      "learning_rate": 7.252358490566037e-07,
+      "loss": 3.1494,
+      "step": 1230
+    },
+    {
+      "epoch": 0.036414565826330535,
+      "grad_norm": 33.5,
+      "learning_rate": 7.28183962264151e-07,
+      "loss": 3.3187,
+      "step": 1235
+    },
+    {
+      "epoch": 0.036561993218339965,
+      "grad_norm": 20.625,
+      "learning_rate": 7.311320754716981e-07,
+      "loss": 3.2469,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0367094206103494,
+      "grad_norm": 20.75,
+      "learning_rate": 7.340801886792453e-07,
+      "loss": 3.2862,
+      "step": 1245
+    },
+    {
+      "epoch": 0.03685684800235884,
+      "grad_norm": 27.875,
+      "learning_rate": 7.370283018867924e-07,
+      "loss": 3.2839,
+      "step": 1250
+    },
+    {
+      "epoch": 0.03700427539436827,
+      "grad_norm": 20.75,
+      "learning_rate": 7.399764150943396e-07,
+      "loss": 3.3897,
+      "step": 1255
+    },
+    {
+      "epoch": 0.03715170278637771,
+      "grad_norm": 27.125,
+      "learning_rate": 7.429245283018868e-07,
+      "loss": 3.1699,
+      "step": 1260
+    },
+    {
+      "epoch": 0.037299130178387145,
+      "grad_norm": 21.75,
+      "learning_rate": 7.45872641509434e-07,
+      "loss": 3.163,
+      "step": 1265
+    },
+    {
+      "epoch": 0.03744655757039658,
+      "grad_norm": 21.625,
+      "learning_rate": 7.488207547169811e-07,
+      "loss": 3.244,
+      "step": 1270
+    },
+    {
+      "epoch": 0.03759398496240601,
+      "grad_norm": 27.75,
+      "learning_rate": 7.517688679245283e-07,
+      "loss": 3.2307,
+      "step": 1275
+    },
+    {
+      "epoch": 0.03774141235441545,
+      "grad_norm": 40.75,
+      "learning_rate": 7.547169811320754e-07,
+      "loss": 3.2905,
+      "step": 1280
+    },
+    {
+      "epoch": 0.03788883974642489,
+      "grad_norm": 25.25,
+      "learning_rate": 7.576650943396226e-07,
+      "loss": 3.3318,
+      "step": 1285
+    },
+    {
+      "epoch": 0.03803626713843432,
+      "grad_norm": 32.5,
+      "learning_rate": 7.606132075471698e-07,
+      "loss": 3.2618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.038183694530443756,
+      "grad_norm": 33.5,
+      "learning_rate": 7.635613207547169e-07,
+      "loss": 3.3272,
+      "step": 1295
+    },
+    {
+      "epoch": 0.03833112192245319,
+      "grad_norm": 28.125,
+      "learning_rate": 7.665094339622641e-07,
+      "loss": 3.1578,
+      "step": 1300
+    },
+    {
+      "epoch": 0.03847854931446263,
+      "grad_norm": 21.375,
+      "learning_rate": 7.694575471698112e-07,
+      "loss": 3.0873,
+      "step": 1305
+    },
+    {
+      "epoch": 0.03862597670647206,
+      "grad_norm": 23.75,
+      "learning_rate": 7.724056603773585e-07,
+      "loss": 3.3165,
+      "step": 1310
+    },
+    {
+      "epoch": 0.0387734040984815,
+      "grad_norm": 21.5,
+      "learning_rate": 7.753537735849056e-07,
+      "loss": 3.1587,
+      "step": 1315
+    },
+    {
+      "epoch": 0.038920831490490936,
+      "grad_norm": 20.5,
+      "learning_rate": 7.783018867924528e-07,
+      "loss": 3.3898,
+      "step": 1320
+    },
+    {
+      "epoch": 0.039068258882500366,
+      "grad_norm": 34.0,
+      "learning_rate": 7.812499999999999e-07,
+      "loss": 3.1372,
+      "step": 1325
+    },
+    {
+      "epoch": 0.0392156862745098,
+      "grad_norm": 18.875,
+      "learning_rate": 7.841981132075472e-07,
+      "loss": 3.2379,
+      "step": 1330
+    },
+    {
+      "epoch": 0.03936311366651924,
+      "grad_norm": 21.125,
+      "learning_rate": 7.871462264150943e-07,
+      "loss": 3.3005,
+      "step": 1335
+    },
+    {
+      "epoch": 0.03951054105852867,
+      "grad_norm": 23.25,
+      "learning_rate": 7.900943396226415e-07,
+      "loss": 3.1946,
+      "step": 1340
+    },
+    {
+      "epoch": 0.03965796845053811,
+      "grad_norm": 23.375,
+      "learning_rate": 7.930424528301887e-07,
+      "loss": 3.0714,
+      "step": 1345
+    },
+    {
+      "epoch": 0.039805395842547546,
+      "grad_norm": 28.375,
+      "learning_rate": 7.959905660377358e-07,
+      "loss": 3.3614,
+      "step": 1350
+    },
+    {
+      "epoch": 0.03995282323455698,
+      "grad_norm": 24.25,
+      "learning_rate": 7.989386792452831e-07,
+      "loss": 3.1599,
+      "step": 1355
+    },
+    {
+      "epoch": 0.040100250626566414,
+      "grad_norm": 20.375,
+      "learning_rate": 8.018867924528302e-07,
+      "loss": 3.1418,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04024767801857585,
+      "grad_norm": 23.5,
+      "learning_rate": 8.048349056603774e-07,
+      "loss": 2.9725,
+      "step": 1365
+    },
+    {
+      "epoch": 0.04039510541058529,
+      "grad_norm": 29.0,
+      "learning_rate": 8.077830188679244e-07,
+      "loss": 3.0553,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04054253280259472,
+      "grad_norm": 29.0,
+      "learning_rate": 8.107311320754716e-07,
+      "loss": 3.1955,
+      "step": 1375
+    },
+    {
+      "epoch": 0.040689960194604156,
+      "grad_norm": 21.75,
+      "learning_rate": 8.136792452830188e-07,
+      "loss": 3.1639,
+      "step": 1380
+    },
+    {
+      "epoch": 0.040837387586613594,
+      "grad_norm": 33.0,
+      "learning_rate": 8.16627358490566e-07,
+      "loss": 3.1981,
+      "step": 1385
+    },
+    {
+      "epoch": 0.04098481497862303,
+      "grad_norm": 25.625,
+      "learning_rate": 8.195754716981131e-07,
+      "loss": 3.081,
+      "step": 1390
+    },
+    {
+      "epoch": 0.04113224237063246,
+      "grad_norm": 24.0,
+      "learning_rate": 8.225235849056603e-07,
+      "loss": 3.0706,
+      "step": 1395
+    },
+    {
+      "epoch": 0.0412796697626419,
+      "grad_norm": 23.0,
+      "learning_rate": 8.254716981132074e-07,
+      "loss": 3.2926,
+      "step": 1400
+    },
+    {
+      "epoch": 0.041427097154651336,
+      "grad_norm": 23.375,
+      "learning_rate": 8.284198113207547e-07,
+      "loss": 3.1699,
+      "step": 1405
+    },
+    {
+      "epoch": 0.04157452454666077,
+      "grad_norm": 30.0,
+      "learning_rate": 8.313679245283019e-07,
+      "loss": 3.145,
+      "step": 1410
+    },
+    {
+      "epoch": 0.041721951938670204,
+      "grad_norm": 29.25,
+      "learning_rate": 8.34316037735849e-07,
+      "loss": 3.2222,
+      "step": 1415
+    },
+    {
+      "epoch": 0.04186937933067964,
+      "grad_norm": 36.5,
+      "learning_rate": 8.372641509433962e-07,
+      "loss": 3.2383,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04201680672268908,
+      "grad_norm": 19.875,
+      "learning_rate": 8.402122641509433e-07,
+      "loss": 2.9331,
+      "step": 1425
+    },
+    {
+      "epoch": 0.04216423411469851,
+      "grad_norm": 24.375,
+      "learning_rate": 8.431603773584906e-07,
+      "loss": 3.076,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04231166150670795,
+      "grad_norm": 24.125,
+      "learning_rate": 8.461084905660377e-07,
+      "loss": 3.1641,
+      "step": 1435
+    },
+    {
+      "epoch": 0.042459088898717384,
+      "grad_norm": 21.375,
+      "learning_rate": 8.490566037735849e-07,
+      "loss": 3.3244,
+      "step": 1440
+    },
+    {
+      "epoch": 0.042606516290726815,
+      "grad_norm": 21.125,
+      "learning_rate": 8.52004716981132e-07,
+      "loss": 3.1178,
+      "step": 1445
+    },
+    {
+      "epoch": 0.04275394368273625,
+      "grad_norm": 27.0,
+      "learning_rate": 8.549528301886793e-07,
+      "loss": 3.1739,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04290137107474569,
+      "grad_norm": 21.625,
+      "learning_rate": 8.579009433962265e-07,
+      "loss": 3.2333,
+      "step": 1455
+    },
+    {
+      "epoch": 0.04304879846675512,
+      "grad_norm": 31.25,
+      "learning_rate": 8.608490566037735e-07,
+      "loss": 3.2602,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04319622585876456,
+      "grad_norm": 22.0,
+      "learning_rate": 8.637971698113207e-07,
+      "loss": 3.2519,
+      "step": 1465
+    },
+    {
+      "epoch": 0.043343653250773995,
+      "grad_norm": 24.625,
+      "learning_rate": 8.667452830188678e-07,
+      "loss": 3.0988,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04349108064278343,
+      "grad_norm": 28.125,
+      "learning_rate": 8.696933962264151e-07,
+      "loss": 3.1193,
+      "step": 1475
+    },
+    {
+      "epoch": 0.04363850803479286,
+      "grad_norm": 20.625,
+      "learning_rate": 8.726415094339622e-07,
+      "loss": 3.2805,
+      "step": 1480
+    },
+    {
+      "epoch": 0.0437859354268023,
+      "grad_norm": 18.875,
+      "learning_rate": 8.755896226415094e-07,
+      "loss": 3.2113,
+      "step": 1485
+    },
+    {
+      "epoch": 0.04393336281881174,
+      "grad_norm": 17.625,
+      "learning_rate": 8.785377358490565e-07,
+      "loss": 3.2982,
+      "step": 1490
+    },
+    {
+      "epoch": 0.04408079021082117,
+      "grad_norm": 17.375,
+      "learning_rate": 8.814858490566037e-07,
+      "loss": 3.187,
+      "step": 1495
+    },
+    {
+      "epoch": 0.044228217602830605,
+      "grad_norm": 21.75,
+      "learning_rate": 8.844339622641509e-07,
+      "loss": 3.1377,
+      "step": 1500
+    },
+    {
+      "epoch": 0.044228217602830605,
+      "eval_loss": 3.5992095470428467,
+      "eval_runtime": 4.6937,
+      "eval_samples_per_second": 84.368,
+      "eval_steps_per_second": 2.77,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04437564499484004,
+      "grad_norm": 188.0,
+      "learning_rate": 8.873820754716981e-07,
+      "loss": 3.1779,
+      "step": 1505
+    },
+    {
+      "epoch": 0.04452307238684948,
+      "grad_norm": 45.5,
+      "learning_rate": 8.903301886792452e-07,
+      "loss": 3.2703,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04467049977885891,
+      "grad_norm": 20.875,
+      "learning_rate": 8.932783018867924e-07,
+      "loss": 3.1839,
+      "step": 1515
+    },
+    {
+      "epoch": 0.04481792717086835,
+      "grad_norm": 19.375,
+      "learning_rate": 8.962264150943396e-07,
+      "loss": 3.0066,
+      "step": 1520
+    },
+    {
+      "epoch": 0.044965354562877785,
+      "grad_norm": 29.5,
+      "learning_rate": 8.991745283018868e-07,
+      "loss": 3.2096,
+      "step": 1525
+    },
+    {
+      "epoch": 0.045112781954887216,
+      "grad_norm": 26.875,
+      "learning_rate": 9.02122641509434e-07,
+      "loss": 3.0473,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04526020934689665,
+      "grad_norm": 19.625,
+      "learning_rate": 9.050707547169811e-07,
+      "loss": 3.0378,
+      "step": 1535
+    },
+    {
+      "epoch": 0.04540763673890609,
+      "grad_norm": 17.5,
+      "learning_rate": 9.080188679245283e-07,
+      "loss": 3.2259,
+      "step": 1540
+    },
+    {
+      "epoch": 0.04555506413091552,
+      "grad_norm": 22.625,
+      "learning_rate": 9.109669811320755e-07,
+      "loss": 3.0831,
+      "step": 1545
+    },
+    {
+      "epoch": 0.04570249152292496,
+      "grad_norm": 19.125,
+      "learning_rate": 9.139150943396226e-07,
+      "loss": 3.1252,
+      "step": 1550
+    },
+    {
+      "epoch": 0.045849918914934396,
+      "grad_norm": 23.25,
+      "learning_rate": 9.168632075471697e-07,
+      "loss": 3.2609,
+      "step": 1555
+    },
+    {
+      "epoch": 0.04599734630694383,
+      "grad_norm": 27.625,
+      "learning_rate": 9.198113207547169e-07,
+      "loss": 3.1041,
+      "step": 1560
+    },
+    {
+      "epoch": 0.04614477369895326,
+      "grad_norm": 22.75,
+      "learning_rate": 9.22759433962264e-07,
+      "loss": 3.289,
+      "step": 1565
+    },
+    {
+      "epoch": 0.0462922010909627,
+      "grad_norm": 19.125,
+      "learning_rate": 9.257075471698113e-07,
+      "loss": 3.2432,
+      "step": 1570
+    },
+    {
+      "epoch": 0.04643962848297214,
+      "grad_norm": 16.5,
+      "learning_rate": 9.286556603773585e-07,
+      "loss": 3.0007,
+      "step": 1575
+    },
+    {
+      "epoch": 0.04658705587498157,
+      "grad_norm": 24.5,
+      "learning_rate": 9.316037735849056e-07,
+      "loss": 3.2563,
+      "step": 1580
+    },
+    {
+      "epoch": 0.046734483266991006,
+      "grad_norm": 19.75,
+      "learning_rate": 9.345518867924528e-07,
+      "loss": 3.0233,
+      "step": 1585
+    },
+    {
+      "epoch": 0.04688191065900044,
+      "grad_norm": 24.0,
+      "learning_rate": 9.374999999999999e-07,
+      "loss": 2.9052,
+      "step": 1590
+    },
+    {
+      "epoch": 0.04702933805100988,
+      "grad_norm": 21.375,
+      "learning_rate": 9.404481132075472e-07,
+      "loss": 3.157,
+      "step": 1595
+    },
+    {
+      "epoch": 0.04717676544301931,
+      "grad_norm": 37.0,
+      "learning_rate": 9.433962264150943e-07,
+      "loss": 2.9625,
+      "step": 1600
+    },
+    {
+      "epoch": 0.04732419283502875,
+      "grad_norm": 16.125,
+      "learning_rate": 9.463443396226415e-07,
+      "loss": 3.1684,
+      "step": 1605
+    },
+    {
+      "epoch": 0.047471620227038186,
+      "grad_norm": 17.0,
+      "learning_rate": 9.492924528301886e-07,
+      "loss": 3.1345,
+      "step": 1610
+    },
+    {
+      "epoch": 0.047619047619047616,
+      "grad_norm": 24.125,
+      "learning_rate": 9.522405660377358e-07,
+      "loss": 3.2707,
+      "step": 1615
+    },
+    {
+      "epoch": 0.047766475011057054,
+      "grad_norm": 19.375,
+      "learning_rate": 9.55188679245283e-07,
+      "loss": 3.1682,
+      "step": 1620
+    },
+    {
+      "epoch": 0.04791390240306649,
+      "grad_norm": 20.125,
+      "learning_rate": 9.581367924528302e-07,
+      "loss": 3.1835,
+      "step": 1625
+    },
+    {
+      "epoch": 0.04806132979507593,
+      "grad_norm": 19.75,
+      "learning_rate": 9.610849056603774e-07,
+      "loss": 3.0946,
+      "step": 1630
+    },
+    {
+      "epoch": 0.04820875718708536,
+      "grad_norm": 27.0,
+      "learning_rate": 9.640330188679245e-07,
+      "loss": 2.9732,
+      "step": 1635
+    },
+    {
+      "epoch": 0.048356184579094796,
+      "grad_norm": 28.0,
+      "learning_rate": 9.669811320754717e-07,
+      "loss": 3.0809,
+      "step": 1640
+    },
+    {
+      "epoch": 0.048503611971104234,
+      "grad_norm": 16.625,
+      "learning_rate": 9.699292452830188e-07,
+      "loss": 2.8662,
+      "step": 1645
+    },
+    {
+      "epoch": 0.048651039363113664,
+      "grad_norm": 25.0,
+      "learning_rate": 9.72877358490566e-07,
+      "loss": 3.1653,
+      "step": 1650
+    },
+    {
+      "epoch": 0.0487984667551231,
+      "grad_norm": 20.875,
+      "learning_rate": 9.758254716981131e-07,
+      "loss": 2.9102,
+      "step": 1655
+    },
+    {
+      "epoch": 0.04894589414713254,
+      "grad_norm": 15.625,
+      "learning_rate": 9.787735849056603e-07,
+      "loss": 3.0904,
+      "step": 1660
+    },
+    {
+      "epoch": 0.04909332153914197,
+      "grad_norm": 26.5,
+      "learning_rate": 9.817216981132074e-07,
+      "loss": 3.2188,
+      "step": 1665
+    },
+    {
+      "epoch": 0.04924074893115141,
+      "grad_norm": 20.0,
+      "learning_rate": 9.846698113207546e-07,
+      "loss": 3.0935,
+      "step": 1670
+    },
+    {
+      "epoch": 0.049388176323160844,
+      "grad_norm": 22.625,
+      "learning_rate": 9.876179245283017e-07,
+      "loss": 3.0112,
+      "step": 1675
+    },
+    {
+      "epoch": 0.04953560371517028,
+      "grad_norm": 28.875,
+      "learning_rate": 9.90566037735849e-07,
+      "loss": 3.0347,
+      "step": 1680
+    },
+    {
+      "epoch": 0.04968303110717971,
+      "grad_norm": 28.0,
+      "learning_rate": 9.935141509433963e-07,
+      "loss": 3.1383,
+      "step": 1685
+    },
+    {
+      "epoch": 0.04983045849918915,
+      "grad_norm": 22.625,
+      "learning_rate": 9.964622641509434e-07,
+      "loss": 3.195,
+      "step": 1690
+    },
+    {
+      "epoch": 0.04997788589119859,
+      "grad_norm": 17.625,
+      "learning_rate": 9.994103773584906e-07,
+      "loss": 3.0536,
+      "step": 1695
+    },
+    {
+      "epoch": 0.05012531328320802,
+      "grad_norm": 20.875,
+      "learning_rate": 1.0023584905660375e-06,
+      "loss": 3.1255,
+      "step": 1700
+    },
+    {
+      "epoch": 0.050272740675217455,
+      "grad_norm": 23.625,
+      "learning_rate": 1.0053066037735849e-06,
+      "loss": 3.0029,
+      "step": 1705
+    },
+    {
+      "epoch": 0.05042016806722689,
+      "grad_norm": 21.875,
+      "learning_rate": 1.008254716981132e-06,
+      "loss": 3.0566,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05056759545923633,
+      "grad_norm": 22.875,
+      "learning_rate": 1.0112028301886792e-06,
+      "loss": 3.235,
+      "step": 1715
+    },
+    {
+      "epoch": 0.05071502285124576,
+      "grad_norm": 21.25,
+      "learning_rate": 1.0141509433962263e-06,
+      "loss": 3.0852,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0508624502432552,
+      "grad_norm": 20.875,
+      "learning_rate": 1.0170990566037737e-06,
+      "loss": 2.9238,
+      "step": 1725
+    },
+    {
+      "epoch": 0.051009877635264635,
+      "grad_norm": 21.0,
+      "learning_rate": 1.0200471698113206e-06,
+      "loss": 2.9247,
+      "step": 1730
+    },
+    {
+      "epoch": 0.051157305027274065,
+      "grad_norm": 22.375,
+      "learning_rate": 1.022995283018868e-06,
+      "loss": 2.9932,
+      "step": 1735
+    },
+    {
+      "epoch": 0.0513047324192835,
+      "grad_norm": 21.625,
+      "learning_rate": 1.025943396226415e-06,
+      "loss": 3.1733,
+      "step": 1740
+    },
+    {
+      "epoch": 0.05145215981129294,
+      "grad_norm": 18.75,
+      "learning_rate": 1.0288915094339623e-06,
+      "loss": 3.0096,
+      "step": 1745
+    },
+    {
+      "epoch": 0.05159958720330237,
+      "grad_norm": 16.5,
+      "learning_rate": 1.0318396226415095e-06,
+      "loss": 3.0647,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05174701459531181,
+      "grad_norm": 20.75,
+      "learning_rate": 1.0347877358490566e-06,
+      "loss": 3.0325,
+      "step": 1755
+    },
+    {
+      "epoch": 0.051894441987321245,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0377358490566038e-06,
+      "loss": 3.0077,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05204186937933068,
+      "grad_norm": 18.625,
+      "learning_rate": 1.040683962264151e-06,
+      "loss": 2.9983,
+      "step": 1765
+    },
+    {
+      "epoch": 0.05218929677134011,
+      "grad_norm": 21.75,
+      "learning_rate": 1.043632075471698e-06,
+      "loss": 3.103,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05233672416334955,
+      "grad_norm": 19.375,
+      "learning_rate": 1.0465801886792452e-06,
+      "loss": 3.0497,
+      "step": 1775
+    },
+    {
+      "epoch": 0.05248415155535899,
+      "grad_norm": 27.125,
+      "learning_rate": 1.0495283018867924e-06,
+      "loss": 3.037,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05263157894736842,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.0524764150943395e-06,
+      "loss": 2.7835,
+      "step": 1785
+    },
+    {
+      "epoch": 0.052779006339377856,
+      "grad_norm": 20.75,
+      "learning_rate": 1.0554245283018867e-06,
+      "loss": 3.0535,
+      "step": 1790
+    },
+    {
+      "epoch": 0.05292643373138729,
+      "grad_norm": 19.375,
+      "learning_rate": 1.0583726415094338e-06,
+      "loss": 3.0061,
+      "step": 1795
+    },
+    {
+      "epoch": 0.05307386112339673,
+      "grad_norm": 18.375,
+      "learning_rate": 1.0613207547169812e-06,
+      "loss": 2.9059,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05322128851540616,
+      "grad_norm": 16.875,
+      "learning_rate": 1.0642688679245282e-06,
+      "loss": 3.0449,
+      "step": 1805
+    },
+    {
+      "epoch": 0.0533687159074156,
+      "grad_norm": 19.875,
+      "learning_rate": 1.0672169811320755e-06,
+      "loss": 2.905,
+      "step": 1810
+    },
+    {
+      "epoch": 0.053516143299425036,
+      "grad_norm": 20.25,
+      "learning_rate": 1.0701650943396225e-06,
+      "loss": 3.2346,
+      "step": 1815
+    },
+    {
+      "epoch": 0.053663570691434466,
+      "grad_norm": 17.875,
+      "learning_rate": 1.0731132075471698e-06,
+      "loss": 2.9078,
+      "step": 1820
+    },
+    {
+      "epoch": 0.0538109980834439,
+      "grad_norm": 20.125,
+      "learning_rate": 1.076061320754717e-06,
+      "loss": 3.1477,
+      "step": 1825
+    },
+    {
+      "epoch": 0.05395842547545334,
+      "grad_norm": 24.0,
+      "learning_rate": 1.0790094339622641e-06,
+      "loss": 2.8281,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05410585286746277,
+      "grad_norm": 20.625,
+      "learning_rate": 1.0819575471698113e-06,
+      "loss": 3.0376,
+      "step": 1835
+    },
+    {
+      "epoch": 0.05425328025947221,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.0849056603773584e-06,
+      "loss": 2.8431,
+      "step": 1840
+    },
+    {
+      "epoch": 0.054400707651481646,
+      "grad_norm": 18.875,
+      "learning_rate": 1.0878537735849056e-06,
+      "loss": 2.9487,
+      "step": 1845
+    },
+    {
+      "epoch": 0.05454813504349108,
+      "grad_norm": 30.75,
+      "learning_rate": 1.090801886792453e-06,
+      "loss": 2.9563,
+      "step": 1850
+    },
+    {
+      "epoch": 0.054695562435500514,
+      "grad_norm": 23.125,
+      "learning_rate": 1.09375e-06,
+      "loss": 2.871,
+      "step": 1855
+    },
+    {
+      "epoch": 0.05484298982750995,
+      "grad_norm": 17.0,
+      "learning_rate": 1.096698113207547e-06,
+      "loss": 3.029,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05499041721951939,
+      "grad_norm": 14.625,
+      "learning_rate": 1.0996462264150942e-06,
+      "loss": 2.898,
+      "step": 1865
+    },
+    {
+      "epoch": 0.05513784461152882,
+      "grad_norm": 21.25,
+      "learning_rate": 1.1025943396226414e-06,
+      "loss": 2.8212,
+      "step": 1870
+    },
+    {
+      "epoch": 0.055285272003538256,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1055424528301887e-06,
+      "loss": 2.9897,
+      "step": 1875
+    },
+    {
+      "epoch": 0.055432699395547694,
+      "grad_norm": 17.125,
+      "learning_rate": 1.1084905660377357e-06,
+      "loss": 2.8868,
+      "step": 1880
+    },
+    {
+      "epoch": 0.05558012678755713,
+      "grad_norm": 19.25,
+      "learning_rate": 1.111438679245283e-06,
+      "loss": 3.0397,
+      "step": 1885
+    },
+    {
+      "epoch": 0.05572755417956656,
+      "grad_norm": 19.75,
+      "learning_rate": 1.11438679245283e-06,
+      "loss": 2.8232,
+      "step": 1890
+    },
+    {
+      "epoch": 0.055874981571576,
+      "grad_norm": 36.75,
+      "learning_rate": 1.1173349056603773e-06,
+      "loss": 2.9727,
+      "step": 1895
+    },
+    {
+      "epoch": 0.056022408963585436,
+      "grad_norm": 39.25,
+      "learning_rate": 1.1202830188679245e-06,
+      "loss": 3.3028,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05616983635559487,
+      "grad_norm": 19.25,
+      "learning_rate": 1.1232311320754717e-06,
+      "loss": 3.1002,
+      "step": 1905
+    },
+    {
+      "epoch": 0.056317263747604304,
+      "grad_norm": 24.0,
+      "learning_rate": 1.1261792452830188e-06,
+      "loss": 3.163,
+      "step": 1910
+    },
+    {
+      "epoch": 0.05646469113961374,
+      "grad_norm": 23.375,
+      "learning_rate": 1.129127358490566e-06,
+      "loss": 2.9574,
+      "step": 1915
+    },
+    {
+      "epoch": 0.05661211853162318,
+      "grad_norm": 62.0,
+      "learning_rate": 1.1320754716981131e-06,
+      "loss": 2.9711,
+      "step": 1920
+    },
+    {
+      "epoch": 0.05675954592363261,
+      "grad_norm": 20.875,
+      "learning_rate": 1.1350235849056605e-06,
+      "loss": 2.9719,
+      "step": 1925
+    },
+    {
+      "epoch": 0.05690697331564205,
+      "grad_norm": 21.0,
+      "learning_rate": 1.1379716981132074e-06,
+      "loss": 2.9451,
+      "step": 1930
+    },
+    {
+      "epoch": 0.057054400707651484,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1409198113207548e-06,
+      "loss": 3.0775,
+      "step": 1935
+    },
+    {
+      "epoch": 0.057201828099660915,
+      "grad_norm": 18.0,
+      "learning_rate": 1.1438679245283017e-06,
+      "loss": 2.9611,
+      "step": 1940
+    },
+    {
+      "epoch": 0.05734925549167035,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1468160377358489e-06,
+      "loss": 2.8335,
+      "step": 1945
+    },
+    {
+      "epoch": 0.05749668288367979,
+      "grad_norm": 19.125,
+      "learning_rate": 1.1497641509433962e-06,
+      "loss": 2.9033,
+      "step": 1950
+    },
+    {
+      "epoch": 0.05764411027568922,
+      "grad_norm": 19.75,
+      "learning_rate": 1.1527122641509432e-06,
+      "loss": 2.957,
+      "step": 1955
+    },
+    {
+      "epoch": 0.05779153766769866,
+      "grad_norm": 21.375,
+      "learning_rate": 1.1556603773584906e-06,
+      "loss": 3.0282,
+      "step": 1960
+    },
+    {
+      "epoch": 0.057938965059708095,
+      "grad_norm": 26.375,
+      "learning_rate": 1.1586084905660375e-06,
+      "loss": 2.9575,
+      "step": 1965
+    },
+    {
+      "epoch": 0.05808639245171753,
+      "grad_norm": 16.375,
+      "learning_rate": 1.1615566037735849e-06,
+      "loss": 2.8847,
+      "step": 1970
+    },
+    {
+      "epoch": 0.05823381984372696,
+      "grad_norm": 21.625,
+      "learning_rate": 1.164504716981132e-06,
+      "loss": 2.8438,
+      "step": 1975
+    },
+    {
+      "epoch": 0.0583812472357364,
+      "grad_norm": 26.375,
+      "learning_rate": 1.1674528301886792e-06,
+      "loss": 3.0832,
+      "step": 1980
+    },
+    {
+      "epoch": 0.05852867462774584,
+      "grad_norm": 17.5,
+      "learning_rate": 1.1704009433962263e-06,
+      "loss": 3.0941,
+      "step": 1985
+    },
+    {
+      "epoch": 0.05867610201975527,
+      "grad_norm": 16.25,
+      "learning_rate": 1.1733490566037737e-06,
+      "loss": 2.716,
+      "step": 1990
+    },
+    {
+      "epoch": 0.058823529411764705,
+      "grad_norm": 71.0,
+      "learning_rate": 1.1762971698113206e-06,
+      "loss": 2.9973,
+      "step": 1995
+    },
+    {
+      "epoch": 0.05897095680377414,
+      "grad_norm": 24.625,
+      "learning_rate": 1.179245283018868e-06,
+      "loss": 2.9267,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05897095680377414,
+      "eval_loss": 3.27067494392395,
+      "eval_runtime": 4.6982,
+      "eval_samples_per_second": 84.287,
+      "eval_steps_per_second": 2.767,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05911838419578358,
+      "grad_norm": 16.5,
+      "learning_rate": 1.182193396226415e-06,
+      "loss": 2.7164,
+      "step": 2005
+    },
+    {
+      "epoch": 0.05926581158779301,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.1851415094339623e-06,
+      "loss": 2.8899,
+      "step": 2010
+    },
+    {
+      "epoch": 0.05941323897980245,
+      "grad_norm": 18.75,
+      "learning_rate": 1.1880896226415095e-06,
+      "loss": 2.926,
+      "step": 2015
+    },
+    {
+      "epoch": 0.059560666371811885,
+      "grad_norm": 23.125,
+      "learning_rate": 1.1910377358490566e-06,
+      "loss": 2.853,
+      "step": 2020
+    },
+    {
+      "epoch": 0.059708093763821316,
+      "grad_norm": 17.75,
+      "learning_rate": 1.1939858490566038e-06,
+      "loss": 2.8584,
+      "step": 2025
+    },
+    {
+      "epoch": 0.05985552115583075,
+      "grad_norm": 20.75,
+      "learning_rate": 1.196933962264151e-06,
+      "loss": 2.8365,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06000294854784019,
+      "grad_norm": 19.0,
+      "learning_rate": 1.199882075471698e-06,
+      "loss": 3.0571,
+      "step": 2035
+    },
+    {
+      "epoch": 0.06015037593984962,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.2028301886792452e-06,
+      "loss": 2.7813,
+      "step": 2040
+    },
+    {
+      "epoch": 0.06029780333185906,
+      "grad_norm": 16.125,
+      "learning_rate": 1.2057783018867924e-06,
+      "loss": 2.744,
+      "step": 2045
+    },
+    {
+      "epoch": 0.060445230723868495,
+      "grad_norm": 18.125,
+      "learning_rate": 1.2087264150943395e-06,
+      "loss": 2.8005,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06059265811587793,
+      "grad_norm": 22.875,
+      "learning_rate": 1.2116745283018867e-06,
+      "loss": 2.9454,
+      "step": 2055
+    },
+    {
+      "epoch": 0.06074008550788736,
+      "grad_norm": 16.875,
+      "learning_rate": 1.2146226415094338e-06,
+      "loss": 3.0735,
+      "step": 2060
+    },
+    {
+      "epoch": 0.0608875128998968,
+      "grad_norm": 40.75,
+      "learning_rate": 1.2175707547169812e-06,
+      "loss": 2.9832,
+      "step": 2065
+    },
+    {
+      "epoch": 0.06103494029190624,
+      "grad_norm": 22.375,
+      "learning_rate": 1.2205188679245281e-06,
+      "loss": 3.0261,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06118236768391567,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2234669811320755e-06,
+      "loss": 3.0268,
+      "step": 2075
+    },
+    {
+      "epoch": 0.061329795075925106,
+      "grad_norm": 22.5,
+      "learning_rate": 1.2264150943396225e-06,
+      "loss": 2.9181,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06147722246793454,
+      "grad_norm": 43.75,
+      "learning_rate": 1.2293632075471698e-06,
+      "loss": 2.7782,
+      "step": 2085
+    },
+    {
+      "epoch": 0.06162464985994398,
+      "grad_norm": 15.125,
+      "learning_rate": 1.232311320754717e-06,
+      "loss": 2.8703,
+      "step": 2090
+    },
+    {
+      "epoch": 0.06177207725195341,
+      "grad_norm": 21.75,
+      "learning_rate": 1.2352594339622641e-06,
+      "loss": 2.972,
+      "step": 2095
+    },
+    {
+      "epoch": 0.06191950464396285,
+      "grad_norm": 16.75,
+      "learning_rate": 1.2382075471698113e-06,
+      "loss": 2.8651,
+      "step": 2100
+    },
+    {
+      "epoch": 0.062066932035972286,
+      "grad_norm": 23.375,
+      "learning_rate": 1.2411556603773584e-06,
+      "loss": 3.1965,
+      "step": 2105
+    },
+    {
+      "epoch": 0.062214359427981716,
+      "grad_norm": 21.125,
+      "learning_rate": 1.2441037735849056e-06,
+      "loss": 2.9804,
+      "step": 2110
+    },
+    {
+      "epoch": 0.062361786819991154,
+      "grad_norm": 23.125,
+      "learning_rate": 1.247051886792453e-06,
+      "loss": 2.9106,
+      "step": 2115
+    },
+    {
+      "epoch": 0.06250921421200059,
+      "grad_norm": 19.5,
+      "learning_rate": 1.2499999999999999e-06,
+      "loss": 2.9487,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06265664160401002,
+      "grad_norm": 43.5,
+      "learning_rate": 1.252948113207547e-06,
+      "loss": 2.9301,
+      "step": 2125
+    },
+    {
+      "epoch": 0.06280406899601947,
+      "grad_norm": 17.75,
+      "learning_rate": 1.2558962264150942e-06,
+      "loss": 2.8569,
+      "step": 2130
+    },
+    {
+      "epoch": 0.0629514963880289,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2588443396226414e-06,
+      "loss": 2.851,
+      "step": 2135
+    },
+    {
+      "epoch": 0.06309892378003833,
+      "grad_norm": 19.0,
+      "learning_rate": 1.2617924528301887e-06,
+      "loss": 2.959,
+      "step": 2140
+    },
+    {
+      "epoch": 0.06324635117204777,
+      "grad_norm": 16.0,
+      "learning_rate": 1.2647405660377357e-06,
+      "loss": 3.0111,
+      "step": 2145
+    },
+    {
+      "epoch": 0.0633937785640572,
+      "grad_norm": 21.25,
+      "learning_rate": 1.267688679245283e-06,
+      "loss": 2.9492,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06354120595606663,
+      "grad_norm": 27.875,
+      "learning_rate": 1.27063679245283e-06,
+      "loss": 2.8556,
+      "step": 2155
+    },
+    {
+      "epoch": 0.06368863334807608,
+      "grad_norm": 292.0,
+      "learning_rate": 1.2735849056603773e-06,
+      "loss": 3.0161,
+      "step": 2160
+    },
+    {
+      "epoch": 0.0638360607400855,
+      "grad_norm": 27.625,
+      "learning_rate": 1.2765330188679245e-06,
+      "loss": 2.8943,
+      "step": 2165
+    },
+    {
+      "epoch": 0.06398348813209494,
+      "grad_norm": 20.5,
+      "learning_rate": 1.2794811320754716e-06,
+      "loss": 2.9569,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06413091552410438,
+      "grad_norm": 16.375,
+      "learning_rate": 1.2824292452830188e-06,
+      "loss": 2.8933,
+      "step": 2175
+    },
+    {
+      "epoch": 0.06427834291611381,
+      "grad_norm": 21.75,
+      "learning_rate": 1.2853773584905662e-06,
+      "loss": 2.8857,
+      "step": 2180
+    },
+    {
+      "epoch": 0.06442577030812324,
+      "grad_norm": 25.375,
+      "learning_rate": 1.288325471698113e-06,
+      "loss": 2.7119,
+      "step": 2185
+    },
+    {
+      "epoch": 0.06457319770013269,
+      "grad_norm": 15.5,
+      "learning_rate": 1.2912735849056605e-06,
+      "loss": 2.9744,
+      "step": 2190
+    },
+    {
+      "epoch": 0.06472062509214212,
+      "grad_norm": 28.0,
+      "learning_rate": 1.2942216981132074e-06,
+      "loss": 2.8828,
+      "step": 2195
+    },
+    {
+      "epoch": 0.06486805248415156,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2971698113207548e-06,
+      "loss": 2.9426,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06501547987616099,
+      "grad_norm": 16.75,
+      "learning_rate": 1.300117924528302e-06,
+      "loss": 2.9991,
+      "step": 2205
+    },
+    {
+      "epoch": 0.06516290726817042,
+      "grad_norm": 32.5,
+      "learning_rate": 1.3030660377358489e-06,
+      "loss": 2.9827,
+      "step": 2210
+    },
+    {
+      "epoch": 0.06531033466017987,
+      "grad_norm": 33.0,
+      "learning_rate": 1.3060141509433962e-06,
+      "loss": 2.7431,
+      "step": 2215
+    },
+    {
+      "epoch": 0.0654577620521893,
+      "grad_norm": 18.125,
+      "learning_rate": 1.3089622641509432e-06,
+      "loss": 2.8863,
+      "step": 2220
+    },
+    {
+      "epoch": 0.06560518944419873,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3119103773584905e-06,
+      "loss": 2.8273,
+      "step": 2225
+    },
+    {
+      "epoch": 0.06575261683620817,
+      "grad_norm": 17.5,
+      "learning_rate": 1.3148584905660377e-06,
+      "loss": 2.8624,
+      "step": 2230
+    },
+    {
+      "epoch": 0.0659000442282176,
+      "grad_norm": 21.0,
+      "learning_rate": 1.3178066037735848e-06,
+      "loss": 2.8465,
+      "step": 2235
+    },
+    {
+      "epoch": 0.06604747162022703,
+      "grad_norm": 25.375,
+      "learning_rate": 1.320754716981132e-06,
+      "loss": 2.8271,
+      "step": 2240
+    },
+    {
+      "epoch": 0.06619489901223648,
+      "grad_norm": 16.375,
+      "learning_rate": 1.3237028301886792e-06,
+      "loss": 2.8661,
+      "step": 2245
+    },
+    {
+      "epoch": 0.06634232640424591,
+      "grad_norm": 15.75,
+      "learning_rate": 1.3266509433962263e-06,
+      "loss": 2.8245,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06648975379625534,
+      "grad_norm": 16.25,
+      "learning_rate": 1.3295990566037737e-06,
+      "loss": 2.8016,
+      "step": 2255
+    },
+    {
+      "epoch": 0.06663718118826478,
+      "grad_norm": 85.5,
+      "learning_rate": 1.3325471698113206e-06,
+      "loss": 3.0405,
+      "step": 2260
+    },
+    {
+      "epoch": 0.06678460858027421,
+      "grad_norm": 16.5,
+      "learning_rate": 1.335495283018868e-06,
+      "loss": 2.941,
+      "step": 2265
+    },
+    {
+      "epoch": 0.06693203597228366,
+      "grad_norm": 18.375,
+      "learning_rate": 1.338443396226415e-06,
+      "loss": 2.9167,
+      "step": 2270
+    },
+    {
+      "epoch": 0.06707946336429309,
+      "grad_norm": 28.25,
+      "learning_rate": 1.3413915094339623e-06,
+      "loss": 2.9053,
+      "step": 2275
+    },
+    {
+      "epoch": 0.06722689075630252,
+      "grad_norm": 20.125,
+      "learning_rate": 1.3443396226415094e-06,
+      "loss": 2.9119,
+      "step": 2280
+    },
+    {
+      "epoch": 0.06737431814831196,
+      "grad_norm": 26.75,
+      "learning_rate": 1.3472877358490566e-06,
+      "loss": 2.8339,
+      "step": 2285
+    },
+    {
+      "epoch": 0.06752174554032139,
+      "grad_norm": 16.75,
+      "learning_rate": 1.3502358490566037e-06,
+      "loss": 2.8336,
+      "step": 2290
+    },
+    {
+      "epoch": 0.06766917293233082,
+      "grad_norm": 14.875,
+      "learning_rate": 1.353183962264151e-06,
+      "loss": 2.7363,
+      "step": 2295
+    },
+    {
+      "epoch": 0.06781660032434027,
+      "grad_norm": 15.625,
+      "learning_rate": 1.356132075471698e-06,
+      "loss": 2.8677,
+      "step": 2300
+    },
+    {
+      "epoch": 0.0679640277163497,
+      "grad_norm": 18.125,
+      "learning_rate": 1.3590801886792452e-06,
+      "loss": 2.8287,
+      "step": 2305
+    },
+    {
+      "epoch": 0.06811145510835913,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3620283018867924e-06,
+      "loss": 2.8503,
+      "step": 2310
+    },
+    {
+      "epoch": 0.06825888250036857,
+      "grad_norm": 18.5,
+      "learning_rate": 1.3649764150943395e-06,
+      "loss": 2.8452,
+      "step": 2315
+    },
+    {
+      "epoch": 0.068406309892378,
+      "grad_norm": 20.125,
+      "learning_rate": 1.3679245283018867e-06,
+      "loss": 2.8188,
+      "step": 2320
+    },
+    {
+      "epoch": 0.06855373728438743,
+      "grad_norm": 21.875,
+      "learning_rate": 1.3708726415094338e-06,
+      "loss": 2.7956,
+      "step": 2325
+    },
+    {
+      "epoch": 0.06870116467639688,
+      "grad_norm": 24.875,
+      "learning_rate": 1.3738207547169812e-06,
+      "loss": 2.8185,
+      "step": 2330
+    },
+    {
+      "epoch": 0.06884859206840631,
+      "grad_norm": 18.25,
+      "learning_rate": 1.3767688679245281e-06,
+      "loss": 2.7711,
+      "step": 2335
+    },
+    {
+      "epoch": 0.06899601946041574,
+      "grad_norm": 20.875,
+      "learning_rate": 1.3797169811320755e-06,
+      "loss": 2.9115,
+      "step": 2340
+    },
+    {
+      "epoch": 0.06914344685242518,
+      "grad_norm": 18.125,
+      "learning_rate": 1.3826650943396224e-06,
+      "loss": 2.8137,
+      "step": 2345
+    },
+    {
+      "epoch": 0.06929087424443461,
+      "grad_norm": 25.5,
+      "learning_rate": 1.3856132075471698e-06,
+      "loss": 2.9505,
+      "step": 2350
+    },
+    {
+      "epoch": 0.06943830163644406,
+      "grad_norm": 24.375,
+      "learning_rate": 1.388561320754717e-06,
+      "loss": 2.8544,
+      "step": 2355
+    },
+    {
+      "epoch": 0.06958572902845349,
+      "grad_norm": 24.875,
+      "learning_rate": 1.3915094339622641e-06,
+      "loss": 2.9782,
+      "step": 2360
+    },
+    {
+      "epoch": 0.06973315642046292,
+      "grad_norm": 19.125,
+      "learning_rate": 1.3944575471698113e-06,
+      "loss": 3.0521,
+      "step": 2365
+    },
+    {
+      "epoch": 0.06988058381247236,
+      "grad_norm": 16.625,
+      "learning_rate": 1.3974056603773584e-06,
+      "loss": 2.7751,
+      "step": 2370
+    },
+    {
+      "epoch": 0.0700280112044818,
+      "grad_norm": 21.375,
+      "learning_rate": 1.4003537735849056e-06,
+      "loss": 2.7304,
+      "step": 2375
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 32.25,
+      "learning_rate": 1.403301886792453e-06,
+      "loss": 2.8306,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07032286598850067,
+      "grad_norm": 20.375,
+      "learning_rate": 1.4062499999999999e-06,
+      "loss": 2.9704,
+      "step": 2385
+    },
+    {
+      "epoch": 0.0704702933805101,
+      "grad_norm": 27.875,
+      "learning_rate": 1.409198113207547e-06,
+      "loss": 2.8213,
+      "step": 2390
+    },
+    {
+      "epoch": 0.07061772077251953,
+      "grad_norm": 27.875,
+      "learning_rate": 1.4121462264150942e-06,
+      "loss": 2.8982,
+      "step": 2395
+    },
+    {
+      "epoch": 0.07076514816452897,
+      "grad_norm": 29.375,
+      "learning_rate": 1.4150943396226413e-06,
+      "loss": 2.8757,
+      "step": 2400
+    },
+    {
+      "epoch": 0.0709125755565384,
+      "grad_norm": 18.125,
+      "learning_rate": 1.4180424528301887e-06,
+      "loss": 2.7914,
+      "step": 2405
+    },
+    {
+      "epoch": 0.07106000294854783,
+      "grad_norm": 63.0,
+      "learning_rate": 1.4209905660377356e-06,
+      "loss": 2.7504,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07120743034055728,
+      "grad_norm": 16.0,
+      "learning_rate": 1.423938679245283e-06,
+      "loss": 2.8089,
+      "step": 2415
+    },
+    {
+      "epoch": 0.07135485773256671,
+      "grad_norm": 19.125,
+      "learning_rate": 1.42688679245283e-06,
+      "loss": 2.7972,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07150228512457614,
+      "grad_norm": 15.0,
+      "learning_rate": 1.4298349056603773e-06,
+      "loss": 2.7051,
+      "step": 2425
+    },
+    {
+      "epoch": 0.07164971251658558,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.4327830188679245e-06,
+      "loss": 2.8048,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07179713990859501,
+      "grad_norm": 18.375,
+      "learning_rate": 1.4357311320754716e-06,
+      "loss": 2.7196,
+      "step": 2435
+    },
+    {
+      "epoch": 0.07194456730060446,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4386792452830188e-06,
+      "loss": 2.695,
+      "step": 2440
+    },
+    {
+      "epoch": 0.07209199469261389,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4416273584905661e-06,
+      "loss": 2.9735,
+      "step": 2445
+    },
+    {
+      "epoch": 0.07223942208462332,
+      "grad_norm": 18.25,
+      "learning_rate": 1.444575471698113e-06,
+      "loss": 2.8195,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07238684947663276,
+      "grad_norm": 17.875,
+      "learning_rate": 1.4475235849056605e-06,
+      "loss": 2.7522,
+      "step": 2455
+    },
+    {
+      "epoch": 0.0725342768686422,
+      "grad_norm": 17.125,
+      "learning_rate": 1.4504716981132074e-06,
+      "loss": 2.8065,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07268170426065163,
+      "grad_norm": 15.625,
+      "learning_rate": 1.4534198113207548e-06,
+      "loss": 2.8917,
+      "step": 2465
+    },
+    {
+      "epoch": 0.07282913165266107,
+      "grad_norm": 19.125,
+      "learning_rate": 1.456367924528302e-06,
+      "loss": 2.8464,
+      "step": 2470
+    },
+    {
+      "epoch": 0.0729765590446705,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4593160377358489e-06,
+      "loss": 2.9117,
+      "step": 2475
+    },
+    {
+      "epoch": 0.07312398643667993,
+      "grad_norm": 16.375,
+      "learning_rate": 1.4622641509433962e-06,
+      "loss": 2.6589,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07327141382868937,
+      "grad_norm": 21.75,
+      "learning_rate": 1.4652122641509432e-06,
+      "loss": 2.915,
+      "step": 2485
+    },
+    {
+      "epoch": 0.0734188412206988,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4681603773584905e-06,
+      "loss": 2.7783,
+      "step": 2490
+    },
+    {
+      "epoch": 0.07356626861270824,
+      "grad_norm": 16.875,
+      "learning_rate": 1.4711084905660377e-06,
+      "loss": 2.7156,
+      "step": 2495
+    },
+    {
+      "epoch": 0.07371369600471768,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4740566037735848e-06,
+      "loss": 2.6785,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07371369600471768,
+      "eval_loss": 3.0259971618652344,
+      "eval_runtime": 4.7168,
+      "eval_samples_per_second": 83.956,
+      "eval_steps_per_second": 2.756,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07386112339672711,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.477004716981132e-06,
+      "loss": 2.6697,
+      "step": 2505
+    },
+    {
+      "epoch": 0.07400855078873654,
+      "grad_norm": 19.125,
+      "learning_rate": 1.4799528301886791e-06,
+      "loss": 2.9352,
+      "step": 2510
+    },
+    {
+      "epoch": 0.07415597818074599,
+      "grad_norm": 24.5,
+      "learning_rate": 1.4829009433962263e-06,
+      "loss": 2.8369,
+      "step": 2515
+    },
+    {
+      "epoch": 0.07430340557275542,
+      "grad_norm": 33.25,
+      "learning_rate": 1.4858490566037737e-06,
+      "loss": 2.7892,
+      "step": 2520
+    },
+    {
+      "epoch": 0.07445083296476486,
+      "grad_norm": 23.0,
+      "learning_rate": 1.4887971698113206e-06,
+      "loss": 2.6221,
+      "step": 2525
+    },
+    {
+      "epoch": 0.07459826035677429,
+      "grad_norm": 17.5,
+      "learning_rate": 1.491745283018868e-06,
+      "loss": 2.7536,
+      "step": 2530
+    },
+    {
+      "epoch": 0.07474568774878372,
+      "grad_norm": 20.0,
+      "learning_rate": 1.494693396226415e-06,
+      "loss": 2.8727,
+      "step": 2535
+    },
+    {
+      "epoch": 0.07489311514079317,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.4976415094339623e-06,
+      "loss": 2.6381,
+      "step": 2540
+    },
+    {
+      "epoch": 0.0750405425328026,
+      "grad_norm": 24.625,
+      "learning_rate": 1.5005896226415094e-06,
+      "loss": 2.8,
+      "step": 2545
+    },
+    {
+      "epoch": 0.07518796992481203,
+      "grad_norm": 21.875,
+      "learning_rate": 1.5035377358490566e-06,
+      "loss": 2.6405,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07533539731682147,
+      "grad_norm": 13.75,
+      "learning_rate": 1.5064858490566037e-06,
+      "loss": 2.7266,
+      "step": 2555
+    },
+    {
+      "epoch": 0.0754828247088309,
+      "grad_norm": 15.25,
+      "learning_rate": 1.5094339622641509e-06,
+      "loss": 2.7492,
+      "step": 2560
+    },
+    {
+      "epoch": 0.07563025210084033,
+      "grad_norm": 19.5,
+      "learning_rate": 1.512382075471698e-06,
+      "loss": 2.7432,
+      "step": 2565
+    },
+    {
+      "epoch": 0.07577767949284978,
+      "grad_norm": 21.0,
+      "learning_rate": 1.5153301886792452e-06,
+      "loss": 2.8012,
+      "step": 2570
+    },
+    {
+      "epoch": 0.0759251068848592,
+      "grad_norm": 18.5,
+      "learning_rate": 1.5182783018867923e-06,
+      "loss": 2.8801,
+      "step": 2575
+    },
+    {
+      "epoch": 0.07607253427686864,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.5212264150943395e-06,
+      "loss": 2.6939,
+      "step": 2580
+    },
+    {
+      "epoch": 0.07621996166887808,
+      "grad_norm": 16.375,
+      "learning_rate": 1.5241745283018867e-06,
+      "loss": 2.7358,
+      "step": 2585
+    },
+    {
+      "epoch": 0.07636738906088751,
+      "grad_norm": 15.375,
+      "learning_rate": 1.5271226415094338e-06,
+      "loss": 2.7957,
+      "step": 2590
+    },
+    {
+      "epoch": 0.07651481645289694,
+      "grad_norm": 18.375,
+      "learning_rate": 1.5300707547169812e-06,
+      "loss": 2.8,
+      "step": 2595
+    },
+    {
+      "epoch": 0.07666224384490639,
+      "grad_norm": 20.625,
+      "learning_rate": 1.5330188679245281e-06,
+      "loss": 2.7041,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07680967123691582,
+      "grad_norm": 17.75,
+      "learning_rate": 1.5359669811320755e-06,
+      "loss": 2.8524,
+      "step": 2605
+    },
+    {
+      "epoch": 0.07695709862892526,
+      "grad_norm": 20.625,
+      "learning_rate": 1.5389150943396224e-06,
+      "loss": 2.7402,
+      "step": 2610
+    },
+    {
+      "epoch": 0.07710452602093469,
+      "grad_norm": 21.25,
+      "learning_rate": 1.5418632075471698e-06,
+      "loss": 2.8244,
+      "step": 2615
+    },
+    {
+      "epoch": 0.07725195341294412,
+      "grad_norm": 24.125,
+      "learning_rate": 1.544811320754717e-06,
+      "loss": 2.7226,
+      "step": 2620
+    },
+    {
+      "epoch": 0.07739938080495357,
+      "grad_norm": 19.625,
+      "learning_rate": 1.547759433962264e-06,
+      "loss": 2.637,
+      "step": 2625
+    },
+    {
+      "epoch": 0.077546808196963,
+      "grad_norm": 22.375,
+      "learning_rate": 1.5507075471698112e-06,
+      "loss": 2.6678,
+      "step": 2630
+    },
+    {
+      "epoch": 0.07769423558897243,
+      "grad_norm": 19.875,
+      "learning_rate": 1.5536556603773586e-06,
+      "loss": 2.5859,
+      "step": 2635
+    },
+    {
+      "epoch": 0.07784166298098187,
+      "grad_norm": 19.75,
+      "learning_rate": 1.5566037735849056e-06,
+      "loss": 2.5884,
+      "step": 2640
+    },
+    {
+      "epoch": 0.0779890903729913,
+      "grad_norm": 15.25,
+      "learning_rate": 1.559551886792453e-06,
+      "loss": 2.8667,
+      "step": 2645
+    },
+    {
+      "epoch": 0.07813651776500073,
+      "grad_norm": 17.0,
+      "learning_rate": 1.5624999999999999e-06,
+      "loss": 2.6761,
+      "step": 2650
+    },
+    {
+      "epoch": 0.07828394515701018,
+      "grad_norm": 21.75,
+      "learning_rate": 1.565448113207547e-06,
+      "loss": 2.6761,
+      "step": 2655
+    },
+    {
+      "epoch": 0.0784313725490196,
+      "grad_norm": 32.25,
+      "learning_rate": 1.5683962264150944e-06,
+      "loss": 2.7592,
+      "step": 2660
+    },
+    {
+      "epoch": 0.07857879994102904,
+      "grad_norm": 43.0,
+      "learning_rate": 1.5713443396226413e-06,
+      "loss": 2.7657,
+      "step": 2665
+    },
+    {
+      "epoch": 0.07872622733303848,
+      "grad_norm": 22.875,
+      "learning_rate": 1.5742924528301887e-06,
+      "loss": 2.8494,
+      "step": 2670
+    },
+    {
+      "epoch": 0.07887365472504791,
+      "grad_norm": 18.25,
+      "learning_rate": 1.5772405660377356e-06,
+      "loss": 2.7864,
+      "step": 2675
+    },
+    {
+      "epoch": 0.07902108211705734,
+      "grad_norm": 24.5,
+      "learning_rate": 1.580188679245283e-06,
+      "loss": 2.598,
+      "step": 2680
+    },
+    {
+      "epoch": 0.07916850950906679,
+      "grad_norm": 27.25,
+      "learning_rate": 1.5831367924528301e-06,
+      "loss": 2.751,
+      "step": 2685
+    },
+    {
+      "epoch": 0.07931593690107622,
+      "grad_norm": 17.625,
+      "learning_rate": 1.5860849056603773e-06,
+      "loss": 2.7719,
+      "step": 2690
+    },
+    {
+      "epoch": 0.07946336429308566,
+      "grad_norm": 20.75,
+      "learning_rate": 1.5890330188679245e-06,
+      "loss": 2.6851,
+      "step": 2695
+    },
+    {
+      "epoch": 0.07961079168509509,
+      "grad_norm": 16.25,
+      "learning_rate": 1.5919811320754716e-06,
+      "loss": 2.7268,
+      "step": 2700
+    },
+    {
+      "epoch": 0.07975821907710452,
+      "grad_norm": 20.75,
+      "learning_rate": 1.5949292452830188e-06,
+      "loss": 2.7603,
+      "step": 2705
+    },
+    {
+      "epoch": 0.07990564646911397,
+      "grad_norm": 17.875,
+      "learning_rate": 1.5978773584905661e-06,
+      "loss": 2.8362,
+      "step": 2710
+    },
+    {
+      "epoch": 0.0800530738611234,
+      "grad_norm": 20.375,
+      "learning_rate": 1.600825471698113e-06,
+      "loss": 2.7863,
+      "step": 2715
+    },
+    {
+      "epoch": 0.08020050125313283,
+      "grad_norm": 16.5,
+      "learning_rate": 1.6037735849056604e-06,
+      "loss": 2.6775,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08034792864514227,
+      "grad_norm": 20.625,
+      "learning_rate": 1.6067216981132074e-06,
+      "loss": 2.7235,
+      "step": 2725
+    },
+    {
+      "epoch": 0.0804953560371517,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.6096698113207547e-06,
+      "loss": 2.655,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08064278342916113,
+      "grad_norm": 23.625,
+      "learning_rate": 1.612617924528302e-06,
+      "loss": 2.8043,
+      "step": 2735
+    },
+    {
+      "epoch": 0.08079021082117058,
+      "grad_norm": 23.375,
+      "learning_rate": 1.6155660377358488e-06,
+      "loss": 2.6753,
+      "step": 2740
+    },
+    {
+      "epoch": 0.08093763821318001,
+      "grad_norm": 16.75,
+      "learning_rate": 1.6185141509433962e-06,
+      "loss": 2.7216,
+      "step": 2745
+    },
+    {
+      "epoch": 0.08108506560518944,
+      "grad_norm": 29.75,
+      "learning_rate": 1.6214622641509431e-06,
+      "loss": 2.893,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08123249299719888,
+      "grad_norm": 18.625,
+      "learning_rate": 1.6244103773584905e-06,
+      "loss": 2.7302,
+      "step": 2755
+    },
+    {
+      "epoch": 0.08137992038920831,
+      "grad_norm": 17.0,
+      "learning_rate": 1.6273584905660377e-06,
+      "loss": 2.7501,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08152734778121776,
+      "grad_norm": 19.5,
+      "learning_rate": 1.6303066037735848e-06,
+      "loss": 2.7171,
+      "step": 2765
+    },
+    {
+      "epoch": 0.08167477517322719,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.633254716981132e-06,
+      "loss": 2.7143,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08182220256523662,
+      "grad_norm": 21.625,
+      "learning_rate": 1.6362028301886791e-06,
+      "loss": 2.632,
+      "step": 2775
+    },
+    {
+      "epoch": 0.08196962995724606,
+      "grad_norm": 17.25,
+      "learning_rate": 1.6391509433962263e-06,
+      "loss": 2.7041,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08211705734925549,
+      "grad_norm": 27.25,
+      "learning_rate": 1.6420990566037736e-06,
+      "loss": 2.8002,
+      "step": 2785
+    },
+    {
+      "epoch": 0.08226448474126492,
+      "grad_norm": 31.25,
+      "learning_rate": 1.6450471698113206e-06,
+      "loss": 2.5959,
+      "step": 2790
+    },
+    {
+      "epoch": 0.08241191213327437,
+      "grad_norm": 19.875,
+      "learning_rate": 1.647995283018868e-06,
+      "loss": 2.7606,
+      "step": 2795
+    },
+    {
+      "epoch": 0.0825593395252838,
+      "grad_norm": 18.5,
+      "learning_rate": 1.6509433962264149e-06,
+      "loss": 3.0455,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08270676691729323,
+      "grad_norm": 22.875,
+      "learning_rate": 1.6538915094339623e-06,
+      "loss": 2.7094,
+      "step": 2805
+    },
+    {
+      "epoch": 0.08285419430930267,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6568396226415094e-06,
+      "loss": 2.4643,
+      "step": 2810
+    },
+    {
+      "epoch": 0.0830016217013121,
+      "grad_norm": 22.0,
+      "learning_rate": 1.6597877358490566e-06,
+      "loss": 2.6835,
+      "step": 2815
+    },
+    {
+      "epoch": 0.08314904909332153,
+      "grad_norm": 18.125,
+      "learning_rate": 1.6627358490566037e-06,
+      "loss": 2.6437,
+      "step": 2820
+    },
+    {
+      "epoch": 0.08329647648533098,
+      "grad_norm": 22.625,
+      "learning_rate": 1.6656839622641509e-06,
+      "loss": 2.6712,
+      "step": 2825
+    },
+    {
+      "epoch": 0.08344390387734041,
+      "grad_norm": 17.0,
+      "learning_rate": 1.668632075471698e-06,
+      "loss": 2.7535,
+      "step": 2830
+    },
+    {
+      "epoch": 0.08359133126934984,
+      "grad_norm": 24.375,
+      "learning_rate": 1.6715801886792452e-06,
+      "loss": 2.6405,
+      "step": 2835
+    },
+    {
+      "epoch": 0.08373875866135928,
+      "grad_norm": 16.875,
+      "learning_rate": 1.6745283018867923e-06,
+      "loss": 2.5578,
+      "step": 2840
+    },
+    {
+      "epoch": 0.08388618605336871,
+      "grad_norm": 17.375,
+      "learning_rate": 1.6774764150943395e-06,
+      "loss": 2.662,
+      "step": 2845
+    },
+    {
+      "epoch": 0.08403361344537816,
+      "grad_norm": 18.0,
+      "learning_rate": 1.6804245283018866e-06,
+      "loss": 2.5877,
+      "step": 2850
+    },
+    {
+      "epoch": 0.08418104083738759,
+      "grad_norm": 22.75,
+      "learning_rate": 1.6833726415094338e-06,
+      "loss": 2.7346,
+      "step": 2855
+    },
+    {
+      "epoch": 0.08432846822939702,
+      "grad_norm": 25.0,
+      "learning_rate": 1.6863207547169812e-06,
+      "loss": 2.7402,
+      "step": 2860
+    },
+    {
+      "epoch": 0.08447589562140646,
+      "grad_norm": 19.0,
+      "learning_rate": 1.689268867924528e-06,
+      "loss": 2.6026,
+      "step": 2865
+    },
+    {
+      "epoch": 0.0846233230134159,
+      "grad_norm": 23.625,
+      "learning_rate": 1.6922169811320755e-06,
+      "loss": 2.6259,
+      "step": 2870
+    },
+    {
+      "epoch": 0.08477075040542532,
+      "grad_norm": 16.625,
+      "learning_rate": 1.6951650943396224e-06,
+      "loss": 2.67,
+      "step": 2875
+    },
+    {
+      "epoch": 0.08491817779743477,
+      "grad_norm": 22.875,
+      "learning_rate": 1.6981132075471698e-06,
+      "loss": 2.4129,
+      "step": 2880
+    },
+    {
+      "epoch": 0.0850656051894442,
+      "grad_norm": 129.0,
+      "learning_rate": 1.701061320754717e-06,
+      "loss": 2.8064,
+      "step": 2885
+    },
+    {
+      "epoch": 0.08521303258145363,
+      "grad_norm": 13.625,
+      "learning_rate": 1.704009433962264e-06,
+      "loss": 2.6248,
+      "step": 2890
+    },
+    {
+      "epoch": 0.08536045997346307,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7069575471698112e-06,
+      "loss": 2.6568,
+      "step": 2895
+    },
+    {
+      "epoch": 0.0855078873654725,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7099056603773586e-06,
+      "loss": 2.7321,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08565531475748193,
+      "grad_norm": 34.5,
+      "learning_rate": 1.7128537735849055e-06,
+      "loss": 2.6696,
+      "step": 2905
+    },
+    {
+      "epoch": 0.08580274214949138,
+      "grad_norm": 18.375,
+      "learning_rate": 1.715801886792453e-06,
+      "loss": 2.7116,
+      "step": 2910
+    },
+    {
+      "epoch": 0.08595016954150081,
+      "grad_norm": 16.125,
+      "learning_rate": 1.7187499999999998e-06,
+      "loss": 2.66,
+      "step": 2915
+    },
+    {
+      "epoch": 0.08609759693351024,
+      "grad_norm": 18.25,
+      "learning_rate": 1.721698113207547e-06,
+      "loss": 2.7062,
+      "step": 2920
+    },
+    {
+      "epoch": 0.08624502432551968,
+      "grad_norm": 19.75,
+      "learning_rate": 1.7246462264150944e-06,
+      "loss": 2.667,
+      "step": 2925
+    },
+    {
+      "epoch": 0.08639245171752911,
+      "grad_norm": 13.125,
+      "learning_rate": 1.7275943396226413e-06,
+      "loss": 2.6131,
+      "step": 2930
+    },
+    {
+      "epoch": 0.08653987910953856,
+      "grad_norm": 20.25,
+      "learning_rate": 1.7305424528301887e-06,
+      "loss": 2.6455,
+      "step": 2935
+    },
+    {
+      "epoch": 0.08668730650154799,
+      "grad_norm": 18.625,
+      "learning_rate": 1.7334905660377356e-06,
+      "loss": 2.6537,
+      "step": 2940
+    },
+    {
+      "epoch": 0.08683473389355742,
+      "grad_norm": 16.75,
+      "learning_rate": 1.736438679245283e-06,
+      "loss": 2.6189,
+      "step": 2945
+    },
+    {
+      "epoch": 0.08698216128556686,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7393867924528301e-06,
+      "loss": 2.6757,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0871295886775763,
+      "grad_norm": 20.5,
+      "learning_rate": 1.7423349056603773e-06,
+      "loss": 2.613,
+      "step": 2955
+    },
+    {
+      "epoch": 0.08727701606958572,
+      "grad_norm": 23.125,
+      "learning_rate": 1.7452830188679244e-06,
+      "loss": 2.6377,
+      "step": 2960
+    },
+    {
+      "epoch": 0.08742444346159517,
+      "grad_norm": 23.25,
+      "learning_rate": 1.7482311320754716e-06,
+      "loss": 2.7463,
+      "step": 2965
+    },
+    {
+      "epoch": 0.0875718708536046,
+      "grad_norm": 69.5,
+      "learning_rate": 1.7511792452830188e-06,
+      "loss": 2.7704,
+      "step": 2970
+    },
+    {
+      "epoch": 0.08771929824561403,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.7541273584905661e-06,
+      "loss": 2.593,
+      "step": 2975
+    },
+    {
+      "epoch": 0.08786672563762347,
+      "grad_norm": 53.25,
+      "learning_rate": 1.757075471698113e-06,
+      "loss": 2.5997,
+      "step": 2980
+    },
+    {
+      "epoch": 0.0880141530296329,
+      "grad_norm": 33.75,
+      "learning_rate": 1.7600235849056604e-06,
+      "loss": 2.7404,
+      "step": 2985
+    },
+    {
+      "epoch": 0.08816158042164234,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7629716981132074e-06,
+      "loss": 2.5596,
+      "step": 2990
+    },
+    {
+      "epoch": 0.08830900781365178,
+      "grad_norm": 24.625,
+      "learning_rate": 1.7659198113207547e-06,
+      "loss": 2.6808,
+      "step": 2995
+    },
+    {
+      "epoch": 0.08845643520566121,
+      "grad_norm": 20.75,
+      "learning_rate": 1.7688679245283019e-06,
+      "loss": 2.534,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08845643520566121,
+      "eval_loss": 2.8255984783172607,
+      "eval_runtime": 4.7091,
+      "eval_samples_per_second": 84.093,
+      "eval_steps_per_second": 2.761,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08860386259767064,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7718160377358488e-06,
+      "loss": 2.6962,
+      "step": 3005
+    },
+    {
+      "epoch": 0.08875128998968008,
+      "grad_norm": 19.125,
+      "learning_rate": 1.7747641509433962e-06,
+      "loss": 2.5769,
+      "step": 3010
+    },
+    {
+      "epoch": 0.08889871738168952,
+      "grad_norm": 23.0,
+      "learning_rate": 1.7777122641509431e-06,
+      "loss": 2.5602,
+      "step": 3015
+    },
+    {
+      "epoch": 0.08904614477369896,
+      "grad_norm": 16.0,
+      "learning_rate": 1.7806603773584905e-06,
+      "loss": 2.5782,
+      "step": 3020
+    },
+    {
+      "epoch": 0.08919357216570839,
+      "grad_norm": 19.375,
+      "learning_rate": 1.7836084905660377e-06,
+      "loss": 2.6285,
+      "step": 3025
+    },
+    {
+      "epoch": 0.08934099955771782,
+      "grad_norm": 21.25,
+      "learning_rate": 1.7865566037735848e-06,
+      "loss": 2.6416,
+      "step": 3030
+    },
+    {
+      "epoch": 0.08948842694972726,
+      "grad_norm": 19.5,
+      "learning_rate": 1.789504716981132e-06,
+      "loss": 2.5563,
+      "step": 3035
+    },
+    {
+      "epoch": 0.0896358543417367,
+      "grad_norm": 19.5,
+      "learning_rate": 1.7924528301886791e-06,
+      "loss": 2.6188,
+      "step": 3040
+    },
+    {
+      "epoch": 0.08978328173374613,
+      "grad_norm": 15.75,
+      "learning_rate": 1.7954009433962263e-06,
+      "loss": 2.6018,
+      "step": 3045
+    },
+    {
+      "epoch": 0.08993070912575557,
+      "grad_norm": 19.625,
+      "learning_rate": 1.7983490566037736e-06,
+      "loss": 2.6601,
+      "step": 3050
+    },
+    {
+      "epoch": 0.090078136517765,
+      "grad_norm": 22.375,
+      "learning_rate": 1.8012971698113206e-06,
+      "loss": 2.6206,
+      "step": 3055
+    },
+    {
+      "epoch": 0.09022556390977443,
+      "grad_norm": 25.625,
+      "learning_rate": 1.804245283018868e-06,
+      "loss": 2.7032,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09037299130178388,
+      "grad_norm": 16.75,
+      "learning_rate": 1.8071933962264149e-06,
+      "loss": 2.6075,
+      "step": 3065
+    },
+    {
+      "epoch": 0.0905204186937933,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8101415094339622e-06,
+      "loss": 2.7258,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09066784608580274,
+      "grad_norm": 23.375,
+      "learning_rate": 1.8130896226415094e-06,
+      "loss": 2.7048,
+      "step": 3075
+    },
+    {
+      "epoch": 0.09081527347781218,
+      "grad_norm": 19.75,
+      "learning_rate": 1.8160377358490566e-06,
+      "loss": 2.4501,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09096270086982161,
+      "grad_norm": 19.125,
+      "learning_rate": 1.8189858490566037e-06,
+      "loss": 2.5144,
+      "step": 3085
+    },
+    {
+      "epoch": 0.09111012826183104,
+      "grad_norm": 17.375,
+      "learning_rate": 1.821933962264151e-06,
+      "loss": 2.7287,
+      "step": 3090
+    },
+    {
+      "epoch": 0.09125755565384049,
+      "grad_norm": 15.125,
+      "learning_rate": 1.824882075471698e-06,
+      "loss": 2.6906,
+      "step": 3095
+    },
+    {
+      "epoch": 0.09140498304584992,
+      "grad_norm": 16.875,
+      "learning_rate": 1.8278301886792452e-06,
+      "loss": 2.6328,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09155241043785936,
+      "grad_norm": 28.375,
+      "learning_rate": 1.8307783018867923e-06,
+      "loss": 2.7216,
+      "step": 3105
+    },
+    {
+      "epoch": 0.09169983782986879,
+      "grad_norm": 20.375,
+      "learning_rate": 1.8337264150943395e-06,
+      "loss": 2.6709,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09184726522187822,
+      "grad_norm": 17.625,
+      "learning_rate": 1.8366745283018868e-06,
+      "loss": 2.6247,
+      "step": 3115
+    },
+    {
+      "epoch": 0.09199469261388767,
+      "grad_norm": 14.625,
+      "learning_rate": 1.8396226415094338e-06,
+      "loss": 2.704,
+      "step": 3120
+    },
+    {
+      "epoch": 0.0921421200058971,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8425707547169811e-06,
+      "loss": 2.6575,
+      "step": 3125
+    },
+    {
+      "epoch": 0.09228954739790653,
+      "grad_norm": 16.75,
+      "learning_rate": 1.845518867924528e-06,
+      "loss": 2.6053,
+      "step": 3130
+    },
+    {
+      "epoch": 0.09243697478991597,
+      "grad_norm": 18.875,
+      "learning_rate": 1.8484669811320755e-06,
+      "loss": 2.6402,
+      "step": 3135
+    },
+    {
+      "epoch": 0.0925844021819254,
+      "grad_norm": 24.375,
+      "learning_rate": 1.8514150943396226e-06,
+      "loss": 2.6393,
+      "step": 3140
+    },
+    {
+      "epoch": 0.09273182957393483,
+      "grad_norm": 23.125,
+      "learning_rate": 1.8543632075471698e-06,
+      "loss": 2.7078,
+      "step": 3145
+    },
+    {
+      "epoch": 0.09287925696594428,
+      "grad_norm": 18.5,
+      "learning_rate": 1.857311320754717e-06,
+      "loss": 2.6088,
+      "step": 3150
+    },
+    {
+      "epoch": 0.0930266843579537,
+      "grad_norm": 17.0,
+      "learning_rate": 1.860259433962264e-06,
+      "loss": 2.6411,
+      "step": 3155
+    },
+    {
+      "epoch": 0.09317411174996314,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8632075471698112e-06,
+      "loss": 2.5609,
+      "step": 3160
+    },
+    {
+      "epoch": 0.09332153914197258,
+      "grad_norm": 19.0,
+      "learning_rate": 1.8661556603773586e-06,
+      "loss": 2.7937,
+      "step": 3165
+    },
+    {
+      "epoch": 0.09346896653398201,
+      "grad_norm": 17.75,
+      "learning_rate": 1.8691037735849055e-06,
+      "loss": 2.5774,
+      "step": 3170
+    },
+    {
+      "epoch": 0.09361639392599144,
+      "grad_norm": 29.125,
+      "learning_rate": 1.8720518867924529e-06,
+      "loss": 2.6118,
+      "step": 3175
+    },
+    {
+      "epoch": 0.09376382131800089,
+      "grad_norm": 34.0,
+      "learning_rate": 1.8749999999999998e-06,
+      "loss": 2.7062,
+      "step": 3180
+    },
+    {
+      "epoch": 0.09391124871001032,
+      "grad_norm": 21.375,
+      "learning_rate": 1.877948113207547e-06,
+      "loss": 2.4245,
+      "step": 3185
+    },
+    {
+      "epoch": 0.09405867610201976,
+      "grad_norm": 21.0,
+      "learning_rate": 1.8808962264150944e-06,
+      "loss": 2.575,
+      "step": 3190
+    },
+    {
+      "epoch": 0.09420610349402919,
+      "grad_norm": 19.75,
+      "learning_rate": 1.8838443396226413e-06,
+      "loss": 2.6233,
+      "step": 3195
+    },
+    {
+      "epoch": 0.09435353088603862,
+      "grad_norm": 17.0,
+      "learning_rate": 1.8867924528301887e-06,
+      "loss": 2.63,
+      "step": 3200
+    },
+    {
+      "epoch": 0.09450095827804807,
+      "grad_norm": 15.5,
+      "learning_rate": 1.8897405660377356e-06,
+      "loss": 2.4527,
+      "step": 3205
+    },
+    {
+      "epoch": 0.0946483856700575,
+      "grad_norm": 28.125,
+      "learning_rate": 1.892688679245283e-06,
+      "loss": 2.6324,
+      "step": 3210
+    },
+    {
+      "epoch": 0.09479581306206693,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8956367924528301e-06,
+      "loss": 2.5413,
+      "step": 3215
+    },
+    {
+      "epoch": 0.09494324045407637,
+      "grad_norm": 20.625,
+      "learning_rate": 1.8985849056603773e-06,
+      "loss": 2.7314,
+      "step": 3220
+    },
+    {
+      "epoch": 0.0950906678460858,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9015330188679244e-06,
+      "loss": 2.5598,
+      "step": 3225
+    },
+    {
+      "epoch": 0.09523809523809523,
+      "grad_norm": 25.75,
+      "learning_rate": 1.9044811320754716e-06,
+      "loss": 2.5823,
+      "step": 3230
+    },
+    {
+      "epoch": 0.09538552263010468,
+      "grad_norm": 14.375,
+      "learning_rate": 1.907429245283019e-06,
+      "loss": 2.6562,
+      "step": 3235
+    },
+    {
+      "epoch": 0.09553295002211411,
+      "grad_norm": 24.75,
+      "learning_rate": 1.910377358490566e-06,
+      "loss": 2.6288,
+      "step": 3240
+    },
+    {
+      "epoch": 0.09568037741412354,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9133254716981133e-06,
+      "loss": 2.5194,
+      "step": 3245
+    },
+    {
+      "epoch": 0.09582780480613298,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9162735849056604e-06,
+      "loss": 2.5139,
+      "step": 3250
+    },
+    {
+      "epoch": 0.09597523219814241,
+      "grad_norm": 20.375,
+      "learning_rate": 1.9192216981132076e-06,
+      "loss": 2.6683,
+      "step": 3255
+    },
+    {
+      "epoch": 0.09612265959015186,
+      "grad_norm": 24.375,
+      "learning_rate": 1.9221698113207547e-06,
+      "loss": 2.4635,
+      "step": 3260
+    },
+    {
+      "epoch": 0.09627008698216129,
+      "grad_norm": 29.75,
+      "learning_rate": 1.925117924528302e-06,
+      "loss": 2.6588,
+      "step": 3265
+    },
+    {
+      "epoch": 0.09641751437417072,
+      "grad_norm": 15.375,
+      "learning_rate": 1.928066037735849e-06,
+      "loss": 2.4663,
+      "step": 3270
+    },
+    {
+      "epoch": 0.09656494176618016,
+      "grad_norm": 17.25,
+      "learning_rate": 1.931014150943396e-06,
+      "loss": 2.5586,
+      "step": 3275
+    },
+    {
+      "epoch": 0.09671236915818959,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9339622641509433e-06,
+      "loss": 2.4494,
+      "step": 3280
+    },
+    {
+      "epoch": 0.09685979655019902,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9369103773584905e-06,
+      "loss": 2.513,
+      "step": 3285
+    },
+    {
+      "epoch": 0.09700722394220847,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9398584905660376e-06,
+      "loss": 2.5481,
+      "step": 3290
+    },
+    {
+      "epoch": 0.0971546513342179,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.942806603773585e-06,
+      "loss": 2.5519,
+      "step": 3295
+    },
+    {
+      "epoch": 0.09730207872622733,
+      "grad_norm": 56.5,
+      "learning_rate": 1.945754716981132e-06,
+      "loss": 2.4955,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09744950611823677,
+      "grad_norm": 20.25,
+      "learning_rate": 1.948702830188679e-06,
+      "loss": 2.5694,
+      "step": 3305
+    },
+    {
+      "epoch": 0.0975969335102462,
+      "grad_norm": 29.875,
+      "learning_rate": 1.9516509433962263e-06,
+      "loss": 2.5207,
+      "step": 3310
+    },
+    {
+      "epoch": 0.09774436090225563,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9545990566037734e-06,
+      "loss": 2.5684,
+      "step": 3315
+    },
+    {
+      "epoch": 0.09789178829426508,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9575471698113206e-06,
+      "loss": 2.5669,
+      "step": 3320
+    },
+    {
+      "epoch": 0.09803921568627451,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9604952830188677e-06,
+      "loss": 2.4171,
+      "step": 3325
+    },
+    {
+      "epoch": 0.09818664307828394,
+      "grad_norm": 77.5,
+      "learning_rate": 1.963443396226415e-06,
+      "loss": 2.6574,
+      "step": 3330
+    },
+    {
+      "epoch": 0.09833407047029338,
+      "grad_norm": 84.5,
+      "learning_rate": 1.9663915094339624e-06,
+      "loss": 2.574,
+      "step": 3335
+    },
+    {
+      "epoch": 0.09848149786230281,
+      "grad_norm": 28.125,
+      "learning_rate": 1.969339622641509e-06,
+      "loss": 2.6212,
+      "step": 3340
+    },
+    {
+      "epoch": 0.09862892525431226,
+      "grad_norm": 30.375,
+      "learning_rate": 1.9722877358490568e-06,
+      "loss": 2.421,
+      "step": 3345
+    },
+    {
+      "epoch": 0.09877635264632169,
+      "grad_norm": 22.875,
+      "learning_rate": 1.9752358490566035e-06,
+      "loss": 2.5433,
+      "step": 3350
+    },
+    {
+      "epoch": 0.09892378003833112,
+      "grad_norm": 24.0,
+      "learning_rate": 1.978183962264151e-06,
+      "loss": 2.733,
+      "step": 3355
+    },
+    {
+      "epoch": 0.09907120743034056,
+      "grad_norm": 16.875,
+      "learning_rate": 1.981132075471698e-06,
+      "loss": 2.599,
+      "step": 3360
+    },
+    {
+      "epoch": 0.09921863482235,
+      "grad_norm": 16.625,
+      "learning_rate": 1.984080188679245e-06,
+      "loss": 2.5051,
+      "step": 3365
+    },
+    {
+      "epoch": 0.09936606221435942,
+      "grad_norm": 22.875,
+      "learning_rate": 1.9870283018867925e-06,
+      "loss": 2.4694,
+      "step": 3370
+    },
+    {
+      "epoch": 0.09951348960636887,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9899764150943392e-06,
+      "loss": 2.5643,
+      "step": 3375
+    },
+    {
+      "epoch": 0.0996609169983783,
+      "grad_norm": 19.75,
+      "learning_rate": 1.992924528301887e-06,
+      "loss": 2.4453,
+      "step": 3380
+    },
+    {
+      "epoch": 0.09980834439038773,
+      "grad_norm": 19.875,
+      "learning_rate": 1.995872641509434e-06,
+      "loss": 2.4782,
+      "step": 3385
+    },
+    {
+      "epoch": 0.09995577178239717,
+      "grad_norm": 14.125,
+      "learning_rate": 1.998820754716981e-06,
+      "loss": 2.5011,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1001031991744066,
+      "grad_norm": 16.625,
+      "learning_rate": 1.999999952328609e-06,
+      "loss": 2.6024,
+      "step": 3395
+    },
+    {
+      "epoch": 0.10025062656641603,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9999996610034596e-06,
+      "loss": 2.5227,
+      "step": 3400
+    },
+    {
+      "epoch": 0.10039805395842548,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9999991048373438e-06,
+      "loss": 2.4102,
+      "step": 3405
+    },
+    {
+      "epoch": 0.10054548135043491,
+      "grad_norm": 22.375,
+      "learning_rate": 1.9999982838304092e-06,
+      "loss": 2.5443,
+      "step": 3410
+    },
+    {
+      "epoch": 0.10069290874244434,
+      "grad_norm": 19.25,
+      "learning_rate": 1.9999971979828727e-06,
+      "loss": 2.5334,
+      "step": 3415
+    },
+    {
+      "epoch": 0.10084033613445378,
+      "grad_norm": 18.5,
+      "learning_rate": 1.999995847295022e-06,
+      "loss": 2.6654,
+      "step": 3420
+    },
+    {
+      "epoch": 0.10098776352646321,
+      "grad_norm": 18.0,
+      "learning_rate": 1.999994231767215e-06,
+      "loss": 2.5571,
+      "step": 3425
+    },
+    {
+      "epoch": 0.10113519091847266,
+      "grad_norm": 13.5,
+      "learning_rate": 1.99999235139988e-06,
+      "loss": 2.5254,
+      "step": 3430
+    },
+    {
+      "epoch": 0.10128261831048209,
+      "grad_norm": 27.375,
+      "learning_rate": 1.999990206193514e-06,
+      "loss": 2.4695,
+      "step": 3435
+    },
+    {
+      "epoch": 0.10143004570249152,
+      "grad_norm": 22.0,
+      "learning_rate": 1.999987796148686e-06,
+      "loss": 2.4044,
+      "step": 3440
+    },
+    {
+      "epoch": 0.10157747309450096,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9999851212660336e-06,
+      "loss": 2.4752,
+      "step": 3445
+    },
+    {
+      "epoch": 0.1017249004865104,
+      "grad_norm": 18.75,
+      "learning_rate": 1.9999821815462655e-06,
+      "loss": 2.6967,
+      "step": 3450
+    },
+    {
+      "epoch": 0.10187232787851982,
+      "grad_norm": 25.0,
+      "learning_rate": 1.9999789769901606e-06,
+      "loss": 2.5363,
+      "step": 3455
+    },
+    {
+      "epoch": 0.10201975527052927,
+      "grad_norm": 26.375,
+      "learning_rate": 1.9999755075985674e-06,
+      "loss": 2.5621,
+      "step": 3460
+    },
+    {
+      "epoch": 0.1021671826625387,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9999717733724043e-06,
+      "loss": 2.5803,
+      "step": 3465
+    },
+    {
+      "epoch": 0.10231461005454813,
+      "grad_norm": 22.625,
+      "learning_rate": 1.9999677743126607e-06,
+      "loss": 2.6209,
+      "step": 3470
+    },
+    {
+      "epoch": 0.10246203744655757,
+      "grad_norm": 17.625,
+      "learning_rate": 1.999963510420396e-06,
+      "loss": 2.4312,
+      "step": 3475
+    },
+    {
+      "epoch": 0.102609464838567,
+      "grad_norm": 41.0,
+      "learning_rate": 1.999958981696739e-06,
+      "loss": 2.5988,
+      "step": 3480
+    },
+    {
+      "epoch": 0.10275689223057644,
+      "grad_norm": 21.375,
+      "learning_rate": 1.999954188142889e-06,
+      "loss": 2.5389,
+      "step": 3485
+    },
+    {
+      "epoch": 0.10290431962258588,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9999491297601154e-06,
+      "loss": 2.5496,
+      "step": 3490
+    },
+    {
+      "epoch": 0.10305174701459531,
+      "grad_norm": 29.0,
+      "learning_rate": 1.9999438065497587e-06,
+      "loss": 2.5188,
+      "step": 3495
+    },
+    {
+      "epoch": 0.10319917440660474,
+      "grad_norm": 18.125,
+      "learning_rate": 1.999938218513228e-06,
+      "loss": 2.4835,
+      "step": 3500
+    },
+    {
+      "epoch": 0.10319917440660474,
+      "eval_loss": 2.641951322555542,
+      "eval_runtime": 4.7225,
+      "eval_samples_per_second": 83.855,
+      "eval_steps_per_second": 2.753,
+      "step": 3500
+    },
+    {
+      "epoch": 0.10334660179861418,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9999323656520037e-06,
+      "loss": 2.4558,
+      "step": 3505
+    },
+    {
+      "epoch": 0.10349402919062362,
+      "grad_norm": 17.375,
+      "learning_rate": 1.999926247967635e-06,
+      "loss": 2.4677,
+      "step": 3510
+    },
+    {
+      "epoch": 0.10364145658263306,
+      "grad_norm": 18.0,
+      "learning_rate": 1.999919865461743e-06,
+      "loss": 2.4587,
+      "step": 3515
+    },
+    {
+      "epoch": 0.10378888397464249,
+      "grad_norm": 20.875,
+      "learning_rate": 1.999913218136018e-06,
+      "loss": 2.5211,
+      "step": 3520
+    },
+    {
+      "epoch": 0.10393631136665192,
+      "grad_norm": 16.25,
+      "learning_rate": 1.99990630599222e-06,
+      "loss": 2.552,
+      "step": 3525
+    },
+    {
+      "epoch": 0.10408373875866136,
+      "grad_norm": 16.875,
+      "learning_rate": 1.99989912903218e-06,
+      "loss": 2.3828,
+      "step": 3530
+    },
+    {
+      "epoch": 0.1042311661506708,
+      "grad_norm": 19.625,
+      "learning_rate": 1.999891687257799e-06,
+      "loss": 2.4554,
+      "step": 3535
+    },
+    {
+      "epoch": 0.10437859354268023,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9998839806710466e-06,
+      "loss": 2.5472,
+      "step": 3540
+    },
+    {
+      "epoch": 0.10452602093468967,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9998760092739654e-06,
+      "loss": 2.4378,
+      "step": 3545
+    },
+    {
+      "epoch": 0.1046734483266991,
+      "grad_norm": 18.5,
+      "learning_rate": 1.999867773068666e-06,
+      "loss": 2.4833,
+      "step": 3550
+    },
+    {
+      "epoch": 0.10482087571870853,
+      "grad_norm": 18.5,
+      "learning_rate": 1.999859272057329e-06,
+      "loss": 2.6006,
+      "step": 3555
+    },
+    {
+      "epoch": 0.10496830311071798,
+      "grad_norm": 20.75,
+      "learning_rate": 1.999850506242207e-06,
+      "loss": 2.5066,
+      "step": 3560
+    },
+    {
+      "epoch": 0.1051157305027274,
+      "grad_norm": 26.875,
+      "learning_rate": 1.9998414756256208e-06,
+      "loss": 2.5387,
+      "step": 3565
+    },
+    {
+      "epoch": 0.10526315789473684,
+      "grad_norm": 34.0,
+      "learning_rate": 1.9998321802099614e-06,
+      "loss": 2.5366,
+      "step": 3570
+    },
+    {
+      "epoch": 0.10541058528674628,
+      "grad_norm": 28.375,
+      "learning_rate": 1.9998226199976925e-06,
+      "loss": 2.6194,
+      "step": 3575
+    },
+    {
+      "epoch": 0.10555801267875571,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9998127949913444e-06,
+      "loss": 2.4925,
+      "step": 3580
+    },
+    {
+      "epoch": 0.10570544007076514,
+      "grad_norm": 15.375,
+      "learning_rate": 1.9998027051935198e-06,
+      "loss": 2.4019,
+      "step": 3585
+    },
+    {
+      "epoch": 0.10585286746277459,
+      "grad_norm": 17.25,
+      "learning_rate": 1.999792350606891e-06,
+      "loss": 2.4669,
+      "step": 3590
+    },
+    {
+      "epoch": 0.10600029485478402,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9997817312342e-06,
+      "loss": 2.6052,
+      "step": 3595
+    },
+    {
+      "epoch": 0.10614772224679346,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9997708470782596e-06,
+      "loss": 2.5472,
+      "step": 3600
+    },
+    {
+      "epoch": 0.10629514963880289,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9997596981419517e-06,
+      "loss": 2.4028,
+      "step": 3605
+    },
+    {
+      "epoch": 0.10644257703081232,
+      "grad_norm": 16.5,
+      "learning_rate": 1.99974828442823e-06,
+      "loss": 2.5456,
+      "step": 3610
+    },
+    {
+      "epoch": 0.10659000442282177,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9997366059401166e-06,
+      "loss": 2.5217,
+      "step": 3615
+    },
+    {
+      "epoch": 0.1067374318148312,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9997246626807045e-06,
+      "loss": 2.5225,
+      "step": 3620
+    },
+    {
+      "epoch": 0.10688485920684063,
+      "grad_norm": 18.25,
+      "learning_rate": 1.9997124546531566e-06,
+      "loss": 2.4429,
+      "step": 3625
+    },
+    {
+      "epoch": 0.10703228659885007,
+      "grad_norm": 27.5,
+      "learning_rate": 1.999699981860707e-06,
+      "loss": 2.5007,
+      "step": 3630
+    },
+    {
+      "epoch": 0.1071797139908595,
+      "grad_norm": 17.0,
+      "learning_rate": 1.999687244306658e-06,
+      "loss": 2.4672,
+      "step": 3635
+    },
+    {
+      "epoch": 0.10732714138286893,
+      "grad_norm": 21.125,
+      "learning_rate": 1.9996742419943834e-06,
+      "loss": 2.4403,
+      "step": 3640
+    },
+    {
+      "epoch": 0.10747456877487838,
+      "grad_norm": 20.625,
+      "learning_rate": 1.9996609749273268e-06,
+      "loss": 2.6091,
+      "step": 3645
+    },
+    {
+      "epoch": 0.1076219961668878,
+      "grad_norm": 22.25,
+      "learning_rate": 1.999647443109002e-06,
+      "loss": 2.4657,
+      "step": 3650
+    },
+    {
+      "epoch": 0.10776942355889724,
+      "grad_norm": 22.0,
+      "learning_rate": 1.9996336465429923e-06,
+      "loss": 2.4482,
+      "step": 3655
+    },
+    {
+      "epoch": 0.10791685095090668,
+      "grad_norm": 18.0,
+      "learning_rate": 1.999619585232952e-06,
+      "loss": 2.4894,
+      "step": 3660
+    },
+    {
+      "epoch": 0.10806427834291611,
+      "grad_norm": 13.25,
+      "learning_rate": 1.999605259182605e-06,
+      "loss": 2.4985,
+      "step": 3665
+    },
+    {
+      "epoch": 0.10821170573492554,
+      "grad_norm": 19.375,
+      "learning_rate": 1.999590668395745e-06,
+      "loss": 2.5532,
+      "step": 3670
+    },
+    {
+      "epoch": 0.10835913312693499,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9995758128762376e-06,
+      "loss": 2.5648,
+      "step": 3675
+    },
+    {
+      "epoch": 0.10850656051894442,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.9995606926280157e-06,
+      "loss": 2.4608,
+      "step": 3680
+    },
+    {
+      "epoch": 0.10865398791095386,
+      "grad_norm": 25.875,
+      "learning_rate": 1.999545307655084e-06,
+      "loss": 2.5516,
+      "step": 3685
+    },
+    {
+      "epoch": 0.10880141530296329,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.999529657961518e-06,
+      "loss": 2.5005,
+      "step": 3690
+    },
+    {
+      "epoch": 0.10894884269497272,
+      "grad_norm": 19.125,
+      "learning_rate": 1.999513743551461e-06,
+      "loss": 2.6021,
+      "step": 3695
+    },
+    {
+      "epoch": 0.10909627008698217,
+      "grad_norm": 18.875,
+      "learning_rate": 1.999497564429129e-06,
+      "loss": 2.5128,
+      "step": 3700
+    },
+    {
+      "epoch": 0.1092436974789916,
+      "grad_norm": 19.25,
+      "learning_rate": 1.9994811205988063e-06,
+      "loss": 2.496,
+      "step": 3705
+    },
+    {
+      "epoch": 0.10939112487100103,
+      "grad_norm": 20.5,
+      "learning_rate": 1.999464412064848e-06,
+      "loss": 2.3362,
+      "step": 3710
+    },
+    {
+      "epoch": 0.10953855226301047,
+      "grad_norm": 21.25,
+      "learning_rate": 1.9994474388316794e-06,
+      "loss": 2.5267,
+      "step": 3715
+    },
+    {
+      "epoch": 0.1096859796550199,
+      "grad_norm": 21.5,
+      "learning_rate": 1.9994302009037957e-06,
+      "loss": 2.4618,
+      "step": 3720
+    },
+    {
+      "epoch": 0.10983340704702933,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9994126982857614e-06,
+      "loss": 2.47,
+      "step": 3725
+    },
+    {
+      "epoch": 0.10998083443903878,
+      "grad_norm": 16.0,
+      "learning_rate": 1.999394930982213e-06,
+      "loss": 2.5356,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11012826183104821,
+      "grad_norm": 18.625,
+      "learning_rate": 1.9993768989978558e-06,
+      "loss": 2.4532,
+      "step": 3735
+    },
+    {
+      "epoch": 0.11027568922305764,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9993586023374645e-06,
+      "loss": 2.5099,
+      "step": 3740
+    },
+    {
+      "epoch": 0.11042311661506708,
+      "grad_norm": 26.0,
+      "learning_rate": 1.9993400410058864e-06,
+      "loss": 2.4453,
+      "step": 3745
+    },
+    {
+      "epoch": 0.11057054400707651,
+      "grad_norm": 17.0,
+      "learning_rate": 1.999321215008036e-06,
+      "loss": 2.5786,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11071797139908596,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9993021243488994e-06,
+      "loss": 2.5121,
+      "step": 3755
+    },
+    {
+      "epoch": 0.11086539879109539,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.999282769033533e-06,
+      "loss": 2.5239,
+      "step": 3760
+    },
+    {
+      "epoch": 0.11101282618310482,
+      "grad_norm": 32.5,
+      "learning_rate": 1.9992631490670623e-06,
+      "loss": 2.3592,
+      "step": 3765
+    },
+    {
+      "epoch": 0.11116025357511426,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9992432644546836e-06,
+      "loss": 2.4312,
+      "step": 3770
+    },
+    {
+      "epoch": 0.11130768096712369,
+      "grad_norm": 38.5,
+      "learning_rate": 1.999223115201664e-06,
+      "loss": 2.5293,
+      "step": 3775
+    },
+    {
+      "epoch": 0.11145510835913312,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9992027013133393e-06,
+      "loss": 2.381,
+      "step": 3780
+    },
+    {
+      "epoch": 0.11160253575114257,
+      "grad_norm": 19.125,
+      "learning_rate": 1.999182022795116e-06,
+      "loss": 2.4684,
+      "step": 3785
+    },
+    {
+      "epoch": 0.111749963143152,
+      "grad_norm": 22.625,
+      "learning_rate": 1.9991610796524697e-06,
+      "loss": 2.367,
+      "step": 3790
+    },
+    {
+      "epoch": 0.11189739053516143,
+      "grad_norm": 18.125,
+      "learning_rate": 1.999139871890948e-06,
+      "loss": 2.5411,
+      "step": 3795
+    },
+    {
+      "epoch": 0.11204481792717087,
+      "grad_norm": 20.625,
+      "learning_rate": 1.999118399516168e-06,
+      "loss": 2.4369,
+      "step": 3800
+    },
+    {
+      "epoch": 0.1121922453191803,
+      "grad_norm": 29.75,
+      "learning_rate": 1.9990966625338154e-06,
+      "loss": 2.588,
+      "step": 3805
+    },
+    {
+      "epoch": 0.11233967271118973,
+      "grad_norm": 30.5,
+      "learning_rate": 1.9990746609496476e-06,
+      "loss": 2.4131,
+      "step": 3810
+    },
+    {
+      "epoch": 0.11248710010319918,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.9990523947694917e-06,
+      "loss": 2.3787,
+      "step": 3815
+    },
+    {
+      "epoch": 0.11263452749520861,
+      "grad_norm": 27.125,
+      "learning_rate": 1.999029863999244e-06,
+      "loss": 2.4659,
+      "step": 3820
+    },
+    {
+      "epoch": 0.11278195488721804,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9990070686448725e-06,
+      "loss": 2.4582,
+      "step": 3825
+    },
+    {
+      "epoch": 0.11292938227922748,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9989840087124134e-06,
+      "loss": 2.3936,
+      "step": 3830
+    },
+    {
+      "epoch": 0.11307680967123691,
+      "grad_norm": 20.375,
+      "learning_rate": 1.9989606842079745e-06,
+      "loss": 2.5457,
+      "step": 3835
+    },
+    {
+      "epoch": 0.11322423706324636,
+      "grad_norm": 25.875,
+      "learning_rate": 1.998937095137733e-06,
+      "loss": 2.5145,
+      "step": 3840
+    },
+    {
+      "epoch": 0.11337166445525579,
+      "grad_norm": 20.0,
+      "learning_rate": 1.998913241507936e-06,
+      "loss": 2.5409,
+      "step": 3845
+    },
+    {
+      "epoch": 0.11351909184726522,
+      "grad_norm": 16.875,
+      "learning_rate": 1.998889123324901e-06,
+      "loss": 2.4614,
+      "step": 3850
+    },
+    {
+      "epoch": 0.11366651923927466,
+      "grad_norm": 25.875,
+      "learning_rate": 1.998864740595016e-06,
+      "loss": 2.432,
+      "step": 3855
+    },
+    {
+      "epoch": 0.1138139466312841,
+      "grad_norm": 19.375,
+      "learning_rate": 1.998840093324738e-06,
+      "loss": 2.5605,
+      "step": 3860
+    },
+    {
+      "epoch": 0.11396137402329352,
+      "grad_norm": 24.625,
+      "learning_rate": 1.998815181520595e-06,
+      "loss": 2.411,
+      "step": 3865
+    },
+    {
+      "epoch": 0.11410880141530297,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9987900051891843e-06,
+      "loss": 2.5734,
+      "step": 3870
+    },
+    {
+      "epoch": 0.1142562288073124,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9987645643371733e-06,
+      "loss": 2.4893,
+      "step": 3875
+    },
+    {
+      "epoch": 0.11440365619932183,
+      "grad_norm": 17.5,
+      "learning_rate": 1.998738858971301e-06,
+      "loss": 2.5551,
+      "step": 3880
+    },
+    {
+      "epoch": 0.11455108359133127,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9987128890983736e-06,
+      "loss": 2.5463,
+      "step": 3885
+    },
+    {
+      "epoch": 0.1146985109833407,
+      "grad_norm": 16.25,
+      "learning_rate": 1.9986866547252704e-06,
+      "loss": 2.3747,
+      "step": 3890
+    },
+    {
+      "epoch": 0.11484593837535013,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9986601558589393e-06,
+      "loss": 2.5215,
+      "step": 3895
+    },
+    {
+      "epoch": 0.11499336576735958,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.9986333925063968e-06,
+      "loss": 2.4341,
+      "step": 3900
+    },
+    {
+      "epoch": 0.11514079315936901,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9986063646747325e-06,
+      "loss": 2.5542,
+      "step": 3905
+    },
+    {
+      "epoch": 0.11528822055137844,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.998579072371104e-06,
+      "loss": 2.5368,
+      "step": 3910
+    },
+    {
+      "epoch": 0.11543564794338788,
+      "grad_norm": 23.625,
+      "learning_rate": 1.998551515602739e-06,
+      "loss": 2.4132,
+      "step": 3915
+    },
+    {
+      "epoch": 0.11558307533539731,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9985236943769358e-06,
+      "loss": 2.5002,
+      "step": 3920
+    },
+    {
+      "epoch": 0.11573050272740676,
+      "grad_norm": 15.625,
+      "learning_rate": 1.9984956087010635e-06,
+      "loss": 2.2823,
+      "step": 3925
+    },
+    {
+      "epoch": 0.11587793011941619,
+      "grad_norm": 18.25,
+      "learning_rate": 1.9984672585825592e-06,
+      "loss": 2.5808,
+      "step": 3930
+    },
+    {
+      "epoch": 0.11602535751142562,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9984386440289315e-06,
+      "loss": 2.5604,
+      "step": 3935
+    },
+    {
+      "epoch": 0.11617278490343506,
+      "grad_norm": 18.5,
+      "learning_rate": 1.998409765047759e-06,
+      "loss": 2.482,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1163202122954445,
+      "grad_norm": 14.5,
+      "learning_rate": 1.99838062164669e-06,
+      "loss": 2.4246,
+      "step": 3945
+    },
+    {
+      "epoch": 0.11646763968745392,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9983512138334425e-06,
+      "loss": 2.4198,
+      "step": 3950
+    },
+    {
+      "epoch": 0.11661506707946337,
+      "grad_norm": 16.25,
+      "learning_rate": 1.998321541615805e-06,
+      "loss": 2.434,
+      "step": 3955
+    },
+    {
+      "epoch": 0.1167624944714728,
+      "grad_norm": 24.75,
+      "learning_rate": 1.9982916050016364e-06,
+      "loss": 2.4377,
+      "step": 3960
+    },
+    {
+      "epoch": 0.11690992186348223,
+      "grad_norm": 21.625,
+      "learning_rate": 1.9982614039988643e-06,
+      "loss": 2.4218,
+      "step": 3965
+    },
+    {
+      "epoch": 0.11705734925549167,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9982309386154884e-06,
+      "loss": 2.4098,
+      "step": 3970
+    },
+    {
+      "epoch": 0.1172047766475011,
+      "grad_norm": 16.5,
+      "learning_rate": 1.998200208859576e-06,
+      "loss": 2.3641,
+      "step": 3975
+    },
+    {
+      "epoch": 0.11735220403951054,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9981692147392655e-06,
+      "loss": 2.4643,
+      "step": 3980
+    },
+    {
+      "epoch": 0.11749963143151998,
+      "grad_norm": 15.625,
+      "learning_rate": 1.998137956262767e-06,
+      "loss": 2.3954,
+      "step": 3985
+    },
+    {
+      "epoch": 0.11764705882352941,
+      "grad_norm": 44.75,
+      "learning_rate": 1.9981064334383577e-06,
+      "loss": 2.4361,
+      "step": 3990
+    },
+    {
+      "epoch": 0.11779448621553884,
+      "grad_norm": 30.875,
+      "learning_rate": 1.998074646274386e-06,
+      "loss": 2.5092,
+      "step": 3995
+    },
+    {
+      "epoch": 0.11794191360754828,
+      "grad_norm": 16.25,
+      "learning_rate": 1.998042594779271e-06,
+      "loss": 2.5085,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11794191360754828,
+      "eval_loss": 2.489173173904419,
+      "eval_runtime": 4.7135,
+      "eval_samples_per_second": 84.014,
+      "eval_steps_per_second": 2.758,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11808934099955772,
+      "grad_norm": 17.875,
+      "learning_rate": 1.9980102789615014e-06,
+      "loss": 2.5217,
+      "step": 4005
+    },
+    {
+      "epoch": 0.11823676839156716,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9979776988296353e-06,
+      "loss": 2.383,
+      "step": 4010
+    },
+    {
+      "epoch": 0.11838419578357659,
+      "grad_norm": 66.5,
+      "learning_rate": 1.997944854392301e-06,
+      "loss": 2.5012,
+      "step": 4015
+    },
+    {
+      "epoch": 0.11853162317558602,
+      "grad_norm": 18.125,
+      "learning_rate": 1.997911745658198e-06,
+      "loss": 2.4182,
+      "step": 4020
+    },
+    {
+      "epoch": 0.11867905056759546,
+      "grad_norm": 22.75,
+      "learning_rate": 1.9978783726360945e-06,
+      "loss": 2.4518,
+      "step": 4025
+    },
+    {
+      "epoch": 0.1188264779596049,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9978447353348287e-06,
+      "loss": 2.3372,
+      "step": 4030
+    },
+    {
+      "epoch": 0.11897390535161433,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9978108337633092e-06,
+      "loss": 2.4263,
+      "step": 4035
+    },
+    {
+      "epoch": 0.11912133274362377,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9977766679305143e-06,
+      "loss": 2.547,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1192687601356332,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9977422378454936e-06,
+      "loss": 2.3247,
+      "step": 4045
+    },
+    {
+      "epoch": 0.11941618752764263,
+      "grad_norm": 15.5625,
+      "learning_rate": 1.9977075435173646e-06,
+      "loss": 2.3781,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11956361491965208,
+      "grad_norm": 23.75,
+      "learning_rate": 1.997672584955316e-06,
+      "loss": 2.3924,
+      "step": 4055
+    },
+    {
+      "epoch": 0.1197110423116615,
+      "grad_norm": 21.75,
+      "learning_rate": 1.997637362168606e-06,
+      "loss": 2.4931,
+      "step": 4060
+    },
+    {
+      "epoch": 0.11985846970367094,
+      "grad_norm": 20.375,
+      "learning_rate": 1.997601875166564e-06,
+      "loss": 2.4232,
+      "step": 4065
+    },
+    {
+      "epoch": 0.12000589709568038,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.9975661239585874e-06,
+      "loss": 2.3576,
+      "step": 4070
+    },
+    {
+      "epoch": 0.12015332448768981,
+      "grad_norm": 16.875,
+      "learning_rate": 1.997530108554145e-06,
+      "loss": 2.439,
+      "step": 4075
+    },
+    {
+      "epoch": 0.12030075187969924,
+      "grad_norm": 19.75,
+      "learning_rate": 1.997493828962775e-06,
+      "loss": 2.4518,
+      "step": 4080
+    },
+    {
+      "epoch": 0.12044817927170869,
+      "grad_norm": 18.125,
+      "learning_rate": 1.997457285194086e-06,
+      "loss": 2.4058,
+      "step": 4085
+    },
+    {
+      "epoch": 0.12059560666371812,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9974204772577557e-06,
+      "loss": 2.4125,
+      "step": 4090
+    },
+    {
+      "epoch": 0.12074303405572756,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9973834051635332e-06,
+      "loss": 2.4462,
+      "step": 4095
+    },
+    {
+      "epoch": 0.12089046144773699,
+      "grad_norm": 14.25,
+      "learning_rate": 1.9973460689212366e-06,
+      "loss": 2.3354,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12103788883974642,
+      "grad_norm": 16.25,
+      "learning_rate": 1.997308468540753e-06,
+      "loss": 2.3519,
+      "step": 4105
+    },
+    {
+      "epoch": 0.12118531623175587,
+      "grad_norm": 16.375,
+      "learning_rate": 1.997270604032042e-06,
+      "loss": 2.3724,
+      "step": 4110
+    },
+    {
+      "epoch": 0.1213327436237653,
+      "grad_norm": 25.75,
+      "learning_rate": 1.9972324754051306e-06,
+      "loss": 2.5401,
+      "step": 4115
+    },
+    {
+      "epoch": 0.12148017101577473,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9971940826701175e-06,
+      "loss": 2.4093,
+      "step": 4120
+    },
+    {
+      "epoch": 0.12162759840778417,
+      "grad_norm": 17.25,
+      "learning_rate": 1.99715542583717e-06,
+      "loss": 2.4163,
+      "step": 4125
+    },
+    {
+      "epoch": 0.1217750257997936,
+      "grad_norm": 14.625,
+      "learning_rate": 1.9971165049165266e-06,
+      "loss": 2.397,
+      "step": 4130
+    },
+    {
+      "epoch": 0.12192245319180303,
+      "grad_norm": 20.625,
+      "learning_rate": 1.997077319918495e-06,
+      "loss": 2.4253,
+      "step": 4135
+    },
+    {
+      "epoch": 0.12206988058381248,
+      "grad_norm": 19.625,
+      "learning_rate": 1.9970378708534527e-06,
+      "loss": 2.3918,
+      "step": 4140
+    },
+    {
+      "epoch": 0.1222173079758219,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.9969981577318476e-06,
+      "loss": 2.4003,
+      "step": 4145
+    },
+    {
+      "epoch": 0.12236473536783134,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9969581805641977e-06,
+      "loss": 2.3088,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12251216275984078,
+      "grad_norm": 17.875,
+      "learning_rate": 1.99691793936109e-06,
+      "loss": 2.3631,
+      "step": 4155
+    },
+    {
+      "epoch": 0.12265959015185021,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.9968774341331828e-06,
+      "loss": 2.3745,
+      "step": 4160
+    },
+    {
+      "epoch": 0.12280701754385964,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9968366648912024e-06,
+      "loss": 2.3856,
+      "step": 4165
+    },
+    {
+      "epoch": 0.12295444493586909,
+      "grad_norm": 22.5,
+      "learning_rate": 1.9967956316459473e-06,
+      "loss": 2.5149,
+      "step": 4170
+    },
+    {
+      "epoch": 0.12310187232787852,
+      "grad_norm": 23.5,
+      "learning_rate": 1.9967543344082845e-06,
+      "loss": 2.3066,
+      "step": 4175
+    },
+    {
+      "epoch": 0.12324929971988796,
+      "grad_norm": 14.125,
+      "learning_rate": 1.996712773189151e-06,
+      "loss": 2.3431,
+      "step": 4180
+    },
+    {
+      "epoch": 0.12339672711189739,
+      "grad_norm": 19.625,
+      "learning_rate": 1.996670947999554e-06,
+      "loss": 2.3903,
+      "step": 4185
+    },
+    {
+      "epoch": 0.12354415450390682,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9966288588505705e-06,
+      "loss": 2.4746,
+      "step": 4190
+    },
+    {
+      "epoch": 0.12369158189591627,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9965865057533474e-06,
+      "loss": 2.313,
+      "step": 4195
+    },
+    {
+      "epoch": 0.1238390092879257,
+      "grad_norm": 17.375,
+      "learning_rate": 1.996543888719101e-06,
+      "loss": 2.5405,
+      "step": 4200
+    },
+    {
+      "epoch": 0.12398643667993513,
+      "grad_norm": 16.0,
+      "learning_rate": 1.996501007759119e-06,
+      "loss": 2.3146,
+      "step": 4205
+    },
+    {
+      "epoch": 0.12413386407194457,
+      "grad_norm": 17.125,
+      "learning_rate": 1.996457862884758e-06,
+      "loss": 2.2757,
+      "step": 4210
+    },
+    {
+      "epoch": 0.124281291463954,
+      "grad_norm": 17.875,
+      "learning_rate": 1.996414454107444e-06,
+      "loss": 2.3661,
+      "step": 4215
+    },
+    {
+      "epoch": 0.12442871885596343,
+      "grad_norm": 18.375,
+      "learning_rate": 1.996370781438674e-06,
+      "loss": 2.3062,
+      "step": 4220
+    },
+    {
+      "epoch": 0.12457614624797288,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9963268448900133e-06,
+      "loss": 2.4657,
+      "step": 4225
+    },
+    {
+      "epoch": 0.12472357363998231,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9962826444730992e-06,
+      "loss": 2.4002,
+      "step": 4230
+    },
+    {
+      "epoch": 0.12487100103199174,
+      "grad_norm": 16.75,
+      "learning_rate": 1.996238180199637e-06,
+      "loss": 2.3283,
+      "step": 4235
+    },
+    {
+      "epoch": 0.12501842842400118,
+      "grad_norm": 14.375,
+      "learning_rate": 1.996193452081403e-06,
+      "loss": 2.3856,
+      "step": 4240
+    },
+    {
+      "epoch": 0.1251658558160106,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.996148460130243e-06,
+      "loss": 2.3721,
+      "step": 4245
+    },
+    {
+      "epoch": 0.12531328320802004,
+      "grad_norm": 18.625,
+      "learning_rate": 1.9961032043580726e-06,
+      "loss": 2.4488,
+      "step": 4250
+    },
+    {
+      "epoch": 0.12546071060002947,
+      "grad_norm": 27.125,
+      "learning_rate": 1.9960576847768784e-06,
+      "loss": 2.3926,
+      "step": 4255
+    },
+    {
+      "epoch": 0.12560813799203893,
+      "grad_norm": 17.625,
+      "learning_rate": 1.996011901398714e-06,
+      "loss": 2.3488,
+      "step": 4260
+    },
+    {
+      "epoch": 0.12575556538404836,
+      "grad_norm": 16.375,
+      "learning_rate": 1.995965854235706e-06,
+      "loss": 2.441,
+      "step": 4265
+    },
+    {
+      "epoch": 0.1259029927760578,
+      "grad_norm": 15.25,
+      "learning_rate": 1.9959195433000496e-06,
+      "loss": 2.4117,
+      "step": 4270
+    },
+    {
+      "epoch": 0.12605042016806722,
+      "grad_norm": 14.75,
+      "learning_rate": 1.995872968604009e-06,
+      "loss": 2.3304,
+      "step": 4275
+    },
+    {
+      "epoch": 0.12619784756007665,
+      "grad_norm": 14.3125,
+      "learning_rate": 1.9958261301599195e-06,
+      "loss": 2.4009,
+      "step": 4280
+    },
+    {
+      "epoch": 0.12634527495208608,
+      "grad_norm": 17.25,
+      "learning_rate": 1.995779027980187e-06,
+      "loss": 2.4338,
+      "step": 4285
+    },
+    {
+      "epoch": 0.12649270234409554,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.9957316620772842e-06,
+      "loss": 2.4749,
+      "step": 4290
+    },
+    {
+      "epoch": 0.12664012973610497,
+      "grad_norm": 18.375,
+      "learning_rate": 1.9956840324637564e-06,
+      "loss": 2.3095,
+      "step": 4295
+    },
+    {
+      "epoch": 0.1267875571281144,
+      "grad_norm": 20.0,
+      "learning_rate": 1.9956361391522177e-06,
+      "loss": 2.3714,
+      "step": 4300
+    },
+    {
+      "epoch": 0.12693498452012383,
+      "grad_norm": 22.125,
+      "learning_rate": 1.995587982155353e-06,
+      "loss": 2.4341,
+      "step": 4305
+    },
+    {
+      "epoch": 0.12708241191213326,
+      "grad_norm": 16.625,
+      "learning_rate": 1.995539561485915e-06,
+      "loss": 2.3509,
+      "step": 4310
+    },
+    {
+      "epoch": 0.12722983930414272,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9954908771567287e-06,
+      "loss": 2.4267,
+      "step": 4315
+    },
+    {
+      "epoch": 0.12737726669615215,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9954419291806865e-06,
+      "loss": 2.3689,
+      "step": 4320
+    },
+    {
+      "epoch": 0.12752469408816158,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.995392717570753e-06,
+      "loss": 2.3899,
+      "step": 4325
+    },
+    {
+      "epoch": 0.127672121480171,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9953432423399606e-06,
+      "loss": 2.2691,
+      "step": 4330
+    },
+    {
+      "epoch": 0.12781954887218044,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.9952935035014126e-06,
+      "loss": 2.4342,
+      "step": 4335
+    },
+    {
+      "epoch": 0.12796697626418987,
+      "grad_norm": 22.0,
+      "learning_rate": 1.995243501068282e-06,
+      "loss": 2.2938,
+      "step": 4340
+    },
+    {
+      "epoch": 0.12811440365619933,
+      "grad_norm": 22.75,
+      "learning_rate": 1.9951932350538113e-06,
+      "loss": 2.3834,
+      "step": 4345
+    },
+    {
+      "epoch": 0.12826183104820876,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9951427054713137e-06,
+      "loss": 2.5038,
+      "step": 4350
+    },
+    {
+      "epoch": 0.1284092584402182,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9950919123341707e-06,
+      "loss": 2.4017,
+      "step": 4355
+    },
+    {
+      "epoch": 0.12855668583222762,
+      "grad_norm": 21.375,
+      "learning_rate": 1.9950408556558344e-06,
+      "loss": 2.3696,
+      "step": 4360
+    },
+    {
+      "epoch": 0.12870411322423705,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9949895354498272e-06,
+      "loss": 2.3681,
+      "step": 4365
+    },
+    {
+      "epoch": 0.12885154061624648,
+      "grad_norm": 13.0,
+      "learning_rate": 1.9949379517297404e-06,
+      "loss": 2.4246,
+      "step": 4370
+    },
+    {
+      "epoch": 0.12899896800825594,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.994886104509236e-06,
+      "loss": 2.4722,
+      "step": 4375
+    },
+    {
+      "epoch": 0.12914639540026537,
+      "grad_norm": 17.25,
+      "learning_rate": 1.994833993802045e-06,
+      "loss": 2.3309,
+      "step": 4380
+    },
+    {
+      "epoch": 0.1292938227922748,
+      "grad_norm": 16.5,
+      "learning_rate": 1.994781619621968e-06,
+      "loss": 2.385,
+      "step": 4385
+    },
+    {
+      "epoch": 0.12944125018428423,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9947289819828764e-06,
+      "loss": 2.301,
+      "step": 4390
+    },
+    {
+      "epoch": 0.12958867757629366,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9946760808987106e-06,
+      "loss": 2.4786,
+      "step": 4395
+    },
+    {
+      "epoch": 0.12973610496830312,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.994622916383481e-06,
+      "loss": 2.2656,
+      "step": 4400
+    },
+    {
+      "epoch": 0.12988353236031255,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.994569488451268e-06,
+      "loss": 2.1932,
+      "step": 4405
+    },
+    {
+      "epoch": 0.13003095975232198,
+      "grad_norm": 14.875,
+      "learning_rate": 1.9945157971162207e-06,
+      "loss": 2.3135,
+      "step": 4410
+    },
+    {
+      "epoch": 0.13017838714433141,
+      "grad_norm": 22.625,
+      "learning_rate": 1.99446184239256e-06,
+      "loss": 2.3792,
+      "step": 4415
+    },
+    {
+      "epoch": 0.13032581453634084,
+      "grad_norm": 20.625,
+      "learning_rate": 1.9944076242945744e-06,
+      "loss": 2.403,
+      "step": 4420
+    },
+    {
+      "epoch": 0.13047324192835028,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9943531428366233e-06,
+      "loss": 2.4211,
+      "step": 4425
+    },
+    {
+      "epoch": 0.13062066932035973,
+      "grad_norm": 16.75,
+      "learning_rate": 1.9942983980331355e-06,
+      "loss": 2.3437,
+      "step": 4430
+    },
+    {
+      "epoch": 0.13076809671236916,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.99424338989861e-06,
+      "loss": 2.3165,
+      "step": 4435
+    },
+    {
+      "epoch": 0.1309155241043786,
+      "grad_norm": 25.0,
+      "learning_rate": 1.994188118447615e-06,
+      "loss": 2.396,
+      "step": 4440
+    },
+    {
+      "epoch": 0.13106295149638802,
+      "grad_norm": 19.125,
+      "learning_rate": 1.9941325836947888e-06,
+      "loss": 2.3832,
+      "step": 4445
+    },
+    {
+      "epoch": 0.13121037888839746,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9940767856548395e-06,
+      "loss": 2.3981,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13135780628040689,
+      "grad_norm": 17.5,
+      "learning_rate": 1.994020724342544e-06,
+      "loss": 2.3459,
+      "step": 4455
+    },
+    {
+      "epoch": 0.13150523367241634,
+      "grad_norm": 15.625,
+      "learning_rate": 1.99396439977275e-06,
+      "loss": 2.4587,
+      "step": 4460
+    },
+    {
+      "epoch": 0.13165266106442577,
+      "grad_norm": 25.125,
+      "learning_rate": 1.9939078119603746e-06,
+      "loss": 2.4407,
+      "step": 4465
+    },
+    {
+      "epoch": 0.1318000884564352,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9938509609204047e-06,
+      "loss": 2.3194,
+      "step": 4470
+    },
+    {
+      "epoch": 0.13194751584844464,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9937938466678967e-06,
+      "loss": 2.4082,
+      "step": 4475
+    },
+    {
+      "epoch": 0.13209494324045407,
+      "grad_norm": 14.5625,
+      "learning_rate": 1.9937364692179764e-06,
+      "loss": 2.2631,
+      "step": 4480
+    },
+    {
+      "epoch": 0.13224237063246352,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.993678828585841e-06,
+      "loss": 2.2747,
+      "step": 4485
+    },
+    {
+      "epoch": 0.13238979802447295,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9936209247867542e-06,
+      "loss": 2.3008,
+      "step": 4490
+    },
+    {
+      "epoch": 0.13253722541648238,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9935627578360526e-06,
+      "loss": 2.3542,
+      "step": 4495
+    },
+    {
+      "epoch": 0.13268465280849182,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9935043277491407e-06,
+      "loss": 2.3161,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13268465280849182,
+      "eval_loss": 2.3908541202545166,
+      "eval_runtime": 4.7136,
+      "eval_samples_per_second": 84.013,
+      "eval_steps_per_second": 2.758,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13283208020050125,
+      "grad_norm": 15.0625,
+      "learning_rate": 1.9934456345414938e-06,
+      "loss": 2.3238,
+      "step": 4505
+    },
+    {
+      "epoch": 0.13297950759251068,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9933866782286553e-06,
+      "loss": 2.2844,
+      "step": 4510
+    },
+    {
+      "epoch": 0.13312693498452013,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.99332745882624e-06,
+      "loss": 2.3055,
+      "step": 4515
+    },
+    {
+      "epoch": 0.13327436237652956,
+      "grad_norm": 15.0,
+      "learning_rate": 1.9932679763499313e-06,
+      "loss": 2.4481,
+      "step": 4520
+    },
+    {
+      "epoch": 0.133421789768539,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.9932082308154833e-06,
+      "loss": 2.2749,
+      "step": 4525
+    },
+    {
+      "epoch": 0.13356921716054843,
+      "grad_norm": 16.5,
+      "learning_rate": 1.993148222238718e-06,
+      "loss": 2.5095,
+      "step": 4530
+    },
+    {
+      "epoch": 0.13371664455255786,
+      "grad_norm": 13.875,
+      "learning_rate": 1.9930879506355285e-06,
+      "loss": 2.4653,
+      "step": 4535
+    },
+    {
+      "epoch": 0.13386407194456731,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9930274160218773e-06,
+      "loss": 2.3229,
+      "step": 4540
+    },
+    {
+      "epoch": 0.13401149933657674,
+      "grad_norm": 21.625,
+      "learning_rate": 1.9929666184137964e-06,
+      "loss": 2.3757,
+      "step": 4545
+    },
+    {
+      "epoch": 0.13415892672858618,
+      "grad_norm": 16.75,
+      "learning_rate": 1.992905557827388e-06,
+      "loss": 2.4258,
+      "step": 4550
+    },
+    {
+      "epoch": 0.1343063541205956,
+      "grad_norm": 13.75,
+      "learning_rate": 1.992844234278823e-06,
+      "loss": 2.3937,
+      "step": 4555
+    },
+    {
+      "epoch": 0.13445378151260504,
+      "grad_norm": 19.375,
+      "learning_rate": 1.9927826477843416e-06,
+      "loss": 2.4173,
+      "step": 4560
+    },
+    {
+      "epoch": 0.13460120890461447,
+      "grad_norm": 14.75,
+      "learning_rate": 1.992720798360256e-06,
+      "loss": 2.3872,
+      "step": 4565
+    },
+    {
+      "epoch": 0.13474863629662392,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9926586860229455e-06,
+      "loss": 2.3229,
+      "step": 4570
+    },
+    {
+      "epoch": 0.13489606368863336,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.99259631078886e-06,
+      "loss": 2.3331,
+      "step": 4575
+    },
+    {
+      "epoch": 0.13504349108064279,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.9925336726745196e-06,
+      "loss": 2.3182,
+      "step": 4580
+    },
+    {
+      "epoch": 0.13519091847265222,
+      "grad_norm": 16.375,
+      "learning_rate": 1.992470771696513e-06,
+      "loss": 2.2982,
+      "step": 4585
+    },
+    {
+      "epoch": 0.13533834586466165,
+      "grad_norm": 15.625,
+      "learning_rate": 1.992407607871499e-06,
+      "loss": 2.3287,
+      "step": 4590
+    },
+    {
+      "epoch": 0.13548577325667108,
+      "grad_norm": 17.5,
+      "learning_rate": 1.992344181216206e-06,
+      "loss": 2.4427,
+      "step": 4595
+    },
+    {
+      "epoch": 0.13563320064868054,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9922804917474316e-06,
+      "loss": 2.3961,
+      "step": 4600
+    },
+    {
+      "epoch": 0.13578062804068997,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9922165394820445e-06,
+      "loss": 2.4671,
+      "step": 4605
+    },
+    {
+      "epoch": 0.1359280554326994,
+      "grad_norm": 22.25,
+      "learning_rate": 1.9921523244369805e-06,
+      "loss": 2.4358,
+      "step": 4610
+    },
+    {
+      "epoch": 0.13607548282470883,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9920878466292473e-06,
+      "loss": 2.4312,
+      "step": 4615
+    },
+    {
+      "epoch": 0.13622291021671826,
+      "grad_norm": 18.5,
+      "learning_rate": 1.9920231060759207e-06,
+      "loss": 2.5213,
+      "step": 4620
+    },
+    {
+      "epoch": 0.13637033760872772,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.9919581027941476e-06,
+      "loss": 2.2932,
+      "step": 4625
+    },
+    {
+      "epoch": 0.13651776500073715,
+      "grad_norm": 20.875,
+      "learning_rate": 1.9918928368011426e-06,
+      "loss": 2.3134,
+      "step": 4630
+    },
+    {
+      "epoch": 0.13666519239274658,
+      "grad_norm": 16.25,
+      "learning_rate": 1.991827308114191e-06,
+      "loss": 2.3506,
+      "step": 4635
+    },
+    {
+      "epoch": 0.136812619784756,
+      "grad_norm": 14.5,
+      "learning_rate": 1.9917615167506477e-06,
+      "loss": 2.3828,
+      "step": 4640
+    },
+    {
+      "epoch": 0.13696004717676544,
+      "grad_norm": 40.25,
+      "learning_rate": 1.9916954627279373e-06,
+      "loss": 2.3665,
+      "step": 4645
+    },
+    {
+      "epoch": 0.13710747456877487,
+      "grad_norm": 21.75,
+      "learning_rate": 1.9916291460635522e-06,
+      "loss": 2.3163,
+      "step": 4650
+    },
+    {
+      "epoch": 0.13725490196078433,
+      "grad_norm": 14.75,
+      "learning_rate": 1.9915625667750577e-06,
+      "loss": 2.3346,
+      "step": 4655
+    },
+    {
+      "epoch": 0.13740232935279376,
+      "grad_norm": 18.125,
+      "learning_rate": 1.991495724880085e-06,
+      "loss": 2.3689,
+      "step": 4660
+    },
+    {
+      "epoch": 0.1375497567448032,
+      "grad_norm": 16.0,
+      "learning_rate": 1.991428620396338e-06,
+      "loss": 2.3924,
+      "step": 4665
+    },
+    {
+      "epoch": 0.13769718413681262,
+      "grad_norm": 15.5,
+      "learning_rate": 1.9913612533415877e-06,
+      "loss": 2.2315,
+      "step": 4670
+    },
+    {
+      "epoch": 0.13784461152882205,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9912936237336764e-06,
+      "loss": 2.3822,
+      "step": 4675
+    },
+    {
+      "epoch": 0.13799203892083148,
+      "grad_norm": 21.5,
+      "learning_rate": 1.9912257315905145e-06,
+      "loss": 2.3552,
+      "step": 4680
+    },
+    {
+      "epoch": 0.13813946631284094,
+      "grad_norm": 19.5,
+      "learning_rate": 1.991157576930083e-06,
+      "loss": 2.4338,
+      "step": 4685
+    },
+    {
+      "epoch": 0.13828689370485037,
+      "grad_norm": 20.875,
+      "learning_rate": 1.991089159770432e-06,
+      "loss": 2.3953,
+      "step": 4690
+    },
+    {
+      "epoch": 0.1384343210968598,
+      "grad_norm": 50.0,
+      "learning_rate": 1.9910204801296814e-06,
+      "loss": 2.3419,
+      "step": 4695
+    },
+    {
+      "epoch": 0.13858174848886923,
+      "grad_norm": 15.25,
+      "learning_rate": 1.99095153802602e-06,
+      "loss": 2.342,
+      "step": 4700
+    },
+    {
+      "epoch": 0.13872917588087866,
+      "grad_norm": 23.25,
+      "learning_rate": 1.9908823334777067e-06,
+      "loss": 2.4395,
+      "step": 4705
+    },
+    {
+      "epoch": 0.13887660327288812,
+      "grad_norm": 17.0,
+      "learning_rate": 1.9908128665030697e-06,
+      "loss": 2.3248,
+      "step": 4710
+    },
+    {
+      "epoch": 0.13902403066489755,
+      "grad_norm": 17.5,
+      "learning_rate": 1.990743137120507e-06,
+      "loss": 2.3639,
+      "step": 4715
+    },
+    {
+      "epoch": 0.13917145805690698,
+      "grad_norm": 18.25,
+      "learning_rate": 1.990673145348485e-06,
+      "loss": 2.3686,
+      "step": 4720
+    },
+    {
+      "epoch": 0.1393188854489164,
+      "grad_norm": 15.5,
+      "learning_rate": 1.990602891205541e-06,
+      "loss": 2.3286,
+      "step": 4725
+    },
+    {
+      "epoch": 0.13946631284092584,
+      "grad_norm": 19.0,
+      "learning_rate": 1.9905323747102813e-06,
+      "loss": 2.3668,
+      "step": 4730
+    },
+    {
+      "epoch": 0.13961374023293527,
+      "grad_norm": 14.125,
+      "learning_rate": 1.9904615958813814e-06,
+      "loss": 2.3007,
+      "step": 4735
+    },
+    {
+      "epoch": 0.13976116762494473,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.990390554737586e-06,
+      "loss": 2.3387,
+      "step": 4740
+    },
+    {
+      "epoch": 0.13990859501695416,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9903192512977104e-06,
+      "loss": 2.3229,
+      "step": 4745
+    },
+    {
+      "epoch": 0.1400560224089636,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9902476855806382e-06,
+      "loss": 2.3936,
+      "step": 4750
+    },
+    {
+      "epoch": 0.14020344980097302,
+      "grad_norm": 16.375,
+      "learning_rate": 1.990175857605323e-06,
+      "loss": 2.4185,
+      "step": 4755
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 13.125,
+      "learning_rate": 1.9901037673907884e-06,
+      "loss": 2.2939,
+      "step": 4760
+    },
+    {
+      "epoch": 0.14049830458499188,
+      "grad_norm": 16.25,
+      "learning_rate": 1.990031414956126e-06,
+      "loss": 2.2509,
+      "step": 4765
+    },
+    {
+      "epoch": 0.14064573197700134,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.989958800320498e-06,
+      "loss": 2.255,
+      "step": 4770
+    },
+    {
+      "epoch": 0.14079315936901077,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.989885923503136e-06,
+      "loss": 2.3749,
+      "step": 4775
+    },
+    {
+      "epoch": 0.1409405867610202,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.98981278452334e-06,
+      "loss": 2.3896,
+      "step": 4780
+    },
+    {
+      "epoch": 0.14108801415302963,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.989739383400481e-06,
+      "loss": 2.1821,
+      "step": 4785
+    },
+    {
+      "epoch": 0.14123544154503906,
+      "grad_norm": 17.5,
+      "learning_rate": 1.989665720153999e-06,
+      "loss": 2.3354,
+      "step": 4790
+    },
+    {
+      "epoch": 0.14138286893704852,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.989591794803402e-06,
+      "loss": 2.2849,
+      "step": 4795
+    },
+    {
+      "epoch": 0.14153029632905795,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.9895176073682685e-06,
+      "loss": 2.3225,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14167772372106738,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9894431578682474e-06,
+      "loss": 2.3356,
+      "step": 4805
+    },
+    {
+      "epoch": 0.1418251511130768,
+      "grad_norm": 25.0,
+      "learning_rate": 1.989368446323055e-06,
+      "loss": 2.3309,
+      "step": 4810
+    },
+    {
+      "epoch": 0.14197257850508624,
+      "grad_norm": 17.875,
+      "learning_rate": 1.989293472752479e-06,
+      "loss": 2.2506,
+      "step": 4815
+    },
+    {
+      "epoch": 0.14212000589709567,
+      "grad_norm": 13.875,
+      "learning_rate": 1.989218237176374e-06,
+      "loss": 2.2593,
+      "step": 4820
+    },
+    {
+      "epoch": 0.14226743328910513,
+      "grad_norm": 17.0,
+      "learning_rate": 1.989142739614667e-06,
+      "loss": 2.3378,
+      "step": 4825
+    },
+    {
+      "epoch": 0.14241486068111456,
+      "grad_norm": 14.1875,
+      "learning_rate": 1.9890669800873518e-06,
+      "loss": 2.2224,
+      "step": 4830
+    },
+    {
+      "epoch": 0.142562288073124,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.9889909586144927e-06,
+      "loss": 2.2454,
+      "step": 4835
+    },
+    {
+      "epoch": 0.14270971546513342,
+      "grad_norm": 20.0,
+      "learning_rate": 1.988914675216224e-06,
+      "loss": 2.3326,
+      "step": 4840
+    },
+    {
+      "epoch": 0.14285714285714285,
+      "grad_norm": 14.5,
+      "learning_rate": 1.9888381299127484e-06,
+      "loss": 2.289,
+      "step": 4845
+    },
+    {
+      "epoch": 0.14300457024915228,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9887613227243377e-06,
+      "loss": 2.3622,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14315199764116174,
+      "grad_norm": 20.0,
+      "learning_rate": 1.9886842536713342e-06,
+      "loss": 2.4091,
+      "step": 4855
+    },
+    {
+      "epoch": 0.14329942503317117,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.988606922774149e-06,
+      "loss": 2.3741,
+      "step": 4860
+    },
+    {
+      "epoch": 0.1434468524251806,
+      "grad_norm": 16.375,
+      "learning_rate": 1.9885293300532623e-06,
+      "loss": 2.3865,
+      "step": 4865
+    },
+    {
+      "epoch": 0.14359427981719003,
+      "grad_norm": 13.625,
+      "learning_rate": 1.9884514755292236e-06,
+      "loss": 2.297,
+      "step": 4870
+    },
+    {
+      "epoch": 0.14374170720919946,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.988373359222652e-06,
+      "loss": 2.3604,
+      "step": 4875
+    },
+    {
+      "epoch": 0.14388913460120892,
+      "grad_norm": 17.375,
+      "learning_rate": 1.9882949811542362e-06,
+      "loss": 2.2758,
+      "step": 4880
+    },
+    {
+      "epoch": 0.14403656199321835,
+      "grad_norm": 15.75,
+      "learning_rate": 1.9882163413447337e-06,
+      "loss": 2.4295,
+      "step": 4885
+    },
+    {
+      "epoch": 0.14418398938522778,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9881374398149715e-06,
+      "loss": 2.4267,
+      "step": 4890
+    },
+    {
+      "epoch": 0.1443314167772372,
+      "grad_norm": 18.0,
+      "learning_rate": 1.988058276585847e-06,
+      "loss": 2.3796,
+      "step": 4895
+    },
+    {
+      "epoch": 0.14447884416924664,
+      "grad_norm": 32.0,
+      "learning_rate": 1.9879788516783242e-06,
+      "loss": 2.2386,
+      "step": 4900
+    },
+    {
+      "epoch": 0.14462627156125607,
+      "grad_norm": 18.0,
+      "learning_rate": 1.9878991651134388e-06,
+      "loss": 2.3415,
+      "step": 4905
+    },
+    {
+      "epoch": 0.14477369895326553,
+      "grad_norm": 13.75,
+      "learning_rate": 1.9878192169122957e-06,
+      "loss": 2.2535,
+      "step": 4910
+    },
+    {
+      "epoch": 0.14492112634527496,
+      "grad_norm": 17.75,
+      "learning_rate": 1.9877390070960677e-06,
+      "loss": 2.2624,
+      "step": 4915
+    },
+    {
+      "epoch": 0.1450685537372844,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9876585356859977e-06,
+      "loss": 2.3719,
+      "step": 4920
+    },
+    {
+      "epoch": 0.14521598112929382,
+      "grad_norm": 25.125,
+      "learning_rate": 1.987577802703398e-06,
+      "loss": 2.3524,
+      "step": 4925
+    },
+    {
+      "epoch": 0.14536340852130325,
+      "grad_norm": 18.75,
+      "learning_rate": 1.98749680816965e-06,
+      "loss": 2.3082,
+      "step": 4930
+    },
+    {
+      "epoch": 0.14551083591331268,
+      "grad_norm": 29.125,
+      "learning_rate": 1.9874155521062047e-06,
+      "loss": 2.2639,
+      "step": 4935
+    },
+    {
+      "epoch": 0.14565826330532214,
+      "grad_norm": 13.375,
+      "learning_rate": 1.9873340345345816e-06,
+      "loss": 2.2192,
+      "step": 4940
+    },
+    {
+      "epoch": 0.14580569069733157,
+      "grad_norm": 15.875,
+      "learning_rate": 1.9872522554763698e-06,
+      "loss": 2.4222,
+      "step": 4945
+    },
+    {
+      "epoch": 0.145953118089341,
+      "grad_norm": 17.625,
+      "learning_rate": 1.987170214953228e-06,
+      "loss": 2.4073,
+      "step": 4950
+    },
+    {
+      "epoch": 0.14610054548135043,
+      "grad_norm": 16.0,
+      "learning_rate": 1.9870879129868842e-06,
+      "loss": 2.3876,
+      "step": 4955
+    },
+    {
+      "epoch": 0.14624797287335986,
+      "grad_norm": 14.25,
+      "learning_rate": 1.987005349599135e-06,
+      "loss": 2.3663,
+      "step": 4960
+    },
+    {
+      "epoch": 0.14639540026536932,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9869225248118463e-06,
+      "loss": 2.5066,
+      "step": 4965
+    },
+    {
+      "epoch": 0.14654282765737875,
+      "grad_norm": 12.5,
+      "learning_rate": 1.9868394386469535e-06,
+      "loss": 2.2031,
+      "step": 4970
+    },
+    {
+      "epoch": 0.14669025504938818,
+      "grad_norm": 13.625,
+      "learning_rate": 1.986756091126462e-06,
+      "loss": 2.308,
+      "step": 4975
+    },
+    {
+      "epoch": 0.1468376824413976,
+      "grad_norm": 17.75,
+      "learning_rate": 1.986672482272445e-06,
+      "loss": 2.3624,
+      "step": 4980
+    },
+    {
+      "epoch": 0.14698510983340704,
+      "grad_norm": 16.875,
+      "learning_rate": 1.9865886121070463e-06,
+      "loss": 2.3513,
+      "step": 4985
+    },
+    {
+      "epoch": 0.14713253722541647,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.986504480652477e-06,
+      "loss": 2.2523,
+      "step": 4990
+    },
+    {
+      "epoch": 0.14727996461742593,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.986420087931019e-06,
+      "loss": 2.3941,
+      "step": 4995
+    },
+    {
+      "epoch": 0.14742739200943536,
+      "grad_norm": 15.8125,
+      "learning_rate": 1.9863354339650234e-06,
+      "loss": 2.2635,
+      "step": 5000
+    },
+    {
+      "epoch": 0.14742739200943536,
+      "eval_loss": 2.3236687183380127,
+      "eval_runtime": 4.7176,
+      "eval_samples_per_second": 83.94,
+      "eval_steps_per_second": 2.756,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 33915,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3986311890783961e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}