{
  "best_metric": 1.203926920890808,
  "best_model_checkpoint": "finetune-checkpoints/checkpoint-24000",
  "epoch": 2.0,
  "eval_steps": 2000,
  "global_step": 24880,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.038585209003216e-05,
      "grad_norm": 1.276573896408081,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.6908,
      "step": 1
    },
    {
      "epoch": 0.0008038585209003215,
      "grad_norm": 1.557997465133667,
      "learning_rate": 1e-05,
      "loss": 1.8209,
      "step": 10
    },
    {
      "epoch": 0.001607717041800643,
      "grad_norm": 1.4084455966949463,
      "learning_rate": 2e-05,
      "loss": 1.7609,
      "step": 20
    },
    {
      "epoch": 0.002411575562700965,
      "grad_norm": 1.5869466066360474,
      "learning_rate": 3e-05,
      "loss": 1.6485,
      "step": 30
    },
    {
      "epoch": 0.003215434083601286,
      "grad_norm": 2.227403163909912,
      "learning_rate": 4e-05,
      "loss": 1.7366,
      "step": 40
    },
    {
      "epoch": 0.0040192926045016075,
      "grad_norm": 1.7820078134536743,
      "learning_rate": 5e-05,
      "loss": 1.5889,
      "step": 50
    },
    {
      "epoch": 0.00482315112540193,
      "grad_norm": 1.197980284690857,
      "learning_rate": 5.9e-05,
      "loss": 1.6265,
      "step": 60
    },
    {
      "epoch": 0.005627009646302251,
      "grad_norm": 1.1902364492416382,
      "learning_rate": 6.9e-05,
      "loss": 1.5198,
      "step": 70
    },
    {
      "epoch": 0.006430868167202572,
      "grad_norm": 2.68631649017334,
      "learning_rate": 7.900000000000001e-05,
      "loss": 1.3046,
      "step": 80
    },
    {
      "epoch": 0.007234726688102894,
      "grad_norm": 1.3876135349273682,
      "learning_rate": 8.900000000000001e-05,
      "loss": 1.4239,
      "step": 90
    },
    {
      "epoch": 0.008038585209003215,
      "grad_norm": 1.8347593545913696,
      "learning_rate": 9.900000000000001e-05,
      "loss": 1.4361,
      "step": 100
    },
    {
      "epoch": 0.008842443729903537,
      "grad_norm": 1.8065059185028076,
      "learning_rate": 9.99636803874092e-05,
      "loss": 1.4075,
      "step": 110
    },
    {
      "epoch": 0.00964630225080386,
      "grad_norm": 1.6402571201324463,
      "learning_rate": 9.992332526230832e-05,
      "loss": 1.4785,
      "step": 120
    },
    {
      "epoch": 0.01045016077170418,
      "grad_norm": 1.8013700246810913,
      "learning_rate": 9.988297013720743e-05,
      "loss": 1.3207,
      "step": 130
    },
    {
      "epoch": 0.011254019292604502,
      "grad_norm": 3.1454508304595947,
      "learning_rate": 9.984261501210655e-05,
      "loss": 1.4448,
      "step": 140
    },
    {
      "epoch": 0.012057877813504822,
      "grad_norm": 2.474430799484253,
      "learning_rate": 9.980225988700565e-05,
      "loss": 1.3956,
      "step": 150
    },
    {
      "epoch": 0.012861736334405145,
      "grad_norm": 1.4373944997787476,
      "learning_rate": 9.976190476190477e-05,
      "loss": 1.2823,
      "step": 160
    },
    {
      "epoch": 0.013665594855305467,
      "grad_norm": 2.847407817840576,
      "learning_rate": 9.972154963680387e-05,
      "loss": 1.3352,
      "step": 170
    },
    {
      "epoch": 0.014469453376205787,
      "grad_norm": 3.7937076091766357,
      "learning_rate": 9.968119451170299e-05,
      "loss": 1.2526,
      "step": 180
    },
    {
      "epoch": 0.01527331189710611,
      "grad_norm": 1.221319556236267,
      "learning_rate": 9.96408393866021e-05,
      "loss": 1.4984,
      "step": 190
    },
    {
      "epoch": 0.01607717041800643,
      "grad_norm": 1.652315378189087,
      "learning_rate": 9.960048426150121e-05,
      "loss": 1.3169,
      "step": 200
    },
    {
      "epoch": 0.016881028938906754,
      "grad_norm": 1.7962154150009155,
      "learning_rate": 9.956012913640033e-05,
      "loss": 1.3491,
      "step": 210
    },
    {
      "epoch": 0.017684887459807074,
      "grad_norm": 1.4589283466339111,
      "learning_rate": 9.951977401129944e-05,
      "loss": 1.3256,
      "step": 220
    },
    {
      "epoch": 0.018488745980707395,
      "grad_norm": 1.7330913543701172,
      "learning_rate": 9.947941888619856e-05,
      "loss": 1.2179,
      "step": 230
    },
    {
      "epoch": 0.01929260450160772,
      "grad_norm": 1.4680728912353516,
      "learning_rate": 9.943906376109766e-05,
      "loss": 1.4964,
      "step": 240
    },
    {
      "epoch": 0.02009646302250804,
      "grad_norm": 1.7182399034500122,
      "learning_rate": 9.939870863599678e-05,
      "loss": 1.4423,
      "step": 250
    },
    {
      "epoch": 0.02090032154340836,
      "grad_norm": 1.5316752195358276,
      "learning_rate": 9.935835351089588e-05,
      "loss": 1.2597,
      "step": 260
    },
    {
      "epoch": 0.021704180064308683,
      "grad_norm": 1.4161688089370728,
      "learning_rate": 9.9317998385795e-05,
      "loss": 1.3918,
      "step": 270
    },
    {
      "epoch": 0.022508038585209004,
      "grad_norm": 2.1475634574890137,
      "learning_rate": 9.92776432606941e-05,
      "loss": 1.3015,
      "step": 280
    },
    {
      "epoch": 0.023311897106109324,
      "grad_norm": 1.2509418725967407,
      "learning_rate": 9.923728813559322e-05,
      "loss": 1.3341,
      "step": 290
    },
    {
      "epoch": 0.024115755627009645,
      "grad_norm": 1.3515390157699585,
      "learning_rate": 9.919693301049233e-05,
      "loss": 1.3216,
      "step": 300
    },
    {
      "epoch": 0.02491961414790997,
      "grad_norm": 1.5757758617401123,
      "learning_rate": 9.915657788539145e-05,
      "loss": 1.3335,
      "step": 310
    },
    {
      "epoch": 0.02572347266881029,
      "grad_norm": 1.8854318857192993,
      "learning_rate": 9.911622276029056e-05,
      "loss": 1.3985,
      "step": 320
    },
    {
      "epoch": 0.02652733118971061,
      "grad_norm": 1.7211475372314453,
      "learning_rate": 9.907586763518968e-05,
      "loss": 1.3353,
      "step": 330
    },
    {
      "epoch": 0.027331189710610933,
      "grad_norm": 1.5294115543365479,
      "learning_rate": 9.903551251008879e-05,
      "loss": 1.3076,
      "step": 340
    },
    {
      "epoch": 0.028135048231511254,
      "grad_norm": 1.2507871389389038,
      "learning_rate": 9.89951573849879e-05,
      "loss": 1.3118,
      "step": 350
    },
    {
      "epoch": 0.028938906752411574,
      "grad_norm": 1.4744722843170166,
      "learning_rate": 9.895480225988701e-05,
      "loss": 1.3776,
      "step": 360
    },
    {
      "epoch": 0.0297427652733119,
      "grad_norm": 1.4481561183929443,
      "learning_rate": 9.891444713478613e-05,
      "loss": 1.3754,
      "step": 370
    },
    {
      "epoch": 0.03054662379421222,
      "grad_norm": 2.999114751815796,
      "learning_rate": 9.887409200968523e-05,
      "loss": 1.3131,
      "step": 380
    },
    {
      "epoch": 0.03135048231511254,
      "grad_norm": 1.4064487218856812,
      "learning_rate": 9.883373688458435e-05,
      "loss": 1.4513,
      "step": 390
    },
    {
      "epoch": 0.03215434083601286,
      "grad_norm": 1.2479560375213623,
      "learning_rate": 9.879338175948346e-05,
      "loss": 1.411,
      "step": 400
    },
    {
      "epoch": 0.03295819935691318,
      "grad_norm": 3.228004217147827,
      "learning_rate": 9.875302663438257e-05,
      "loss": 1.355,
      "step": 410
    },
    {
      "epoch": 0.03376205787781351,
      "grad_norm": 2.090229034423828,
      "learning_rate": 9.871267150928168e-05,
      "loss": 1.3551,
      "step": 420
    },
    {
      "epoch": 0.03456591639871383,
      "grad_norm": 1.4778763055801392,
      "learning_rate": 9.86723163841808e-05,
      "loss": 1.3192,
      "step": 430
    },
    {
      "epoch": 0.03536977491961415,
      "grad_norm": 1.4477226734161377,
      "learning_rate": 9.863196125907991e-05,
      "loss": 1.3218,
      "step": 440
    },
    {
      "epoch": 0.03617363344051447,
      "grad_norm": 2.188595771789551,
      "learning_rate": 9.859160613397902e-05,
      "loss": 1.2708,
      "step": 450
    },
    {
      "epoch": 0.03697749196141479,
      "grad_norm": 1.2476807832717896,
      "learning_rate": 9.855125100887814e-05,
      "loss": 1.3206,
      "step": 460
    },
    {
      "epoch": 0.03778135048231511,
      "grad_norm": 1.279171109199524,
      "learning_rate": 9.851089588377724e-05,
      "loss": 1.2349,
      "step": 470
    },
    {
      "epoch": 0.03858520900321544,
      "grad_norm": 1.156655192375183,
      "learning_rate": 9.847054075867636e-05,
      "loss": 1.2511,
      "step": 480
    },
    {
      "epoch": 0.03938906752411576,
      "grad_norm": 1.1249313354492188,
      "learning_rate": 9.843018563357546e-05,
      "loss": 1.2895,
      "step": 490
    },
    {
      "epoch": 0.04019292604501608,
      "grad_norm": 2.300807237625122,
      "learning_rate": 9.838983050847458e-05,
      "loss": 1.2176,
      "step": 500
    },
    {
      "epoch": 0.0409967845659164,
      "grad_norm": 1.7253382205963135,
      "learning_rate": 9.834947538337369e-05,
      "loss": 1.2453,
      "step": 510
    },
    {
      "epoch": 0.04180064308681672,
      "grad_norm": 1.288348913192749,
      "learning_rate": 9.83091202582728e-05,
      "loss": 1.3414,
      "step": 520
    },
    {
      "epoch": 0.04260450160771704,
      "grad_norm": 1.2001973390579224,
      "learning_rate": 9.826876513317191e-05,
      "loss": 1.228,
      "step": 530
    },
    {
      "epoch": 0.04340836012861737,
      "grad_norm": 1.375143051147461,
      "learning_rate": 9.822841000807103e-05,
      "loss": 1.4148,
      "step": 540
    },
    {
      "epoch": 0.04421221864951769,
      "grad_norm": 2.0278618335723877,
      "learning_rate": 9.818805488297013e-05,
      "loss": 1.3985,
      "step": 550
    },
    {
      "epoch": 0.04501607717041801,
      "grad_norm": 1.3718616962432861,
      "learning_rate": 9.814769975786925e-05,
      "loss": 1.2803,
      "step": 560
    },
    {
      "epoch": 0.04581993569131833,
      "grad_norm": 1.0574828386306763,
      "learning_rate": 9.810734463276836e-05,
      "loss": 1.157,
      "step": 570
    },
    {
      "epoch": 0.04662379421221865,
      "grad_norm": 1.2841317653656006,
      "learning_rate": 9.806698950766749e-05,
      "loss": 1.3258,
      "step": 580
    },
    {
      "epoch": 0.04742765273311897,
      "grad_norm": 1.2227920293807983,
      "learning_rate": 9.802663438256659e-05,
      "loss": 1.3772,
      "step": 590
    },
    {
      "epoch": 0.04823151125401929,
      "grad_norm": 1.8267626762390137,
      "learning_rate": 9.798627925746571e-05,
      "loss": 1.4531,
      "step": 600
    },
    {
      "epoch": 0.04903536977491962,
      "grad_norm": 1.2855032682418823,
      "learning_rate": 9.794592413236481e-05,
      "loss": 1.3173,
      "step": 610
    },
    {
      "epoch": 0.04983922829581994,
      "grad_norm": 1.6092931032180786,
      "learning_rate": 9.790556900726393e-05,
      "loss": 1.2729,
      "step": 620
    },
    {
      "epoch": 0.05064308681672026,
      "grad_norm": 1.3694730997085571,
      "learning_rate": 9.786521388216304e-05,
      "loss": 1.3944,
      "step": 630
    },
    {
      "epoch": 0.05144694533762058,
      "grad_norm": 1.0532931089401245,
      "learning_rate": 9.782485875706216e-05,
      "loss": 1.404,
      "step": 640
    },
    {
      "epoch": 0.0522508038585209,
      "grad_norm": 2.206791877746582,
      "learning_rate": 9.778450363196126e-05,
      "loss": 1.2867,
      "step": 650
    },
    {
      "epoch": 0.05305466237942122,
      "grad_norm": 1.2418971061706543,
      "learning_rate": 9.774414850686038e-05,
      "loss": 1.3017,
      "step": 660
    },
    {
      "epoch": 0.053858520900321546,
      "grad_norm": 1.8438310623168945,
      "learning_rate": 9.77037933817595e-05,
      "loss": 1.2832,
      "step": 670
    },
    {
      "epoch": 0.05466237942122187,
      "grad_norm": 1.0188833475112915,
      "learning_rate": 9.76634382566586e-05,
      "loss": 1.3811,
      "step": 680
    },
    {
      "epoch": 0.05546623794212219,
      "grad_norm": 1.3860092163085938,
      "learning_rate": 9.762308313155772e-05,
      "loss": 1.2406,
      "step": 690
    },
    {
      "epoch": 0.05627009646302251,
      "grad_norm": 1.2227293252944946,
      "learning_rate": 9.758272800645682e-05,
      "loss": 1.3276,
      "step": 700
    },
    {
      "epoch": 0.05707395498392283,
      "grad_norm": 1.1025124788284302,
      "learning_rate": 9.754237288135594e-05,
      "loss": 1.3964,
      "step": 710
    },
    {
      "epoch": 0.05787781350482315,
      "grad_norm": 1.0877604484558105,
      "learning_rate": 9.750201775625505e-05,
      "loss": 1.3527,
      "step": 720
    },
    {
      "epoch": 0.058681672025723476,
      "grad_norm": 1.8921070098876953,
      "learning_rate": 9.746166263115416e-05,
      "loss": 1.2887,
      "step": 730
    },
    {
      "epoch": 0.0594855305466238,
      "grad_norm": 1.6759284734725952,
      "learning_rate": 9.742130750605327e-05,
      "loss": 1.3984,
      "step": 740
    },
    {
      "epoch": 0.06028938906752412,
      "grad_norm": 0.9962583780288696,
      "learning_rate": 9.738095238095239e-05,
      "loss": 1.2688,
      "step": 750
    },
    {
      "epoch": 0.06109324758842444,
      "grad_norm": 2.284313917160034,
      "learning_rate": 9.734059725585149e-05,
      "loss": 1.1844,
      "step": 760
    },
    {
      "epoch": 0.06189710610932476,
      "grad_norm": 1.3981996774673462,
      "learning_rate": 9.730024213075061e-05,
      "loss": 1.3251,
      "step": 770
    },
    {
      "epoch": 0.06270096463022508,
      "grad_norm": 1.0155847072601318,
      "learning_rate": 9.725988700564971e-05,
      "loss": 1.1357,
      "step": 780
    },
    {
      "epoch": 0.0635048231511254,
      "grad_norm": 1.4152178764343262,
      "learning_rate": 9.721953188054883e-05,
      "loss": 1.2166,
      "step": 790
    },
    {
      "epoch": 0.06430868167202572,
      "grad_norm": 1.5149229764938354,
      "learning_rate": 9.717917675544794e-05,
      "loss": 1.176,
      "step": 800
    },
    {
      "epoch": 0.06511254019292605,
      "grad_norm": 1.0455445051193237,
      "learning_rate": 9.713882163034706e-05,
      "loss": 1.2231,
      "step": 810
    },
    {
      "epoch": 0.06591639871382636,
      "grad_norm": 1.8290674686431885,
      "learning_rate": 9.709846650524616e-05,
      "loss": 1.3406,
      "step": 820
    },
    {
      "epoch": 0.06672025723472669,
      "grad_norm": 1.7942872047424316,
      "learning_rate": 9.705811138014528e-05,
      "loss": 1.4025,
      "step": 830
    },
    {
      "epoch": 0.06752411575562701,
      "grad_norm": 1.2038955688476562,
      "learning_rate": 9.70177562550444e-05,
      "loss": 1.2591,
      "step": 840
    },
    {
      "epoch": 0.06832797427652733,
      "grad_norm": 1.5919041633605957,
      "learning_rate": 9.697740112994351e-05,
      "loss": 1.3773,
      "step": 850
    },
    {
      "epoch": 0.06913183279742766,
      "grad_norm": 1.2200759649276733,
      "learning_rate": 9.693704600484262e-05,
      "loss": 1.326,
      "step": 860
    },
    {
      "epoch": 0.06993569131832797,
      "grad_norm": 1.3108922243118286,
      "learning_rate": 9.689669087974174e-05,
      "loss": 1.3047,
      "step": 870
    },
    {
      "epoch": 0.0707395498392283,
      "grad_norm": 1.160473346710205,
      "learning_rate": 9.685633575464084e-05,
      "loss": 1.2789,
      "step": 880
    },
    {
      "epoch": 0.07154340836012862,
      "grad_norm": 1.5282626152038574,
      "learning_rate": 9.681598062953996e-05,
      "loss": 1.1021,
      "step": 890
    },
    {
      "epoch": 0.07234726688102894,
      "grad_norm": 2.892331838607788,
      "learning_rate": 9.677562550443908e-05,
      "loss": 1.3719,
      "step": 900
    },
    {
      "epoch": 0.07315112540192927,
      "grad_norm": 1.1138297319412231,
      "learning_rate": 9.673527037933818e-05,
      "loss": 1.1974,
      "step": 910
    },
    {
      "epoch": 0.07395498392282958,
      "grad_norm": 1.3960531949996948,
      "learning_rate": 9.66949152542373e-05,
      "loss": 1.3427,
      "step": 920
    },
    {
      "epoch": 0.0747588424437299,
      "grad_norm": 1.2433034181594849,
      "learning_rate": 9.66545601291364e-05,
      "loss": 1.2951,
      "step": 930
    },
    {
      "epoch": 0.07556270096463022,
      "grad_norm": 1.0800739526748657,
      "learning_rate": 9.661420500403552e-05,
      "loss": 1.2975,
      "step": 940
    },
    {
      "epoch": 0.07636655948553055,
      "grad_norm": 2.4013397693634033,
      "learning_rate": 9.657384987893463e-05,
      "loss": 1.382,
      "step": 950
    },
    {
      "epoch": 0.07717041800643087,
      "grad_norm": 1.9743378162384033,
      "learning_rate": 9.653349475383375e-05,
      "loss": 1.3482,
      "step": 960
    },
    {
      "epoch": 0.07797427652733119,
      "grad_norm": 1.1982871294021606,
      "learning_rate": 9.649313962873285e-05,
      "loss": 1.432,
      "step": 970
    },
    {
      "epoch": 0.07877813504823152,
      "grad_norm": 1.0801359415054321,
      "learning_rate": 9.645278450363197e-05,
      "loss": 1.2414,
      "step": 980
    },
    {
      "epoch": 0.07958199356913183,
      "grad_norm": 1.329847812652588,
      "learning_rate": 9.641242937853107e-05,
      "loss": 1.2838,
      "step": 990
    },
    {
      "epoch": 0.08038585209003216,
      "grad_norm": 2.138526678085327,
      "learning_rate": 9.637207425343019e-05,
      "loss": 1.3736,
      "step": 1000
    },
    {
      "epoch": 0.08118971061093247,
      "grad_norm": 1.342236876487732,
      "learning_rate": 9.63317191283293e-05,
      "loss": 1.3163,
      "step": 1010
    },
    {
      "epoch": 0.0819935691318328,
      "grad_norm": 2.2968924045562744,
      "learning_rate": 9.629136400322841e-05,
      "loss": 1.3696,
      "step": 1020
    },
    {
      "epoch": 0.08279742765273312,
      "grad_norm": 1.244811773300171,
      "learning_rate": 9.625100887812752e-05,
      "loss": 1.196,
      "step": 1030
    },
    {
      "epoch": 0.08360128617363344,
      "grad_norm": 1.9975718259811401,
      "learning_rate": 9.621065375302664e-05,
      "loss": 1.2744,
      "step": 1040
    },
    {
      "epoch": 0.08440514469453377,
      "grad_norm": 1.5397216081619263,
      "learning_rate": 9.617029862792574e-05,
      "loss": 1.2667,
      "step": 1050
    },
    {
      "epoch": 0.08520900321543408,
      "grad_norm": 1.3428692817687988,
      "learning_rate": 9.612994350282486e-05,
      "loss": 1.1593,
      "step": 1060
    },
    {
      "epoch": 0.0860128617363344,
      "grad_norm": 1.1173874139785767,
      "learning_rate": 9.608958837772398e-05,
      "loss": 1.3026,
      "step": 1070
    },
    {
      "epoch": 0.08681672025723473,
      "grad_norm": 2.408698081970215,
      "learning_rate": 9.604923325262308e-05,
      "loss": 1.2274,
      "step": 1080
    },
    {
      "epoch": 0.08762057877813505,
      "grad_norm": 2.4228882789611816,
      "learning_rate": 9.60088781275222e-05,
      "loss": 1.1629,
      "step": 1090
    },
    {
      "epoch": 0.08842443729903537,
      "grad_norm": 1.572268009185791,
      "learning_rate": 9.59685230024213e-05,
      "loss": 1.214,
      "step": 1100
    },
    {
      "epoch": 0.08922829581993569,
      "grad_norm": 1.5269432067871094,
      "learning_rate": 9.592816787732042e-05,
      "loss": 1.2515,
      "step": 1110
    },
    {
      "epoch": 0.09003215434083602,
      "grad_norm": 1.0226801633834839,
      "learning_rate": 9.588781275221954e-05,
      "loss": 1.2765,
      "step": 1120
    },
    {
      "epoch": 0.09083601286173633,
      "grad_norm": 1.3933486938476562,
      "learning_rate": 9.584745762711866e-05,
      "loss": 1.1333,
      "step": 1130
    },
    {
      "epoch": 0.09163987138263666,
      "grad_norm": 1.4505245685577393,
      "learning_rate": 9.580710250201776e-05,
      "loss": 1.2681,
      "step": 1140
    },
    {
      "epoch": 0.09244372990353698,
      "grad_norm": 0.9488227963447571,
      "learning_rate": 9.576674737691688e-05,
      "loss": 1.2356,
      "step": 1150
    },
    {
      "epoch": 0.0932475884244373,
      "grad_norm": 1.6163320541381836,
      "learning_rate": 9.572639225181599e-05,
      "loss": 1.3831,
      "step": 1160
    },
    {
      "epoch": 0.09405144694533762,
      "grad_norm": 2.110747814178467,
      "learning_rate": 9.56860371267151e-05,
      "loss": 1.3194,
      "step": 1170
    },
    {
      "epoch": 0.09485530546623794,
      "grad_norm": 1.251051902770996,
      "learning_rate": 9.564568200161421e-05,
      "loss": 1.2746,
      "step": 1180
    },
    {
      "epoch": 0.09565916398713827,
      "grad_norm": 0.9723356366157532,
      "learning_rate": 9.560532687651333e-05,
      "loss": 1.323,
      "step": 1190
    },
    {
      "epoch": 0.09646302250803858,
      "grad_norm": 1.8033671379089355,
      "learning_rate": 9.556497175141243e-05,
      "loss": 1.1776,
      "step": 1200
    },
    {
      "epoch": 0.0972668810289389,
      "grad_norm": 1.8983824253082275,
      "learning_rate": 9.552461662631155e-05,
      "loss": 1.251,
      "step": 1210
    },
    {
      "epoch": 0.09807073954983923,
      "grad_norm": 1.7401349544525146,
      "learning_rate": 9.548426150121066e-05,
      "loss": 1.2079,
      "step": 1220
    },
    {
      "epoch": 0.09887459807073955,
      "grad_norm": 1.1335679292678833,
      "learning_rate": 9.544390637610977e-05,
      "loss": 1.1727,
      "step": 1230
    },
    {
      "epoch": 0.09967845659163987,
      "grad_norm": 1.8288633823394775,
      "learning_rate": 9.540355125100888e-05,
      "loss": 1.2381,
      "step": 1240
    },
    {
      "epoch": 0.10048231511254019,
      "grad_norm": 1.2624421119689941,
      "learning_rate": 9.5363196125908e-05,
      "loss": 1.3436,
      "step": 1250
    },
    {
      "epoch": 0.10128617363344052,
      "grad_norm": 1.6780205965042114,
      "learning_rate": 9.53228410008071e-05,
      "loss": 1.2763,
      "step": 1260
    },
    {
      "epoch": 0.10209003215434084,
      "grad_norm": 1.6315609216690063,
      "learning_rate": 9.528248587570622e-05,
      "loss": 1.2907,
      "step": 1270
    },
    {
      "epoch": 0.10289389067524116,
      "grad_norm": 1.821751594543457,
      "learning_rate": 9.524213075060532e-05,
      "loss": 1.3109,
      "step": 1280
    },
    {
      "epoch": 0.10369774919614148,
      "grad_norm": 1.9027079343795776,
      "learning_rate": 9.520177562550444e-05,
      "loss": 1.2388,
      "step": 1290
    },
    {
      "epoch": 0.1045016077170418,
      "grad_norm": 0.9719260334968567,
      "learning_rate": 9.516142050040356e-05,
      "loss": 1.325,
      "step": 1300
    },
    {
      "epoch": 0.10530546623794212,
      "grad_norm": 1.5324716567993164,
      "learning_rate": 9.512106537530266e-05,
      "loss": 1.2834,
      "step": 1310
    },
    {
      "epoch": 0.10610932475884244,
      "grad_norm": 1.4811768531799316,
      "learning_rate": 9.508071025020178e-05,
      "loss": 1.2229,
      "step": 1320
    },
    {
      "epoch": 0.10691318327974277,
      "grad_norm": 1.5108731985092163,
      "learning_rate": 9.504035512510089e-05,
      "loss": 1.326,
      "step": 1330
    },
    {
      "epoch": 0.10771704180064309,
      "grad_norm": 1.3529566526412964,
      "learning_rate": 9.5e-05,
      "loss": 1.1673,
      "step": 1340
    },
    {
      "epoch": 0.1085209003215434,
      "grad_norm": 1.7689844369888306,
      "learning_rate": 9.495964487489911e-05,
      "loss": 1.1217,
      "step": 1350
    },
    {
      "epoch": 0.10932475884244373,
      "grad_norm": 1.4722760915756226,
      "learning_rate": 9.491928974979823e-05,
      "loss": 1.3527,
      "step": 1360
    },
    {
      "epoch": 0.11012861736334405,
      "grad_norm": 1.8713065385818481,
      "learning_rate": 9.487893462469735e-05,
      "loss": 1.2872,
      "step": 1370
    },
    {
      "epoch": 0.11093247588424437,
      "grad_norm": 1.243556261062622,
      "learning_rate": 9.483857949959646e-05,
      "loss": 1.2113,
      "step": 1380
    },
    {
      "epoch": 0.11173633440514469,
      "grad_norm": 1.129428505897522,
      "learning_rate": 9.479822437449557e-05,
      "loss": 1.2766,
      "step": 1390
    },
    {
      "epoch": 0.11254019292604502,
      "grad_norm": 2.1960277557373047,
      "learning_rate": 9.475786924939469e-05,
      "loss": 1.3027,
      "step": 1400
    },
    {
      "epoch": 0.11334405144694534,
      "grad_norm": 1.2851146459579468,
      "learning_rate": 9.471751412429379e-05,
      "loss": 1.2294,
      "step": 1410
    },
    {
      "epoch": 0.11414790996784566,
      "grad_norm": 1.365645170211792,
      "learning_rate": 9.467715899919291e-05,
      "loss": 1.4006,
      "step": 1420
    },
    {
      "epoch": 0.11495176848874598,
      "grad_norm": 1.5098659992218018,
      "learning_rate": 9.463680387409201e-05,
      "loss": 1.3402,
      "step": 1430
    },
    {
      "epoch": 0.1157556270096463,
      "grad_norm": 1.6049816608428955,
      "learning_rate": 9.459644874899113e-05,
      "loss": 1.3056,
      "step": 1440
    },
    {
      "epoch": 0.11655948553054662,
      "grad_norm": 1.306289553642273,
      "learning_rate": 9.455609362389024e-05,
      "loss": 1.3096,
      "step": 1450
    },
    {
      "epoch": 0.11736334405144695,
      "grad_norm": 1.9932974576950073,
      "learning_rate": 9.451573849878936e-05,
      "loss": 1.3128,
      "step": 1460
    },
    {
      "epoch": 0.11816720257234727,
      "grad_norm": 1.2543416023254395,
      "learning_rate": 9.447538337368846e-05,
      "loss": 1.2531,
      "step": 1470
    },
    {
      "epoch": 0.1189710610932476,
      "grad_norm": 1.874577522277832,
      "learning_rate": 9.443502824858758e-05,
      "loss": 1.4227,
      "step": 1480
    },
    {
      "epoch": 0.1197749196141479,
      "grad_norm": 1.5035979747772217,
      "learning_rate": 9.439467312348668e-05,
      "loss": 1.3016,
      "step": 1490
    },
    {
      "epoch": 0.12057877813504823,
      "grad_norm": 0.9851014614105225,
      "learning_rate": 9.43543179983858e-05,
      "loss": 1.2965,
      "step": 1500
    },
    {
      "epoch": 0.12138263665594855,
      "grad_norm": 1.2649171352386475,
      "learning_rate": 9.43139628732849e-05,
      "loss": 1.2061,
      "step": 1510
    },
    {
      "epoch": 0.12218649517684887,
      "grad_norm": 2.586120843887329,
      "learning_rate": 9.427360774818402e-05,
      "loss": 1.2258,
      "step": 1520
    },
    {
      "epoch": 0.1229903536977492,
      "grad_norm": 1.7649742364883423,
      "learning_rate": 9.423325262308314e-05,
      "loss": 1.2618,
      "step": 1530
    },
    {
      "epoch": 0.12379421221864952,
      "grad_norm": 1.172960877418518,
      "learning_rate": 9.419289749798225e-05,
      "loss": 1.2505,
      "step": 1540
    },
    {
      "epoch": 0.12459807073954984,
      "grad_norm": 1.0227694511413574,
      "learning_rate": 9.415254237288136e-05,
      "loss": 1.1528,
      "step": 1550
    },
    {
      "epoch": 0.12540192926045016,
      "grad_norm": 1.686591386795044,
      "learning_rate": 9.411218724778047e-05,
      "loss": 1.2418,
      "step": 1560
    },
    {
      "epoch": 0.12620578778135047,
      "grad_norm": 1.7634520530700684,
      "learning_rate": 9.407183212267959e-05,
      "loss": 1.2923,
      "step": 1570
    },
    {
      "epoch": 0.1270096463022508,
      "grad_norm": 1.1225353479385376,
      "learning_rate": 9.403147699757869e-05,
      "loss": 1.3068,
      "step": 1580
    },
    {
      "epoch": 0.12781350482315113,
      "grad_norm": 1.3780168294906616,
      "learning_rate": 9.399112187247781e-05,
      "loss": 1.2754,
      "step": 1590
    },
    {
      "epoch": 0.12861736334405144,
      "grad_norm": 3.1708788871765137,
      "learning_rate": 9.395076674737691e-05,
      "loss": 1.2333,
      "step": 1600
    },
    {
      "epoch": 0.12942122186495178,
      "grad_norm": 1.2919507026672363,
      "learning_rate": 9.391041162227603e-05,
      "loss": 1.269,
      "step": 1610
    },
    {
      "epoch": 0.1302250803858521,
      "grad_norm": 1.1988024711608887,
      "learning_rate": 9.387005649717514e-05,
      "loss": 1.3041,
      "step": 1620
    },
    {
      "epoch": 0.1310289389067524,
      "grad_norm": 1.3553966283798218,
      "learning_rate": 9.382970137207427e-05,
      "loss": 1.36,
      "step": 1630
    },
    {
      "epoch": 0.13183279742765272,
      "grad_norm": 1.6519627571105957,
      "learning_rate": 9.378934624697337e-05,
      "loss": 1.2687,
      "step": 1640
    },
    {
      "epoch": 0.13263665594855306,
      "grad_norm": 1.3935558795928955,
      "learning_rate": 9.374899112187249e-05,
      "loss": 1.2566,
      "step": 1650
    },
    {
      "epoch": 0.13344051446945338,
      "grad_norm": 1.8146889209747314,
      "learning_rate": 9.37086359967716e-05,
      "loss": 1.3103,
      "step": 1660
    },
    {
      "epoch": 0.1342443729903537,
      "grad_norm": 1.1725800037384033,
      "learning_rate": 9.366828087167071e-05,
      "loss": 1.28,
      "step": 1670
    },
    {
      "epoch": 0.13504823151125403,
      "grad_norm": 1.4799379110336304,
      "learning_rate": 9.362792574656982e-05,
      "loss": 1.2874,
      "step": 1680
    },
    {
      "epoch": 0.13585209003215434,
      "grad_norm": 1.1668813228607178,
      "learning_rate": 9.358757062146894e-05,
      "loss": 1.2449,
      "step": 1690
    },
    {
      "epoch": 0.13665594855305466,
      "grad_norm": 1.3387470245361328,
      "learning_rate": 9.354721549636804e-05,
      "loss": 1.336,
      "step": 1700
    },
    {
      "epoch": 0.13745980707395497,
      "grad_norm": 1.42355477809906,
      "learning_rate": 9.350686037126716e-05,
      "loss": 1.3606,
      "step": 1710
    },
    {
      "epoch": 0.1382636655948553,
      "grad_norm": 1.2502082586288452,
      "learning_rate": 9.346650524616626e-05,
      "loss": 1.2905,
      "step": 1720
    },
    {
      "epoch": 0.13906752411575563,
      "grad_norm": 1.7439844608306885,
      "learning_rate": 9.342615012106538e-05,
      "loss": 1.2083,
      "step": 1730
    },
    {
      "epoch": 0.13987138263665594,
      "grad_norm": 0.9318333268165588,
      "learning_rate": 9.338579499596449e-05,
      "loss": 1.2035,
      "step": 1740
    },
    {
      "epoch": 0.14067524115755628,
      "grad_norm": 1.819989800453186,
      "learning_rate": 9.33454398708636e-05,
      "loss": 1.2072,
      "step": 1750
    },
    {
      "epoch": 0.1414790996784566,
      "grad_norm": 1.5602185726165771,
      "learning_rate": 9.330508474576271e-05,
      "loss": 1.1104,
      "step": 1760
    },
    {
      "epoch": 0.1422829581993569,
      "grad_norm": 1.0436007976531982,
      "learning_rate": 9.326472962066183e-05,
      "loss": 1.188,
      "step": 1770
    },
    {
      "epoch": 0.14308681672025725,
      "grad_norm": 3.7814908027648926,
      "learning_rate": 9.322437449556095e-05,
      "loss": 1.2902,
      "step": 1780
    },
    {
      "epoch": 0.14389067524115756,
      "grad_norm": 1.4683253765106201,
      "learning_rate": 9.318401937046005e-05,
      "loss": 1.25,
      "step": 1790
    },
    {
      "epoch": 0.14469453376205788,
      "grad_norm": 1.266164779663086,
      "learning_rate": 9.314366424535917e-05,
      "loss": 1.309,
      "step": 1800
    },
    {
      "epoch": 0.1454983922829582,
      "grad_norm": 8.560582160949707,
      "learning_rate": 9.310330912025827e-05,
      "loss": 1.373,
      "step": 1810
    },
    {
      "epoch": 0.14630225080385853,
      "grad_norm": 1.4244352579116821,
      "learning_rate": 9.306295399515739e-05,
      "loss": 1.2828,
      "step": 1820
    },
    {
      "epoch": 0.14710610932475884,
      "grad_norm": 1.052838683128357,
      "learning_rate": 9.30225988700565e-05,
      "loss": 1.2791,
      "step": 1830
    },
    {
      "epoch": 0.14790996784565916,
      "grad_norm": 1.3596222400665283,
      "learning_rate": 9.298224374495561e-05,
      "loss": 1.0957,
      "step": 1840
    },
    {
      "epoch": 0.1487138263665595,
      "grad_norm": 1.2217438220977783,
      "learning_rate": 9.294188861985472e-05,
      "loss": 1.2423,
      "step": 1850
    },
    {
      "epoch": 0.1495176848874598,
      "grad_norm": 1.2153493165969849,
      "learning_rate": 9.290153349475384e-05,
      "loss": 1.3722,
      "step": 1860
    },
    {
      "epoch": 0.15032154340836013,
      "grad_norm": 1.7717721462249756,
      "learning_rate": 9.286117836965294e-05,
      "loss": 1.2418,
      "step": 1870
    },
    {
      "epoch": 0.15112540192926044,
      "grad_norm": 1.3134626150131226,
      "learning_rate": 9.282082324455206e-05,
      "loss": 1.221,
      "step": 1880
    },
    {
      "epoch": 0.15192926045016078,
      "grad_norm": 1.7526469230651855,
      "learning_rate": 9.278046811945116e-05,
      "loss": 1.4151,
      "step": 1890
    },
    {
      "epoch": 0.1527331189710611,
      "grad_norm": 1.1325948238372803,
      "learning_rate": 9.27401129943503e-05,
      "loss": 1.2002,
      "step": 1900
    },
    {
      "epoch": 0.1535369774919614,
      "grad_norm": 1.3506640195846558,
      "learning_rate": 9.26997578692494e-05,
      "loss": 1.2874,
      "step": 1910
    },
    {
      "epoch": 0.15434083601286175,
      "grad_norm": 1.7926762104034424,
      "learning_rate": 9.265940274414852e-05,
      "loss": 1.2386,
      "step": 1920
    },
    {
      "epoch": 0.15514469453376206,
      "grad_norm": 1.4644490480422974,
      "learning_rate": 9.261904761904762e-05,
      "loss": 1.1797,
      "step": 1930
    },
    {
      "epoch": 0.15594855305466238,
      "grad_norm": 1.57390296459198,
      "learning_rate": 9.257869249394674e-05,
      "loss": 1.2558,
      "step": 1940
    },
    {
      "epoch": 0.1567524115755627,
      "grad_norm": 1.5351284742355347,
      "learning_rate": 9.253833736884585e-05,
      "loss": 1.2979,
      "step": 1950
    },
    {
      "epoch": 0.15755627009646303,
      "grad_norm": 1.4718937873840332,
      "learning_rate": 9.249798224374496e-05,
      "loss": 1.3425,
      "step": 1960
    },
    {
      "epoch": 0.15836012861736334,
      "grad_norm": 0.9524021744728088,
      "learning_rate": 9.245762711864407e-05,
      "loss": 1.1977,
      "step": 1970
    },
    {
      "epoch": 0.15916398713826366,
      "grad_norm": 1.8549355268478394,
      "learning_rate": 9.241727199354319e-05,
      "loss": 1.2554,
      "step": 1980
    },
    {
      "epoch": 0.159967845659164,
      "grad_norm": 1.3382093906402588,
      "learning_rate": 9.237691686844229e-05,
      "loss": 1.2411,
      "step": 1990
    },
    {
      "epoch": 0.1607717041800643,
      "grad_norm": 1.112624168395996,
      "learning_rate": 9.233656174334141e-05,
      "loss": 1.2603,
      "step": 2000
    },
    {
      "epoch": 0.1607717041800643,
      "eval_yahma/alpaca-cleaned_loss": 1.284648060798645,
      "eval_yahma/alpaca-cleaned_runtime": 115.7666,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.276,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.16,
      "step": 2000
    },
    {
      "epoch": 0.16157556270096463,
      "grad_norm": 1.0333826541900635,
      "learning_rate": 9.229620661824053e-05,
      "loss": 1.423,
      "step": 2010
    },
    {
      "epoch": 0.16237942122186494,
      "grad_norm": 0.9243280291557312,
      "learning_rate": 9.225585149313963e-05,
      "loss": 1.3014,
      "step": 2020
    },
    {
      "epoch": 0.16318327974276528,
      "grad_norm": 1.417212963104248,
      "learning_rate": 9.221549636803875e-05,
      "loss": 1.3641,
      "step": 2030
    },
    {
      "epoch": 0.1639871382636656,
      "grad_norm": 1.8156367540359497,
      "learning_rate": 9.217514124293785e-05,
      "loss": 1.3503,
      "step": 2040
    },
    {
      "epoch": 0.1647909967845659,
      "grad_norm": 1.642484188079834,
      "learning_rate": 9.213478611783697e-05,
      "loss": 1.2603,
      "step": 2050
    },
    {
      "epoch": 0.16559485530546625,
      "grad_norm": 1.0917778015136719,
      "learning_rate": 9.209443099273608e-05,
      "loss": 1.2754,
      "step": 2060
    },
    {
      "epoch": 0.16639871382636656,
      "grad_norm": 2.1152048110961914,
      "learning_rate": 9.20540758676352e-05,
      "loss": 1.2619,
      "step": 2070
    },
    {
      "epoch": 0.16720257234726688,
      "grad_norm": 1.556593418121338,
      "learning_rate": 9.20137207425343e-05,
      "loss": 1.1498,
      "step": 2080
    },
    {
      "epoch": 0.1680064308681672,
      "grad_norm": 1.4882763624191284,
      "learning_rate": 9.197336561743342e-05,
      "loss": 1.2363,
      "step": 2090
    },
    {
      "epoch": 0.16881028938906753,
      "grad_norm": 1.1114956140518188,
      "learning_rate": 9.193301049233252e-05,
      "loss": 1.2216,
      "step": 2100
    },
    {
      "epoch": 0.16961414790996784,
      "grad_norm": 1.2593668699264526,
      "learning_rate": 9.189265536723164e-05,
      "loss": 1.3039,
      "step": 2110
    },
    {
      "epoch": 0.17041800643086816,
      "grad_norm": 1.329266905784607,
      "learning_rate": 9.185230024213075e-05,
      "loss": 1.2402,
      "step": 2120
    },
    {
      "epoch": 0.1712218649517685,
      "grad_norm": 1.7473903894424438,
      "learning_rate": 9.181194511702986e-05,
      "loss": 1.2572,
      "step": 2130
    },
    {
      "epoch": 0.1720257234726688,
      "grad_norm": 1.2216113805770874,
      "learning_rate": 9.177158999192897e-05,
      "loss": 1.2478,
      "step": 2140
    },
    {
      "epoch": 0.17282958199356913,
      "grad_norm": 1.2641884088516235,
      "learning_rate": 9.173123486682809e-05,
      "loss": 1.2347,
      "step": 2150
    },
    {
      "epoch": 0.17363344051446947,
      "grad_norm": 2.2486350536346436,
      "learning_rate": 9.16908797417272e-05,
      "loss": 1.2163,
      "step": 2160
    },
    {
      "epoch": 0.17443729903536978,
      "grad_norm": 1.4562492370605469,
      "learning_rate": 9.165052461662632e-05,
      "loss": 1.1854,
      "step": 2170
    },
    {
      "epoch": 0.1752411575562701,
      "grad_norm": 1.8461923599243164,
      "learning_rate": 9.161016949152543e-05,
      "loss": 1.2042,
      "step": 2180
    },
    {
      "epoch": 0.1760450160771704,
      "grad_norm": 1.1238946914672852,
      "learning_rate": 9.156981436642455e-05,
      "loss": 1.3557,
      "step": 2190
    },
    {
      "epoch": 0.17684887459807075,
      "grad_norm": 1.1480857133865356,
      "learning_rate": 9.152945924132365e-05,
      "loss": 1.24,
      "step": 2200
    },
    {
      "epoch": 0.17765273311897106,
      "grad_norm": 1.1996605396270752,
      "learning_rate": 9.148910411622277e-05,
      "loss": 1.2744,
      "step": 2210
    },
    {
      "epoch": 0.17845659163987138,
      "grad_norm": 1.267026662826538,
      "learning_rate": 9.144874899112187e-05,
      "loss": 1.2034,
      "step": 2220
    },
    {
      "epoch": 0.17926045016077172,
      "grad_norm": 0.9613250494003296,
      "learning_rate": 9.140839386602099e-05,
      "loss": 1.2365,
      "step": 2230
    },
    {
      "epoch": 0.18006430868167203,
      "grad_norm": 1.1929056644439697,
      "learning_rate": 9.137207425343019e-05,
      "loss": 1.1804,
      "step": 2240
    },
    {
      "epoch": 0.18086816720257234,
      "grad_norm": 1.337989330291748,
      "learning_rate": 9.133171912832931e-05,
      "loss": 1.1404,
      "step": 2250
    },
    {
      "epoch": 0.18167202572347266,
      "grad_norm": 1.1448936462402344,
      "learning_rate": 9.129136400322841e-05,
      "loss": 1.2866,
      "step": 2260
    },
    {
      "epoch": 0.182475884244373,
      "grad_norm": 1.0559418201446533,
      "learning_rate": 9.125100887812753e-05,
      "loss": 1.2936,
      "step": 2270
    },
    {
      "epoch": 0.1832797427652733,
      "grad_norm": 1.2513796091079712,
      "learning_rate": 9.121065375302664e-05,
      "loss": 1.1852,
      "step": 2280
    },
    {
      "epoch": 0.18408360128617363,
      "grad_norm": 1.0850704908370972,
      "learning_rate": 9.117029862792576e-05,
      "loss": 1.3189,
      "step": 2290
    },
    {
      "epoch": 0.18488745980707397,
      "grad_norm": 1.1448137760162354,
      "learning_rate": 9.112994350282486e-05,
      "loss": 1.1907,
      "step": 2300
    },
    {
      "epoch": 0.18569131832797428,
      "grad_norm": 1.3059879541397095,
      "learning_rate": 9.108958837772398e-05,
      "loss": 1.2661,
      "step": 2310
    },
    {
      "epoch": 0.1864951768488746,
      "grad_norm": 1.2946072816848755,
      "learning_rate": 9.104923325262308e-05,
      "loss": 1.4031,
      "step": 2320
    },
    {
      "epoch": 0.1872990353697749,
      "grad_norm": 1.630731463432312,
      "learning_rate": 9.10088781275222e-05,
      "loss": 1.2734,
      "step": 2330
    },
    {
      "epoch": 0.18810289389067525,
      "grad_norm": 1.1476322412490845,
      "learning_rate": 9.096852300242132e-05,
      "loss": 1.1943,
      "step": 2340
    },
    {
      "epoch": 0.18890675241157556,
      "grad_norm": 1.2856926918029785,
      "learning_rate": 9.092816787732042e-05,
      "loss": 1.3245,
      "step": 2350
    },
    {
      "epoch": 0.18971061093247588,
      "grad_norm": 1.6066474914550781,
      "learning_rate": 9.088781275221954e-05,
      "loss": 1.3207,
      "step": 2360
    },
    {
      "epoch": 0.19051446945337622,
      "grad_norm": 1.4631816148757935,
      "learning_rate": 9.084745762711865e-05,
      "loss": 1.2339,
      "step": 2370
    },
    {
      "epoch": 0.19131832797427653,
      "grad_norm": 1.5194926261901855,
      "learning_rate": 9.080710250201776e-05,
      "loss": 1.2414,
      "step": 2380
    },
    {
      "epoch": 0.19212218649517684,
      "grad_norm": 3.205920934677124,
      "learning_rate": 9.076674737691687e-05,
      "loss": 1.2412,
      "step": 2390
    },
    {
      "epoch": 0.19292604501607716,
      "grad_norm": 1.5582493543624878,
      "learning_rate": 9.072639225181599e-05,
      "loss": 1.245,
      "step": 2400
    },
    {
      "epoch": 0.1937299035369775,
      "grad_norm": 1.579376220703125,
      "learning_rate": 9.068603712671509e-05,
      "loss": 1.1683,
      "step": 2410
    },
    {
      "epoch": 0.1945337620578778,
      "grad_norm": 1.1779841184616089,
      "learning_rate": 9.064568200161421e-05,
      "loss": 1.2371,
      "step": 2420
    },
    {
      "epoch": 0.19533762057877813,
      "grad_norm": 1.2082445621490479,
      "learning_rate": 9.060532687651331e-05,
      "loss": 1.2501,
      "step": 2430
    },
    {
      "epoch": 0.19614147909967847,
      "grad_norm": 1.2248579263687134,
      "learning_rate": 9.056497175141243e-05,
      "loss": 1.2724,
      "step": 2440
    },
    {
      "epoch": 0.19694533762057878,
      "grad_norm": 1.2531076669692993,
      "learning_rate": 9.052461662631154e-05,
      "loss": 1.2786,
      "step": 2450
    },
    {
      "epoch": 0.1977491961414791,
      "grad_norm": 1.2422891855239868,
      "learning_rate": 9.048426150121066e-05,
      "loss": 1.4194,
      "step": 2460
    },
    {
      "epoch": 0.1985530546623794,
      "grad_norm": 1.63975191116333,
      "learning_rate": 9.044390637610976e-05,
      "loss": 1.2917,
      "step": 2470
    },
    {
      "epoch": 0.19935691318327975,
      "grad_norm": 1.3208664655685425,
      "learning_rate": 9.040355125100888e-05,
      "loss": 1.2646,
      "step": 2480
    },
    {
      "epoch": 0.20016077170418006,
      "grad_norm": 3.9294943809509277,
      "learning_rate": 9.036319612590798e-05,
      "loss": 1.3366,
      "step": 2490
    },
    {
      "epoch": 0.20096463022508038,
      "grad_norm": 1.5396499633789062,
      "learning_rate": 9.03228410008071e-05,
      "loss": 1.2358,
      "step": 2500
    },
    {
      "epoch": 0.20176848874598072,
      "grad_norm": 1.4243323802947998,
      "learning_rate": 9.028248587570622e-05,
      "loss": 1.2169,
      "step": 2510
    },
    {
      "epoch": 0.20257234726688103,
      "grad_norm": 1.5669913291931152,
      "learning_rate": 9.024213075060534e-05,
      "loss": 1.2263,
      "step": 2520
    },
    {
      "epoch": 0.20337620578778134,
      "grad_norm": 0.9774581789970398,
      "learning_rate": 9.020177562550444e-05,
      "loss": 1.239,
      "step": 2530
    },
    {
      "epoch": 0.20418006430868169,
      "grad_norm": 1.010597586631775,
      "learning_rate": 9.016142050040356e-05,
      "loss": 1.2691,
      "step": 2540
    },
    {
      "epoch": 0.204983922829582,
      "grad_norm": 1.1401910781860352,
      "learning_rate": 9.012106537530266e-05,
      "loss": 1.3039,
      "step": 2550
    },
    {
      "epoch": 0.2057877813504823,
      "grad_norm": 1.491416573524475,
      "learning_rate": 9.008071025020178e-05,
      "loss": 1.3013,
      "step": 2560
    },
    {
      "epoch": 0.20659163987138263,
      "grad_norm": 1.2823078632354736,
      "learning_rate": 9.00403551251009e-05,
      "loss": 1.279,
      "step": 2570
    },
    {
      "epoch": 0.20739549839228297,
      "grad_norm": 1.1345791816711426,
      "learning_rate": 9e-05,
      "loss": 1.2173,
      "step": 2580
    },
    {
      "epoch": 0.20819935691318328,
      "grad_norm": 2.610827922821045,
      "learning_rate": 8.995964487489912e-05,
      "loss": 1.3519,
      "step": 2590
    },
    {
      "epoch": 0.2090032154340836,
      "grad_norm": 1.0789318084716797,
      "learning_rate": 8.991928974979823e-05,
      "loss": 1.3307,
      "step": 2600
    },
    {
      "epoch": 0.20980707395498394,
      "grad_norm": 1.324279546737671,
      "learning_rate": 8.987893462469735e-05,
      "loss": 1.2548,
      "step": 2610
    },
    {
      "epoch": 0.21061093247588425,
      "grad_norm": 1.199806571006775,
      "learning_rate": 8.983857949959645e-05,
      "loss": 1.208,
      "step": 2620
    },
    {
      "epoch": 0.21141479099678456,
      "grad_norm": 1.218712568283081,
      "learning_rate": 8.979822437449557e-05,
      "loss": 1.2557,
      "step": 2630
    },
    {
      "epoch": 0.21221864951768488,
      "grad_norm": 1.133314847946167,
      "learning_rate": 8.975786924939467e-05,
      "loss": 1.2158,
      "step": 2640
    },
    {
      "epoch": 0.21302250803858522,
      "grad_norm": 2.35660719871521,
      "learning_rate": 8.971751412429379e-05,
      "loss": 1.148,
      "step": 2650
    },
    {
      "epoch": 0.21382636655948553,
      "grad_norm": 2.8322184085845947,
      "learning_rate": 8.96771589991929e-05,
      "loss": 1.1055,
      "step": 2660
    },
    {
      "epoch": 0.21463022508038584,
      "grad_norm": 1.280400276184082,
      "learning_rate": 8.963680387409201e-05,
      "loss": 1.3694,
      "step": 2670
    },
    {
      "epoch": 0.21543408360128619,
      "grad_norm": 1.163424015045166,
      "learning_rate": 8.959644874899112e-05,
      "loss": 1.2949,
      "step": 2680
    },
    {
      "epoch": 0.2162379421221865,
      "grad_norm": 1.504102349281311,
      "learning_rate": 8.955609362389024e-05,
      "loss": 1.3514,
      "step": 2690
    },
    {
      "epoch": 0.2170418006430868,
      "grad_norm": 1.1156712770462036,
      "learning_rate": 8.951573849878934e-05,
      "loss": 1.2545,
      "step": 2700
    },
    {
      "epoch": 0.21784565916398713,
      "grad_norm": 1.1746556758880615,
      "learning_rate": 8.947538337368846e-05,
      "loss": 1.2149,
      "step": 2710
    },
    {
      "epoch": 0.21864951768488747,
      "grad_norm": 1.2077722549438477,
      "learning_rate": 8.943502824858756e-05,
      "loss": 1.2764,
      "step": 2720
    },
    {
      "epoch": 0.21945337620578778,
      "grad_norm": 1.1626636981964111,
      "learning_rate": 8.939467312348668e-05,
      "loss": 1.3789,
      "step": 2730
    },
    {
      "epoch": 0.2202572347266881,
      "grad_norm": 2.6447031497955322,
      "learning_rate": 8.93543179983858e-05,
      "loss": 1.2829,
      "step": 2740
    },
    {
      "epoch": 0.22106109324758844,
      "grad_norm": 1.1910282373428345,
      "learning_rate": 8.93139628732849e-05,
      "loss": 1.2785,
      "step": 2750
    },
    {
      "epoch": 0.22186495176848875,
      "grad_norm": 1.3396004438400269,
      "learning_rate": 8.927360774818402e-05,
      "loss": 1.1879,
      "step": 2760
    },
    {
      "epoch": 0.22266881028938906,
      "grad_norm": 1.88126802444458,
      "learning_rate": 8.923325262308314e-05,
      "loss": 1.1849,
      "step": 2770
    },
    {
      "epoch": 0.22347266881028938,
      "grad_norm": 1.8967148065567017,
      "learning_rate": 8.919289749798225e-05,
      "loss": 1.2796,
      "step": 2780
    },
    {
      "epoch": 0.22427652733118972,
      "grad_norm": 1.174562692642212,
      "learning_rate": 8.915254237288136e-05,
      "loss": 1.2017,
      "step": 2790
    },
    {
      "epoch": 0.22508038585209003,
      "grad_norm": 1.0899947881698608,
      "learning_rate": 8.911218724778048e-05,
      "loss": 1.2023,
      "step": 2800
    },
    {
      "epoch": 0.22588424437299034,
      "grad_norm": 3.6643030643463135,
      "learning_rate": 8.907183212267959e-05,
      "loss": 1.1986,
      "step": 2810
    },
    {
      "epoch": 0.2266881028938907,
      "grad_norm": 2.480713367462158,
      "learning_rate": 8.90314769975787e-05,
      "loss": 1.3729,
      "step": 2820
    },
    {
      "epoch": 0.227491961414791,
      "grad_norm": 1.0885637998580933,
      "learning_rate": 8.899112187247781e-05,
      "loss": 1.1907,
      "step": 2830
    },
    {
      "epoch": 0.2282958199356913,
      "grad_norm": 1.0519663095474243,
      "learning_rate": 8.895076674737693e-05,
      "loss": 1.1776,
      "step": 2840
    },
    {
      "epoch": 0.22909967845659163,
      "grad_norm": 2.0941977500915527,
      "learning_rate": 8.891041162227603e-05,
      "loss": 1.2108,
      "step": 2850
    },
    {
      "epoch": 0.22990353697749197,
      "grad_norm": 1.2095335721969604,
      "learning_rate": 8.887005649717515e-05,
      "loss": 1.1837,
      "step": 2860
    },
    {
      "epoch": 0.23070739549839228,
      "grad_norm": 1.6254414319992065,
      "learning_rate": 8.882970137207426e-05,
      "loss": 1.302,
      "step": 2870
    },
    {
      "epoch": 0.2315112540192926,
      "grad_norm": 1.7986174821853638,
      "learning_rate": 8.878934624697337e-05,
      "loss": 1.3875,
      "step": 2880
    },
    {
      "epoch": 0.23231511254019294,
      "grad_norm": 1.1640440225601196,
      "learning_rate": 8.874899112187248e-05,
      "loss": 1.2783,
      "step": 2890
    },
    {
      "epoch": 0.23311897106109325,
      "grad_norm": 1.3544087409973145,
      "learning_rate": 8.87086359967716e-05,
      "loss": 1.1829,
      "step": 2900
    },
    {
      "epoch": 0.23392282958199356,
      "grad_norm": 1.7265479564666748,
      "learning_rate": 8.86682808716707e-05,
      "loss": 1.3421,
      "step": 2910
    },
    {
      "epoch": 0.2347266881028939,
      "grad_norm": 1.130656123161316,
      "learning_rate": 8.862792574656982e-05,
      "loss": 1.35,
      "step": 2920
    },
    {
      "epoch": 0.23553054662379422,
      "grad_norm": 1.722943663597107,
      "learning_rate": 8.858757062146892e-05,
      "loss": 1.2528,
      "step": 2930
    },
    {
      "epoch": 0.23633440514469453,
      "grad_norm": 0.9455968737602234,
      "learning_rate": 8.854721549636804e-05,
      "loss": 1.2752,
      "step": 2940
    },
    {
      "epoch": 0.23713826366559485,
      "grad_norm": 1.1642229557037354,
      "learning_rate": 8.850686037126715e-05,
      "loss": 1.2127,
      "step": 2950
    },
    {
      "epoch": 0.2379421221864952,
      "grad_norm": 1.1500279903411865,
      "learning_rate": 8.846650524616626e-05,
      "loss": 1.1981,
      "step": 2960
    },
    {
      "epoch": 0.2387459807073955,
      "grad_norm": 1.352089762687683,
      "learning_rate": 8.842615012106538e-05,
      "loss": 1.3122,
      "step": 2970
    },
    {
      "epoch": 0.2395498392282958,
      "grad_norm": 1.4196809530258179,
      "learning_rate": 8.838579499596449e-05,
      "loss": 1.3015,
      "step": 2980
    },
    {
      "epoch": 0.24035369774919615,
      "grad_norm": 1.7181886434555054,
      "learning_rate": 8.83454398708636e-05,
      "loss": 1.3678,
      "step": 2990
    },
    {
      "epoch": 0.24115755627009647,
      "grad_norm": 2.6380298137664795,
      "learning_rate": 8.830508474576271e-05,
      "loss": 1.2444,
      "step": 3000
    },
    {
      "epoch": 0.24196141479099678,
      "grad_norm": 1.2690985202789307,
      "learning_rate": 8.826472962066183e-05,
      "loss": 1.2668,
      "step": 3010
    },
    {
      "epoch": 0.2427652733118971,
      "grad_norm": 1.4184331893920898,
      "learning_rate": 8.822437449556093e-05,
      "loss": 1.1698,
      "step": 3020
    },
    {
      "epoch": 0.24356913183279744,
      "grad_norm": 4.190485954284668,
      "learning_rate": 8.818401937046006e-05,
      "loss": 1.2451,
      "step": 3030
    },
    {
      "epoch": 0.24437299035369775,
      "grad_norm": 1.2775496244430542,
      "learning_rate": 8.814366424535917e-05,
      "loss": 1.1782,
      "step": 3040
    },
    {
      "epoch": 0.24517684887459806,
      "grad_norm": 1.435091257095337,
      "learning_rate": 8.810330912025829e-05,
      "loss": 1.3203,
      "step": 3050
    },
    {
      "epoch": 0.2459807073954984,
      "grad_norm": 1.2115564346313477,
      "learning_rate": 8.806295399515739e-05,
      "loss": 1.1335,
      "step": 3060
    },
    {
      "epoch": 0.24678456591639872,
      "grad_norm": 1.3692493438720703,
      "learning_rate": 8.802259887005651e-05,
      "loss": 1.2182,
      "step": 3070
    },
    {
      "epoch": 0.24758842443729903,
      "grad_norm": 1.2258083820343018,
      "learning_rate": 8.798224374495561e-05,
      "loss": 1.1362,
      "step": 3080
    },
    {
      "epoch": 0.24839228295819935,
      "grad_norm": 1.7599858045578003,
      "learning_rate": 8.794188861985473e-05,
      "loss": 1.3197,
      "step": 3090
    },
    {
      "epoch": 0.2491961414790997,
      "grad_norm": 1.6278449296951294,
      "learning_rate": 8.790153349475384e-05,
      "loss": 1.2382,
      "step": 3100
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.995894432067871,
      "learning_rate": 8.786117836965296e-05,
      "loss": 1.1695,
      "step": 3110
    },
    {
      "epoch": 0.2508038585209003,
      "grad_norm": 1.095329999923706,
      "learning_rate": 8.782082324455206e-05,
      "loss": 1.2429,
      "step": 3120
    },
    {
      "epoch": 0.2516077170418006,
      "grad_norm": 1.7307472229003906,
      "learning_rate": 8.778046811945118e-05,
      "loss": 1.2102,
      "step": 3130
    },
    {
      "epoch": 0.25241157556270094,
      "grad_norm": 1.4965832233428955,
      "learning_rate": 8.774011299435028e-05,
      "loss": 1.2401,
      "step": 3140
    },
    {
      "epoch": 0.2532154340836013,
      "grad_norm": 1.0890928506851196,
      "learning_rate": 8.76997578692494e-05,
      "loss": 1.2583,
      "step": 3150
    },
    {
      "epoch": 0.2540192926045016,
      "grad_norm": 1.1239075660705566,
      "learning_rate": 8.76594027441485e-05,
      "loss": 1.2871,
      "step": 3160
    },
    {
      "epoch": 0.25482315112540194,
      "grad_norm": 1.455148458480835,
      "learning_rate": 8.761904761904762e-05,
      "loss": 1.2707,
      "step": 3170
    },
    {
      "epoch": 0.25562700964630225,
      "grad_norm": 1.3527302742004395,
      "learning_rate": 8.757869249394673e-05,
      "loss": 1.2797,
      "step": 3180
    },
    {
      "epoch": 0.25643086816720256,
      "grad_norm": 1.3919832706451416,
      "learning_rate": 8.753833736884585e-05,
      "loss": 1.1467,
      "step": 3190
    },
    {
      "epoch": 0.2572347266881029,
      "grad_norm": 1.2856230735778809,
      "learning_rate": 8.749798224374496e-05,
      "loss": 1.2301,
      "step": 3200
    },
    {
      "epoch": 0.2580385852090032,
      "grad_norm": 1.550368070602417,
      "learning_rate": 8.745762711864407e-05,
      "loss": 1.1609,
      "step": 3210
    },
    {
      "epoch": 0.25884244372990356,
      "grad_norm": 0.9304008483886719,
      "learning_rate": 8.741727199354319e-05,
      "loss": 1.2375,
      "step": 3220
    },
    {
      "epoch": 0.2596463022508039,
      "grad_norm": 1.2073613405227661,
      "learning_rate": 8.737691686844229e-05,
      "loss": 1.2715,
      "step": 3230
    },
    {
      "epoch": 0.2604501607717042,
      "grad_norm": 1.894003987312317,
      "learning_rate": 8.733656174334141e-05,
      "loss": 1.205,
      "step": 3240
    },
    {
      "epoch": 0.2612540192926045,
      "grad_norm": 1.4571211338043213,
      "learning_rate": 8.729620661824051e-05,
      "loss": 1.2137,
      "step": 3250
    },
    {
      "epoch": 0.2620578778135048,
      "grad_norm": 1.2148561477661133,
      "learning_rate": 8.725585149313963e-05,
      "loss": 1.3079,
      "step": 3260
    },
    {
      "epoch": 0.2628617363344051,
      "grad_norm": 1.1078846454620361,
      "learning_rate": 8.721549636803874e-05,
      "loss": 1.1524,
      "step": 3270
    },
    {
      "epoch": 0.26366559485530544,
      "grad_norm": 1.6308702230453491,
      "learning_rate": 8.717514124293786e-05,
      "loss": 1.1884,
      "step": 3280
    },
    {
      "epoch": 0.2644694533762058,
      "grad_norm": 1.2123280763626099,
      "learning_rate": 8.713478611783696e-05,
      "loss": 1.2499,
      "step": 3290
    },
    {
      "epoch": 0.2652733118971061,
      "grad_norm": 1.2395896911621094,
      "learning_rate": 8.709443099273609e-05,
      "loss": 1.1591,
      "step": 3300
    },
    {
      "epoch": 0.26607717041800644,
      "grad_norm": 1.0542362928390503,
      "learning_rate": 8.70540758676352e-05,
      "loss": 1.203,
      "step": 3310
    },
    {
      "epoch": 0.26688102893890675,
      "grad_norm": 1.354406476020813,
      "learning_rate": 8.701372074253431e-05,
      "loss": 1.2953,
      "step": 3320
    },
    {
      "epoch": 0.26768488745980706,
      "grad_norm": 1.0819886922836304,
      "learning_rate": 8.697336561743342e-05,
      "loss": 1.2686,
      "step": 3330
    },
    {
      "epoch": 0.2684887459807074,
      "grad_norm": 1.5589094161987305,
      "learning_rate": 8.693301049233254e-05,
      "loss": 1.2366,
      "step": 3340
    },
    {
      "epoch": 0.2692926045016077,
      "grad_norm": 2.0046586990356445,
      "learning_rate": 8.689265536723164e-05,
      "loss": 1.2515,
      "step": 3350
    },
    {
      "epoch": 0.27009646302250806,
      "grad_norm": 2.3307278156280518,
      "learning_rate": 8.685230024213076e-05,
      "loss": 1.2391,
      "step": 3360
    },
    {
      "epoch": 0.2709003215434084,
      "grad_norm": 2.4583561420440674,
      "learning_rate": 8.681194511702986e-05,
      "loss": 1.1907,
      "step": 3370
    },
    {
      "epoch": 0.2717041800643087,
      "grad_norm": 1.040289044380188,
      "learning_rate": 8.677158999192898e-05,
      "loss": 1.1758,
      "step": 3380
    },
    {
      "epoch": 0.272508038585209,
      "grad_norm": 1.7027485370635986,
      "learning_rate": 8.673123486682809e-05,
      "loss": 1.3229,
      "step": 3390
    },
    {
      "epoch": 0.2733118971061093,
      "grad_norm": 1.3173311948776245,
      "learning_rate": 8.66908797417272e-05,
      "loss": 1.3225,
      "step": 3400
    },
    {
      "epoch": 0.2741157556270096,
      "grad_norm": 1.4691309928894043,
      "learning_rate": 8.665052461662631e-05,
      "loss": 1.0976,
      "step": 3410
    },
    {
      "epoch": 0.27491961414790994,
      "grad_norm": 1.385856032371521,
      "learning_rate": 8.661016949152543e-05,
      "loss": 1.2308,
      "step": 3420
    },
    {
      "epoch": 0.2757234726688103,
      "grad_norm": 1.038273811340332,
      "learning_rate": 8.656981436642455e-05,
      "loss": 1.1296,
      "step": 3430
    },
    {
      "epoch": 0.2765273311897106,
      "grad_norm": 1.1960220336914062,
      "learning_rate": 8.652945924132365e-05,
      "loss": 1.2165,
      "step": 3440
    },
    {
      "epoch": 0.27733118971061094,
      "grad_norm": 1.353964924812317,
      "learning_rate": 8.648910411622277e-05,
      "loss": 1.2996,
      "step": 3450
    },
    {
      "epoch": 0.27813504823151125,
      "grad_norm": 1.173205018043518,
      "learning_rate": 8.644874899112187e-05,
      "loss": 1.295,
      "step": 3460
    },
    {
      "epoch": 0.27893890675241156,
      "grad_norm": 2.594675302505493,
      "learning_rate": 8.640839386602099e-05,
      "loss": 1.1145,
      "step": 3470
    },
    {
      "epoch": 0.2797427652733119,
      "grad_norm": 1.1430559158325195,
      "learning_rate": 8.63680387409201e-05,
      "loss": 1.1369,
      "step": 3480
    },
    {
      "epoch": 0.2805466237942122,
      "grad_norm": 1.6096569299697876,
      "learning_rate": 8.632768361581921e-05,
      "loss": 1.1668,
      "step": 3490
    },
    {
      "epoch": 0.28135048231511256,
      "grad_norm": 2.040187120437622,
      "learning_rate": 8.628732849071832e-05,
      "loss": 1.2364,
      "step": 3500
    },
    {
      "epoch": 0.2821543408360129,
      "grad_norm": 1.2003670930862427,
      "learning_rate": 8.624697336561744e-05,
      "loss": 1.2538,
      "step": 3510
    },
    {
      "epoch": 0.2829581993569132,
      "grad_norm": 1.2703702449798584,
      "learning_rate": 8.620661824051654e-05,
      "loss": 1.3134,
      "step": 3520
    },
    {
      "epoch": 0.2837620578778135,
      "grad_norm": 1.3621017932891846,
      "learning_rate": 8.616626311541566e-05,
      "loss": 1.2114,
      "step": 3530
    },
    {
      "epoch": 0.2845659163987138,
      "grad_norm": 2.1689460277557373,
      "learning_rate": 8.612590799031476e-05,
      "loss": 1.393,
      "step": 3540
    },
    {
      "epoch": 0.2853697749196141,
      "grad_norm": 1.1933314800262451,
      "learning_rate": 8.608555286521388e-05,
      "loss": 1.1932,
      "step": 3550
    },
    {
      "epoch": 0.2861736334405145,
      "grad_norm": 1.2700861692428589,
      "learning_rate": 8.6045197740113e-05,
      "loss": 1.4143,
      "step": 3560
    },
    {
      "epoch": 0.2869774919614148,
      "grad_norm": 1.3436776399612427,
      "learning_rate": 8.600484261501212e-05,
      "loss": 1.2522,
      "step": 3570
    },
    {
      "epoch": 0.2877813504823151,
      "grad_norm": 1.207179307937622,
      "learning_rate": 8.596448748991122e-05,
      "loss": 1.2819,
      "step": 3580
    },
    {
      "epoch": 0.28858520900321544,
      "grad_norm": 1.681864619255066,
      "learning_rate": 8.592413236481034e-05,
      "loss": 1.1955,
      "step": 3590
    },
    {
      "epoch": 0.28938906752411575,
      "grad_norm": 1.4330998659133911,
      "learning_rate": 8.588377723970945e-05,
      "loss": 1.2732,
      "step": 3600
    },
    {
      "epoch": 0.29019292604501606,
      "grad_norm": 1.481811285018921,
      "learning_rate": 8.584342211460856e-05,
      "loss": 1.3336,
      "step": 3610
    },
    {
      "epoch": 0.2909967845659164,
      "grad_norm": 1.147080898284912,
      "learning_rate": 8.580306698950767e-05,
      "loss": 1.2514,
      "step": 3620
    },
    {
      "epoch": 0.29180064308681675,
      "grad_norm": 1.0554629564285278,
      "learning_rate": 8.576271186440679e-05,
      "loss": 1.2695,
      "step": 3630
    },
    {
      "epoch": 0.29260450160771706,
      "grad_norm": 1.670669674873352,
      "learning_rate": 8.572235673930589e-05,
      "loss": 1.3654,
      "step": 3640
    },
    {
      "epoch": 0.2934083601286174,
      "grad_norm": 1.398838996887207,
      "learning_rate": 8.568200161420501e-05,
      "loss": 1.2102,
      "step": 3650
    },
    {
      "epoch": 0.2942122186495177,
      "grad_norm": 1.1837551593780518,
      "learning_rate": 8.564164648910413e-05,
      "loss": 1.3595,
      "step": 3660
    },
    {
      "epoch": 0.295016077170418,
      "grad_norm": 1.2118239402770996,
      "learning_rate": 8.560129136400323e-05,
      "loss": 1.2297,
      "step": 3670
    },
    {
      "epoch": 0.2958199356913183,
      "grad_norm": 1.3268067836761475,
      "learning_rate": 8.556093623890235e-05,
      "loss": 1.3314,
      "step": 3680
    },
    {
      "epoch": 0.2966237942122186,
      "grad_norm": 1.6145943403244019,
      "learning_rate": 8.552058111380146e-05,
      "loss": 1.183,
      "step": 3690
    },
    {
      "epoch": 0.297427652733119,
      "grad_norm": 1.7863456010818481,
      "learning_rate": 8.548022598870057e-05,
      "loss": 1.2523,
      "step": 3700
    },
    {
      "epoch": 0.2982315112540193,
      "grad_norm": 1.243323802947998,
      "learning_rate": 8.543987086359968e-05,
      "loss": 1.2971,
      "step": 3710
    },
    {
      "epoch": 0.2990353697749196,
      "grad_norm": 1.4287739992141724,
      "learning_rate": 8.53995157384988e-05,
      "loss": 1.1036,
      "step": 3720
    },
    {
      "epoch": 0.29983922829581994,
      "grad_norm": 1.366148591041565,
      "learning_rate": 8.53591606133979e-05,
      "loss": 1.2176,
      "step": 3730
    },
    {
      "epoch": 0.30064308681672025,
      "grad_norm": 1.3919678926467896,
      "learning_rate": 8.531880548829702e-05,
      "loss": 1.1227,
      "step": 3740
    },
    {
      "epoch": 0.30144694533762056,
      "grad_norm": 1.0967152118682861,
      "learning_rate": 8.527845036319612e-05,
      "loss": 1.2714,
      "step": 3750
    },
    {
      "epoch": 0.3022508038585209,
      "grad_norm": 1.183366060256958,
      "learning_rate": 8.523809523809524e-05,
      "loss": 1.2636,
      "step": 3760
    },
    {
      "epoch": 0.30305466237942125,
      "grad_norm": 1.0084779262542725,
      "learning_rate": 8.519774011299435e-05,
      "loss": 1.4423,
      "step": 3770
    },
    {
      "epoch": 0.30385852090032156,
      "grad_norm": 1.0512844324111938,
      "learning_rate": 8.515738498789346e-05,
      "loss": 1.2682,
      "step": 3780
    },
    {
      "epoch": 0.3046623794212219,
      "grad_norm": 1.1441655158996582,
      "learning_rate": 8.511702986279257e-05,
      "loss": 1.2442,
      "step": 3790
    },
    {
      "epoch": 0.3054662379421222,
      "grad_norm": 1.0703250169754028,
      "learning_rate": 8.507667473769169e-05,
      "loss": 1.263,
      "step": 3800
    },
    {
      "epoch": 0.3062700964630225,
      "grad_norm": 1.2896406650543213,
      "learning_rate": 8.503631961259079e-05,
      "loss": 1.1945,
      "step": 3810
    },
    {
      "epoch": 0.3070739549839228,
      "grad_norm": 1.516065239906311,
      "learning_rate": 8.499596448748992e-05,
      "loss": 1.1396,
      "step": 3820
    },
    {
      "epoch": 0.30787781350482313,
      "grad_norm": 1.256030559539795,
      "learning_rate": 8.495560936238903e-05,
      "loss": 1.2744,
      "step": 3830
    },
    {
      "epoch": 0.3086816720257235,
      "grad_norm": 1.2016384601593018,
      "learning_rate": 8.491525423728815e-05,
      "loss": 1.1509,
      "step": 3840
    },
    {
      "epoch": 0.3094855305466238,
      "grad_norm": 1.8608025312423706,
      "learning_rate": 8.487489911218725e-05,
      "loss": 1.1685,
      "step": 3850
    },
    {
      "epoch": 0.3102893890675241,
      "grad_norm": 1.5279998779296875,
      "learning_rate": 8.483454398708637e-05,
      "loss": 1.2004,
      "step": 3860
    },
    {
      "epoch": 0.31109324758842444,
      "grad_norm": 1.5688719749450684,
      "learning_rate": 8.479418886198547e-05,
      "loss": 1.2176,
      "step": 3870
    },
    {
      "epoch": 0.31189710610932475,
      "grad_norm": 1.5599457025527954,
      "learning_rate": 8.475383373688459e-05,
      "loss": 1.2146,
      "step": 3880
    },
    {
      "epoch": 0.31270096463022506,
      "grad_norm": 1.1148357391357422,
      "learning_rate": 8.471347861178371e-05,
      "loss": 1.3145,
      "step": 3890
    },
    {
      "epoch": 0.3135048231511254,
      "grad_norm": 1.0745151042938232,
      "learning_rate": 8.467312348668281e-05,
      "loss": 1.0896,
      "step": 3900
    },
    {
      "epoch": 0.31430868167202575,
      "grad_norm": 1.194466233253479,
      "learning_rate": 8.463276836158193e-05,
      "loss": 1.1872,
      "step": 3910
    },
    {
      "epoch": 0.31511254019292606,
      "grad_norm": 2.6574392318725586,
      "learning_rate": 8.459241323648104e-05,
      "loss": 1.3583,
      "step": 3920
    },
    {
      "epoch": 0.3159163987138264,
      "grad_norm": 1.4396333694458008,
      "learning_rate": 8.455205811138016e-05,
      "loss": 1.4009,
      "step": 3930
    },
    {
      "epoch": 0.3167202572347267,
      "grad_norm": 1.4764877557754517,
      "learning_rate": 8.451170298627926e-05,
      "loss": 1.312,
      "step": 3940
    },
    {
      "epoch": 0.317524115755627,
      "grad_norm": 1.3817880153656006,
      "learning_rate": 8.447134786117838e-05,
      "loss": 1.2372,
      "step": 3950
    },
    {
      "epoch": 0.3183279742765273,
      "grad_norm": 2.030385971069336,
      "learning_rate": 8.443099273607748e-05,
      "loss": 1.2477,
      "step": 3960
    },
    {
      "epoch": 0.31913183279742763,
      "grad_norm": 2.089034080505371,
      "learning_rate": 8.43906376109766e-05,
      "loss": 1.1984,
      "step": 3970
    },
    {
      "epoch": 0.319935691318328,
      "grad_norm": 1.7432914972305298,
      "learning_rate": 8.43502824858757e-05,
      "loss": 1.1436,
      "step": 3980
    },
    {
      "epoch": 0.3207395498392283,
      "grad_norm": 1.3305381536483765,
      "learning_rate": 8.430992736077482e-05,
      "loss": 1.2557,
      "step": 3990
    },
    {
      "epoch": 0.3215434083601286,
      "grad_norm": 1.6194047927856445,
      "learning_rate": 8.426957223567393e-05,
      "loss": 1.3163,
      "step": 4000
    },
    {
      "epoch": 0.3215434083601286,
      "eval_yahma/alpaca-cleaned_loss": 1.2568495273590088,
      "eval_yahma/alpaca-cleaned_runtime": 115.5517,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.308,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.164,
      "step": 4000
    },
    {
      "epoch": 0.32234726688102894,
      "grad_norm": 0.9996694922447205,
      "learning_rate": 8.422921711057305e-05,
      "loss": 1.1647,
      "step": 4010
    },
    {
      "epoch": 0.32315112540192925,
      "grad_norm": 1.4061142206192017,
      "learning_rate": 8.418886198547215e-05,
      "loss": 1.2166,
      "step": 4020
    },
    {
      "epoch": 0.32395498392282956,
      "grad_norm": 1.2130415439605713,
      "learning_rate": 8.414850686037127e-05,
      "loss": 1.1794,
      "step": 4030
    },
    {
      "epoch": 0.3247588424437299,
      "grad_norm": 1.2440122365951538,
      "learning_rate": 8.410815173527037e-05,
      "loss": 1.2449,
      "step": 4040
    },
    {
      "epoch": 0.32556270096463025,
      "grad_norm": 1.5796377658843994,
      "learning_rate": 8.406779661016949e-05,
      "loss": 1.2174,
      "step": 4050
    },
    {
      "epoch": 0.32636655948553056,
      "grad_norm": 1.8361690044403076,
      "learning_rate": 8.402744148506861e-05,
      "loss": 1.2374,
      "step": 4060
    },
    {
      "epoch": 0.3271704180064309,
      "grad_norm": 1.2278058528900146,
      "learning_rate": 8.398708635996771e-05,
      "loss": 1.1239,
      "step": 4070
    },
    {
      "epoch": 0.3279742765273312,
      "grad_norm": 1.4516998529434204,
      "learning_rate": 8.394673123486683e-05,
      "loss": 1.1387,
      "step": 4080
    },
    {
      "epoch": 0.3287781350482315,
      "grad_norm": 1.1614266633987427,
      "learning_rate": 8.390637610976595e-05,
      "loss": 1.2378,
      "step": 4090
    },
    {
      "epoch": 0.3295819935691318,
      "grad_norm": 1.7242281436920166,
      "learning_rate": 8.386602098466506e-05,
      "loss": 1.2562,
      "step": 4100
    },
    {
      "epoch": 0.33038585209003213,
      "grad_norm": 1.3742716312408447,
      "learning_rate": 8.382566585956417e-05,
      "loss": 1.1921,
      "step": 4110
    },
    {
      "epoch": 0.3311897106109325,
      "grad_norm": 1.2327730655670166,
      "learning_rate": 8.378531073446329e-05,
      "loss": 1.1925,
      "step": 4120
    },
    {
      "epoch": 0.3319935691318328,
      "grad_norm": 1.3033393621444702,
      "learning_rate": 8.37449556093624e-05,
      "loss": 1.2141,
      "step": 4130
    },
    {
      "epoch": 0.3327974276527331,
      "grad_norm": 1.1770600080490112,
      "learning_rate": 8.370460048426151e-05,
      "loss": 1.2859,
      "step": 4140
    },
    {
      "epoch": 0.33360128617363344,
      "grad_norm": 1.2515443563461304,
      "learning_rate": 8.366424535916062e-05,
      "loss": 1.1253,
      "step": 4150
    },
    {
      "epoch": 0.33440514469453375,
      "grad_norm": 1.2094742059707642,
      "learning_rate": 8.362389023405974e-05,
      "loss": 1.2391,
      "step": 4160
    },
    {
      "epoch": 0.33520900321543406,
      "grad_norm": 1.5006566047668457,
      "learning_rate": 8.358353510895884e-05,
      "loss": 1.2237,
      "step": 4170
    },
    {
      "epoch": 0.3360128617363344,
      "grad_norm": 1.1610201597213745,
      "learning_rate": 8.354317998385796e-05,
      "loss": 1.2401,
      "step": 4180
    },
    {
      "epoch": 0.33681672025723475,
      "grad_norm": 1.9824213981628418,
      "learning_rate": 8.350282485875706e-05,
      "loss": 1.3003,
      "step": 4190
    },
    {
      "epoch": 0.33762057877813506,
      "grad_norm": 1.5451104640960693,
      "learning_rate": 8.346246973365618e-05,
      "loss": 1.2139,
      "step": 4200
    },
    {
      "epoch": 0.3384244372990354,
      "grad_norm": 1.7514973878860474,
      "learning_rate": 8.342211460855529e-05,
      "loss": 1.1285,
      "step": 4210
    },
    {
      "epoch": 0.3392282958199357,
      "grad_norm": 1.2136577367782593,
      "learning_rate": 8.33817594834544e-05,
      "loss": 1.2802,
      "step": 4220
    },
    {
      "epoch": 0.340032154340836,
      "grad_norm": 1.0182433128356934,
      "learning_rate": 8.334140435835351e-05,
      "loss": 1.2155,
      "step": 4230
    },
    {
      "epoch": 0.3408360128617363,
      "grad_norm": 0.9786971807479858,
      "learning_rate": 8.330104923325263e-05,
      "loss": 1.0544,
      "step": 4240
    },
    {
      "epoch": 0.34163987138263663,
      "grad_norm": 1.7164803743362427,
      "learning_rate": 8.326069410815173e-05,
      "loss": 1.1538,
      "step": 4250
    },
    {
      "epoch": 0.342443729903537,
      "grad_norm": 1.1870179176330566,
      "learning_rate": 8.322033898305085e-05,
      "loss": 1.2086,
      "step": 4260
    },
    {
      "epoch": 0.3432475884244373,
      "grad_norm": 1.4695736169815063,
      "learning_rate": 8.317998385794996e-05,
      "loss": 0.9652,
      "step": 4270
    },
    {
      "epoch": 0.3440514469453376,
      "grad_norm": 1.3140954971313477,
      "learning_rate": 8.313962873284907e-05,
      "loss": 1.1532,
      "step": 4280
    },
    {
      "epoch": 0.34485530546623794,
      "grad_norm": 1.1161473989486694,
      "learning_rate": 8.309927360774819e-05,
      "loss": 1.1443,
      "step": 4290
    },
    {
      "epoch": 0.34565916398713825,
      "grad_norm": 1.8414597511291504,
      "learning_rate": 8.30589184826473e-05,
      "loss": 1.2339,
      "step": 4300
    },
    {
      "epoch": 0.34646302250803857,
      "grad_norm": 1.5777239799499512,
      "learning_rate": 8.301856335754641e-05,
      "loss": 1.2191,
      "step": 4310
    },
    {
      "epoch": 0.34726688102893893,
      "grad_norm": 1.560793161392212,
      "learning_rate": 8.297820823244552e-05,
      "loss": 1.2505,
      "step": 4320
    },
    {
      "epoch": 0.34807073954983925,
      "grad_norm": 1.2204498052597046,
      "learning_rate": 8.293785310734464e-05,
      "loss": 1.1672,
      "step": 4330
    },
    {
      "epoch": 0.34887459807073956,
      "grad_norm": 1.1328129768371582,
      "learning_rate": 8.289749798224374e-05,
      "loss": 1.3139,
      "step": 4340
    },
    {
      "epoch": 0.3496784565916399,
      "grad_norm": 2.0714643001556396,
      "learning_rate": 8.285714285714287e-05,
      "loss": 1.2314,
      "step": 4350
    },
    {
      "epoch": 0.3504823151125402,
      "grad_norm": 1.0736972093582153,
      "learning_rate": 8.281678773204198e-05,
      "loss": 1.3055,
      "step": 4360
    },
    {
      "epoch": 0.3512861736334405,
      "grad_norm": 1.7276315689086914,
      "learning_rate": 8.27764326069411e-05,
      "loss": 1.157,
      "step": 4370
    },
    {
      "epoch": 0.3520900321543408,
      "grad_norm": 1.200021505355835,
      "learning_rate": 8.27360774818402e-05,
      "loss": 1.1639,
      "step": 4380
    },
    {
      "epoch": 0.3528938906752412,
      "grad_norm": 1.3981629610061646,
      "learning_rate": 8.269572235673932e-05,
      "loss": 1.2204,
      "step": 4390
    },
    {
      "epoch": 0.3536977491961415,
      "grad_norm": 1.264017105102539,
      "learning_rate": 8.265536723163842e-05,
      "loss": 1.3322,
      "step": 4400
    },
    {
      "epoch": 0.3545016077170418,
      "grad_norm": 1.5476160049438477,
      "learning_rate": 8.261501210653754e-05,
      "loss": 1.1365,
      "step": 4410
    },
    {
      "epoch": 0.3553054662379421,
      "grad_norm": 2.6459197998046875,
      "learning_rate": 8.257465698143665e-05,
      "loss": 1.2962,
      "step": 4420
    },
    {
      "epoch": 0.35610932475884244,
      "grad_norm": 1.1030009984970093,
      "learning_rate": 8.253430185633576e-05,
      "loss": 1.1737,
      "step": 4430
    },
    {
      "epoch": 0.35691318327974275,
      "grad_norm": 1.3585678339004517,
      "learning_rate": 8.249394673123487e-05,
      "loss": 1.1757,
      "step": 4440
    },
    {
      "epoch": 0.35771704180064307,
      "grad_norm": 1.4126050472259521,
      "learning_rate": 8.245359160613399e-05,
      "loss": 1.2081,
      "step": 4450
    },
    {
      "epoch": 0.35852090032154343,
      "grad_norm": 1.3890093564987183,
      "learning_rate": 8.241323648103309e-05,
      "loss": 1.1337,
      "step": 4460
    },
    {
      "epoch": 0.35932475884244375,
      "grad_norm": 1.015844702720642,
      "learning_rate": 8.237288135593221e-05,
      "loss": 1.1043,
      "step": 4470
    },
    {
      "epoch": 0.36012861736334406,
      "grad_norm": 1.4223793745040894,
      "learning_rate": 8.233252623083131e-05,
      "loss": 1.1537,
      "step": 4480
    },
    {
      "epoch": 0.3609324758842444,
      "grad_norm": 1.3807647228240967,
      "learning_rate": 8.229217110573043e-05,
      "loss": 1.1781,
      "step": 4490
    },
    {
      "epoch": 0.3617363344051447,
      "grad_norm": 1.4933511018753052,
      "learning_rate": 8.225181598062954e-05,
      "loss": 1.1658,
      "step": 4500
    },
    {
      "epoch": 0.362540192926045,
      "grad_norm": 1.1804993152618408,
      "learning_rate": 8.221146085552865e-05,
      "loss": 1.2211,
      "step": 4510
    },
    {
      "epoch": 0.3633440514469453,
      "grad_norm": 1.9485111236572266,
      "learning_rate": 8.217110573042777e-05,
      "loss": 1.216,
      "step": 4520
    },
    {
      "epoch": 0.3641479099678457,
      "grad_norm": 1.215474009513855,
      "learning_rate": 8.213075060532688e-05,
      "loss": 1.2775,
      "step": 4530
    },
    {
      "epoch": 0.364951768488746,
      "grad_norm": 1.2956700325012207,
      "learning_rate": 8.2090395480226e-05,
      "loss": 1.2052,
      "step": 4540
    },
    {
      "epoch": 0.3657556270096463,
      "grad_norm": 1.3250538110733032,
      "learning_rate": 8.20500403551251e-05,
      "loss": 1.345,
      "step": 4550
    },
    {
      "epoch": 0.3665594855305466,
      "grad_norm": 1.5776617527008057,
      "learning_rate": 8.200968523002422e-05,
      "loss": 1.2108,
      "step": 4560
    },
    {
      "epoch": 0.36736334405144694,
      "grad_norm": 1.3848649263381958,
      "learning_rate": 8.196933010492332e-05,
      "loss": 1.1932,
      "step": 4570
    },
    {
      "epoch": 0.36816720257234725,
      "grad_norm": 1.2446579933166504,
      "learning_rate": 8.192897497982244e-05,
      "loss": 1.299,
      "step": 4580
    },
    {
      "epoch": 0.36897106109324757,
      "grad_norm": 1.8784652948379517,
      "learning_rate": 8.188861985472155e-05,
      "loss": 1.2617,
      "step": 4590
    },
    {
      "epoch": 0.36977491961414793,
      "grad_norm": 1.3336700201034546,
      "learning_rate": 8.184826472962066e-05,
      "loss": 1.2814,
      "step": 4600
    },
    {
      "epoch": 0.37057877813504825,
      "grad_norm": 1.560197353363037,
      "learning_rate": 8.180790960451978e-05,
      "loss": 1.2565,
      "step": 4610
    },
    {
      "epoch": 0.37138263665594856,
      "grad_norm": 1.5231118202209473,
      "learning_rate": 8.17675544794189e-05,
      "loss": 1.3068,
      "step": 4620
    },
    {
      "epoch": 0.3721864951768489,
      "grad_norm": 1.2317482233047485,
      "learning_rate": 8.1727199354318e-05,
      "loss": 1.2264,
      "step": 4630
    },
    {
      "epoch": 0.3729903536977492,
      "grad_norm": 1.2745100259780884,
      "learning_rate": 8.168684422921712e-05,
      "loss": 1.4129,
      "step": 4640
    },
    {
      "epoch": 0.3737942122186495,
      "grad_norm": 1.971482753753662,
      "learning_rate": 8.164648910411623e-05,
      "loss": 1.1744,
      "step": 4650
    },
    {
      "epoch": 0.3745980707395498,
      "grad_norm": 1.0844069719314575,
      "learning_rate": 8.160613397901535e-05,
      "loss": 1.2031,
      "step": 4660
    },
    {
      "epoch": 0.3754019292604502,
      "grad_norm": 1.330795168876648,
      "learning_rate": 8.156577885391445e-05,
      "loss": 1.0976,
      "step": 4670
    },
    {
      "epoch": 0.3762057877813505,
      "grad_norm": 1.6659519672393799,
      "learning_rate": 8.152542372881357e-05,
      "loss": 1.2434,
      "step": 4680
    },
    {
      "epoch": 0.3770096463022508,
      "grad_norm": 1.3720048666000366,
      "learning_rate": 8.148506860371267e-05,
      "loss": 1.1822,
      "step": 4690
    },
    {
      "epoch": 0.3778135048231511,
      "grad_norm": 1.275566577911377,
      "learning_rate": 8.144471347861179e-05,
      "loss": 1.295,
      "step": 4700
    },
    {
      "epoch": 0.37861736334405144,
      "grad_norm": 1.3874690532684326,
      "learning_rate": 8.14043583535109e-05,
      "loss": 1.1628,
      "step": 4710
    },
    {
      "epoch": 0.37942122186495175,
      "grad_norm": 1.2212285995483398,
      "learning_rate": 8.136400322841001e-05,
      "loss": 1.2601,
      "step": 4720
    },
    {
      "epoch": 0.38022508038585207,
      "grad_norm": 3.2551138401031494,
      "learning_rate": 8.132364810330912e-05,
      "loss": 1.2103,
      "step": 4730
    },
    {
      "epoch": 0.38102893890675243,
      "grad_norm": 1.1511696577072144,
      "learning_rate": 8.128329297820824e-05,
      "loss": 1.1613,
      "step": 4740
    },
    {
      "epoch": 0.38183279742765275,
      "grad_norm": 1.658243179321289,
      "learning_rate": 8.124293785310735e-05,
      "loss": 1.314,
      "step": 4750
    },
    {
      "epoch": 0.38263665594855306,
      "grad_norm": 1.3986806869506836,
      "learning_rate": 8.120258272800646e-05,
      "loss": 1.1491,
      "step": 4760
    },
    {
      "epoch": 0.3834405144694534,
      "grad_norm": 1.2337511777877808,
      "learning_rate": 8.116222760290558e-05,
      "loss": 1.2881,
      "step": 4770
    },
    {
      "epoch": 0.3842443729903537,
      "grad_norm": 1.68783438205719,
      "learning_rate": 8.112187247780468e-05,
      "loss": 1.2599,
      "step": 4780
    },
    {
      "epoch": 0.385048231511254,
      "grad_norm": 1.1215641498565674,
      "learning_rate": 8.10815173527038e-05,
      "loss": 1.2125,
      "step": 4790
    },
    {
      "epoch": 0.3858520900321543,
      "grad_norm": 1.1371439695358276,
      "learning_rate": 8.10411622276029e-05,
      "loss": 1.2937,
      "step": 4800
    },
    {
      "epoch": 0.3866559485530547,
      "grad_norm": 1.5197312831878662,
      "learning_rate": 8.100080710250202e-05,
      "loss": 1.1736,
      "step": 4810
    },
    {
      "epoch": 0.387459807073955,
      "grad_norm": 1.3680319786071777,
      "learning_rate": 8.096045197740113e-05,
      "loss": 1.2799,
      "step": 4820
    },
    {
      "epoch": 0.3882636655948553,
      "grad_norm": 1.2305307388305664,
      "learning_rate": 8.092009685230025e-05,
      "loss": 1.1474,
      "step": 4830
    },
    {
      "epoch": 0.3890675241157556,
      "grad_norm": 1.3372670412063599,
      "learning_rate": 8.087974172719935e-05,
      "loss": 1.2025,
      "step": 4840
    },
    {
      "epoch": 0.38987138263665594,
      "grad_norm": 1.3374559879302979,
      "learning_rate": 8.083938660209847e-05,
      "loss": 1.2011,
      "step": 4850
    },
    {
      "epoch": 0.39067524115755625,
      "grad_norm": 1.3609942197799683,
      "learning_rate": 8.079903147699757e-05,
      "loss": 1.3043,
      "step": 4860
    },
    {
      "epoch": 0.39147909967845657,
      "grad_norm": 1.3824046850204468,
      "learning_rate": 8.075867635189669e-05,
      "loss": 1.2488,
      "step": 4870
    },
    {
      "epoch": 0.39228295819935693,
      "grad_norm": 1.2782840728759766,
      "learning_rate": 8.071832122679581e-05,
      "loss": 1.1891,
      "step": 4880
    },
    {
      "epoch": 0.39308681672025725,
      "grad_norm": 1.4103243350982666,
      "learning_rate": 8.067796610169493e-05,
      "loss": 1.3044,
      "step": 4890
    },
    {
      "epoch": 0.39389067524115756,
      "grad_norm": 1.2876780033111572,
      "learning_rate": 8.063761097659403e-05,
      "loss": 1.1633,
      "step": 4900
    },
    {
      "epoch": 0.3946945337620579,
      "grad_norm": 1.3857921361923218,
      "learning_rate": 8.059725585149315e-05,
      "loss": 1.2069,
      "step": 4910
    },
    {
      "epoch": 0.3954983922829582,
      "grad_norm": 1.3935562372207642,
      "learning_rate": 8.055690072639225e-05,
      "loss": 1.1609,
      "step": 4920
    },
    {
      "epoch": 0.3963022508038585,
      "grad_norm": 2.10925030708313,
      "learning_rate": 8.051654560129137e-05,
      "loss": 1.3154,
      "step": 4930
    },
    {
      "epoch": 0.3971061093247588,
      "grad_norm": 1.2593389749526978,
      "learning_rate": 8.047619047619048e-05,
      "loss": 1.1579,
      "step": 4940
    },
    {
      "epoch": 0.3979099678456592,
      "grad_norm": 1.4237825870513916,
      "learning_rate": 8.04358353510896e-05,
      "loss": 1.2519,
      "step": 4950
    },
    {
      "epoch": 0.3987138263665595,
      "grad_norm": 1.0136818885803223,
      "learning_rate": 8.03954802259887e-05,
      "loss": 1.236,
      "step": 4960
    },
    {
      "epoch": 0.3995176848874598,
      "grad_norm": 1.361685872077942,
      "learning_rate": 8.035512510088782e-05,
      "loss": 1.091,
      "step": 4970
    },
    {
      "epoch": 0.4003215434083601,
      "grad_norm": 1.5500177145004272,
      "learning_rate": 8.031476997578694e-05,
      "loss": 1.2047,
      "step": 4980
    },
    {
      "epoch": 0.40112540192926044,
      "grad_norm": 1.5440484285354614,
      "learning_rate": 8.027441485068604e-05,
      "loss": 1.2541,
      "step": 4990
    },
    {
      "epoch": 0.40192926045016075,
      "grad_norm": 1.7334917783737183,
      "learning_rate": 8.023405972558516e-05,
      "loss": 1.126,
      "step": 5000
    },
    {
      "epoch": 0.40273311897106107,
      "grad_norm": 1.1905109882354736,
      "learning_rate": 8.019370460048426e-05,
      "loss": 1.2036,
      "step": 5010
    },
    {
      "epoch": 0.40353697749196143,
      "grad_norm": 1.008507251739502,
      "learning_rate": 8.015334947538338e-05,
      "loss": 1.3405,
      "step": 5020
    },
    {
      "epoch": 0.40434083601286175,
      "grad_norm": 1.1107839345932007,
      "learning_rate": 8.011299435028249e-05,
      "loss": 1.2577,
      "step": 5030
    },
    {
      "epoch": 0.40514469453376206,
      "grad_norm": 1.179787039756775,
      "learning_rate": 8.00726392251816e-05,
      "loss": 1.1532,
      "step": 5040
    },
    {
      "epoch": 0.4059485530546624,
      "grad_norm": 1.4130208492279053,
      "learning_rate": 8.003228410008071e-05,
      "loss": 1.2049,
      "step": 5050
    },
    {
      "epoch": 0.4067524115755627,
      "grad_norm": 1.2334232330322266,
      "learning_rate": 7.999192897497983e-05,
      "loss": 1.1509,
      "step": 5060
    },
    {
      "epoch": 0.407556270096463,
      "grad_norm": 1.5695637464523315,
      "learning_rate": 7.995157384987893e-05,
      "loss": 1.2461,
      "step": 5070
    },
    {
      "epoch": 0.40836012861736337,
      "grad_norm": 1.1055469512939453,
      "learning_rate": 7.991121872477805e-05,
      "loss": 1.0805,
      "step": 5080
    },
    {
      "epoch": 0.4091639871382637,
      "grad_norm": 1.6603456735610962,
      "learning_rate": 7.987086359967715e-05,
      "loss": 1.0994,
      "step": 5090
    },
    {
      "epoch": 0.409967845659164,
      "grad_norm": 1.1670210361480713,
      "learning_rate": 7.983050847457627e-05,
      "loss": 1.2261,
      "step": 5100
    },
    {
      "epoch": 0.4107717041800643,
      "grad_norm": 1.5910462141036987,
      "learning_rate": 7.979015334947538e-05,
      "loss": 1.1899,
      "step": 5110
    },
    {
      "epoch": 0.4115755627009646,
      "grad_norm": 2.9151222705841064,
      "learning_rate": 7.97497982243745e-05,
      "loss": 1.1837,
      "step": 5120
    },
    {
      "epoch": 0.41237942122186494,
      "grad_norm": 1.3645329475402832,
      "learning_rate": 7.97094430992736e-05,
      "loss": 1.2522,
      "step": 5130
    },
    {
      "epoch": 0.41318327974276525,
      "grad_norm": 1.454911708831787,
      "learning_rate": 7.966908797417273e-05,
      "loss": 1.1749,
      "step": 5140
    },
    {
      "epoch": 0.4139871382636656,
      "grad_norm": 2.2036333084106445,
      "learning_rate": 7.962873284907184e-05,
      "loss": 1.2823,
      "step": 5150
    },
    {
      "epoch": 0.41479099678456594,
      "grad_norm": 1.2479572296142578,
      "learning_rate": 7.958837772397095e-05,
      "loss": 1.2181,
      "step": 5160
    },
    {
      "epoch": 0.41559485530546625,
      "grad_norm": 1.1498271226882935,
      "learning_rate": 7.954802259887006e-05,
      "loss": 1.2351,
      "step": 5170
    },
    {
      "epoch": 0.41639871382636656,
      "grad_norm": 1.4739044904708862,
      "learning_rate": 7.950766747376918e-05,
      "loss": 1.2791,
      "step": 5180
    },
    {
      "epoch": 0.4172025723472669,
      "grad_norm": 1.3020864725112915,
      "learning_rate": 7.946731234866828e-05,
      "loss": 1.1077,
      "step": 5190
    },
    {
      "epoch": 0.4180064308681672,
      "grad_norm": 1.4907817840576172,
      "learning_rate": 7.94269572235674e-05,
      "loss": 1.3331,
      "step": 5200
    },
    {
      "epoch": 0.4188102893890675,
      "grad_norm": 1.3892371654510498,
      "learning_rate": 7.938660209846652e-05,
      "loss": 1.2152,
      "step": 5210
    },
    {
      "epoch": 0.41961414790996787,
      "grad_norm": 0.9748413562774658,
      "learning_rate": 7.934624697336562e-05,
      "loss": 1.2838,
      "step": 5220
    },
    {
      "epoch": 0.4204180064308682,
      "grad_norm": 1.5303418636322021,
      "learning_rate": 7.930589184826474e-05,
      "loss": 1.2055,
      "step": 5230
    },
    {
      "epoch": 0.4212218649517685,
      "grad_norm": 1.7131788730621338,
      "learning_rate": 7.926553672316385e-05,
      "loss": 1.2611,
      "step": 5240
    },
    {
      "epoch": 0.4220257234726688,
      "grad_norm": 1.1032283306121826,
      "learning_rate": 7.922518159806296e-05,
      "loss": 1.2017,
      "step": 5250
    },
    {
      "epoch": 0.4228295819935691,
      "grad_norm": 1.971808910369873,
      "learning_rate": 7.918482647296207e-05,
      "loss": 1.2185,
      "step": 5260
    },
    {
      "epoch": 0.42363344051446944,
      "grad_norm": 1.2828810214996338,
      "learning_rate": 7.914447134786119e-05,
      "loss": 1.1686,
      "step": 5270
    },
    {
      "epoch": 0.42443729903536975,
      "grad_norm": 1.083824872970581,
      "learning_rate": 7.910411622276029e-05,
      "loss": 1.2189,
      "step": 5280
    },
    {
      "epoch": 0.4252411575562701,
      "grad_norm": 1.4020578861236572,
      "learning_rate": 7.906376109765941e-05,
      "loss": 1.1495,
      "step": 5290
    },
    {
      "epoch": 0.42604501607717044,
      "grad_norm": 1.1178410053253174,
      "learning_rate": 7.902340597255851e-05,
      "loss": 1.2211,
      "step": 5300
    },
    {
      "epoch": 0.42684887459807075,
      "grad_norm": 1.4398339986801147,
      "learning_rate": 7.898305084745763e-05,
      "loss": 1.2003,
      "step": 5310
    },
    {
      "epoch": 0.42765273311897106,
      "grad_norm": 1.1164844036102295,
      "learning_rate": 7.894269572235674e-05,
      "loss": 1.1514,
      "step": 5320
    },
    {
      "epoch": 0.4284565916398714,
      "grad_norm": 1.5604671239852905,
      "learning_rate": 7.890234059725585e-05,
      "loss": 1.3449,
      "step": 5330
    },
    {
      "epoch": 0.4292604501607717,
      "grad_norm": 1.345582127571106,
      "learning_rate": 7.886198547215496e-05,
      "loss": 1.144,
      "step": 5340
    },
    {
      "epoch": 0.430064308681672,
      "grad_norm": 1.3933559656143188,
      "learning_rate": 7.882163034705408e-05,
      "loss": 1.1806,
      "step": 5350
    },
    {
      "epoch": 0.43086816720257237,
      "grad_norm": 1.3390429019927979,
      "learning_rate": 7.878127522195318e-05,
      "loss": 1.218,
      "step": 5360
    },
    {
      "epoch": 0.4316720257234727,
      "grad_norm": 2.075464963912964,
      "learning_rate": 7.87409200968523e-05,
      "loss": 1.0676,
      "step": 5370
    },
    {
      "epoch": 0.432475884244373,
      "grad_norm": 1.098965048789978,
      "learning_rate": 7.870056497175142e-05,
      "loss": 1.1328,
      "step": 5380
    },
    {
      "epoch": 0.4332797427652733,
      "grad_norm": 2.4418158531188965,
      "learning_rate": 7.866020984665052e-05,
      "loss": 1.169,
      "step": 5390
    },
    {
      "epoch": 0.4340836012861736,
      "grad_norm": 1.4988317489624023,
      "learning_rate": 7.861985472154964e-05,
      "loss": 1.1963,
      "step": 5400
    },
    {
      "epoch": 0.43488745980707394,
      "grad_norm": 1.1517032384872437,
      "learning_rate": 7.857949959644876e-05,
      "loss": 1.2728,
      "step": 5410
    },
    {
      "epoch": 0.43569131832797425,
      "grad_norm": 1.3836979866027832,
      "learning_rate": 7.853914447134786e-05,
      "loss": 1.2674,
      "step": 5420
    },
    {
      "epoch": 0.4364951768488746,
      "grad_norm": 1.0794776678085327,
      "learning_rate": 7.849878934624698e-05,
      "loss": 1.1862,
      "step": 5430
    },
    {
      "epoch": 0.43729903536977494,
      "grad_norm": 1.3821418285369873,
      "learning_rate": 7.845843422114609e-05,
      "loss": 1.2164,
      "step": 5440
    },
    {
      "epoch": 0.43810289389067525,
      "grad_norm": 1.051756739616394,
      "learning_rate": 7.84180790960452e-05,
      "loss": 1.2235,
      "step": 5450
    },
    {
      "epoch": 0.43890675241157556,
      "grad_norm": 1.1937540769577026,
      "learning_rate": 7.837772397094432e-05,
      "loss": 1.2094,
      "step": 5460
    },
    {
      "epoch": 0.4397106109324759,
      "grad_norm": 1.652544617652893,
      "learning_rate": 7.833736884584343e-05,
      "loss": 1.3222,
      "step": 5470
    },
    {
      "epoch": 0.4405144694533762,
      "grad_norm": 1.5041916370391846,
      "learning_rate": 7.829701372074255e-05,
      "loss": 1.1524,
      "step": 5480
    },
    {
      "epoch": 0.4413183279742765,
      "grad_norm": 2.3479011058807373,
      "learning_rate": 7.825665859564165e-05,
      "loss": 1.2593,
      "step": 5490
    },
    {
      "epoch": 0.44212218649517687,
      "grad_norm": 1.3248958587646484,
      "learning_rate": 7.821630347054077e-05,
      "loss": 1.1777,
      "step": 5500
    },
    {
      "epoch": 0.4429260450160772,
      "grad_norm": 1.0060685873031616,
      "learning_rate": 7.817594834543987e-05,
      "loss": 1.2084,
      "step": 5510
    },
    {
      "epoch": 0.4437299035369775,
      "grad_norm": 1.3419932126998901,
      "learning_rate": 7.813559322033899e-05,
      "loss": 1.1381,
      "step": 5520
    },
    {
      "epoch": 0.4445337620578778,
      "grad_norm": 1.8088581562042236,
      "learning_rate": 7.80952380952381e-05,
      "loss": 1.1101,
      "step": 5530
    },
    {
      "epoch": 0.4453376205787781,
      "grad_norm": 1.0692769289016724,
      "learning_rate": 7.805488297013721e-05,
      "loss": 1.2082,
      "step": 5540
    },
    {
      "epoch": 0.44614147909967844,
      "grad_norm": 1.071608066558838,
      "learning_rate": 7.801452784503632e-05,
      "loss": 1.165,
      "step": 5550
    },
    {
      "epoch": 0.44694533762057875,
      "grad_norm": 1.3941121101379395,
      "learning_rate": 7.797417271993544e-05,
      "loss": 1.2301,
      "step": 5560
    },
    {
      "epoch": 0.4477491961414791,
      "grad_norm": 1.166388750076294,
      "learning_rate": 7.793381759483454e-05,
      "loss": 1.1386,
      "step": 5570
    },
    {
      "epoch": 0.44855305466237944,
      "grad_norm": 1.871343731880188,
      "learning_rate": 7.789346246973366e-05,
      "loss": 1.2296,
      "step": 5580
    },
    {
      "epoch": 0.44935691318327975,
      "grad_norm": 1.4270875453948975,
      "learning_rate": 7.785310734463276e-05,
      "loss": 1.2229,
      "step": 5590
    },
    {
      "epoch": 0.45016077170418006,
      "grad_norm": 1.0972518920898438,
      "learning_rate": 7.781275221953188e-05,
      "loss": 1.1807,
      "step": 5600
    },
    {
      "epoch": 0.4509646302250804,
      "grad_norm": 2.3128669261932373,
      "learning_rate": 7.7772397094431e-05,
      "loss": 1.3243,
      "step": 5610
    },
    {
      "epoch": 0.4517684887459807,
      "grad_norm": 1.4300646781921387,
      "learning_rate": 7.77320419693301e-05,
      "loss": 1.2877,
      "step": 5620
    },
    {
      "epoch": 0.452572347266881,
      "grad_norm": 1.5658491849899292,
      "learning_rate": 7.769168684422922e-05,
      "loss": 1.206,
      "step": 5630
    },
    {
      "epoch": 0.4533762057877814,
      "grad_norm": 1.6786673069000244,
      "learning_rate": 7.765133171912833e-05,
      "loss": 1.1521,
      "step": 5640
    },
    {
      "epoch": 0.4541800643086817,
      "grad_norm": 1.7415214776992798,
      "learning_rate": 7.761097659402745e-05,
      "loss": 1.0989,
      "step": 5650
    },
    {
      "epoch": 0.454983922829582,
      "grad_norm": 1.5428780317306519,
      "learning_rate": 7.757062146892655e-05,
      "loss": 1.2544,
      "step": 5660
    },
    {
      "epoch": 0.4557877813504823,
      "grad_norm": 1.2579231262207031,
      "learning_rate": 7.753026634382567e-05,
      "loss": 1.2282,
      "step": 5670
    },
    {
      "epoch": 0.4565916398713826,
      "grad_norm": 1.019351601600647,
      "learning_rate": 7.748991121872479e-05,
      "loss": 1.1731,
      "step": 5680
    },
    {
      "epoch": 0.45739549839228294,
      "grad_norm": 1.2908672094345093,
      "learning_rate": 7.74495560936239e-05,
      "loss": 1.0881,
      "step": 5690
    },
    {
      "epoch": 0.45819935691318325,
      "grad_norm": 1.3027087450027466,
      "learning_rate": 7.740920096852301e-05,
      "loss": 1.2435,
      "step": 5700
    },
    {
      "epoch": 0.4590032154340836,
      "grad_norm": 1.603829264640808,
      "learning_rate": 7.736884584342213e-05,
      "loss": 1.2332,
      "step": 5710
    },
    {
      "epoch": 0.45980707395498394,
      "grad_norm": 1.3460705280303955,
      "learning_rate": 7.732849071832123e-05,
      "loss": 1.2056,
      "step": 5720
    },
    {
      "epoch": 0.46061093247588425,
      "grad_norm": 1.7570997476577759,
      "learning_rate": 7.728813559322035e-05,
      "loss": 1.2024,
      "step": 5730
    },
    {
      "epoch": 0.46141479099678456,
      "grad_norm": 1.3634543418884277,
      "learning_rate": 7.724778046811945e-05,
      "loss": 1.1674,
      "step": 5740
    },
    {
      "epoch": 0.4622186495176849,
      "grad_norm": 1.3680329322814941,
      "learning_rate": 7.720742534301857e-05,
      "loss": 1.2917,
      "step": 5750
    },
    {
      "epoch": 0.4630225080385852,
      "grad_norm": 1.585895299911499,
      "learning_rate": 7.716707021791768e-05,
      "loss": 1.202,
      "step": 5760
    },
    {
      "epoch": 0.4638263665594855,
      "grad_norm": 1.8435193300247192,
      "learning_rate": 7.71267150928168e-05,
      "loss": 1.2317,
      "step": 5770
    },
    {
      "epoch": 0.4646302250803859,
      "grad_norm": 0.9803140759468079,
      "learning_rate": 7.70863599677159e-05,
      "loss": 1.1616,
      "step": 5780
    },
    {
      "epoch": 0.4654340836012862,
      "grad_norm": 1.3502463102340698,
      "learning_rate": 7.704600484261502e-05,
      "loss": 1.2768,
      "step": 5790
    },
    {
      "epoch": 0.4662379421221865,
      "grad_norm": 1.094741940498352,
      "learning_rate": 7.700564971751412e-05,
      "loss": 1.1815,
      "step": 5800
    },
    {
      "epoch": 0.4670418006430868,
      "grad_norm": 1.1807562112808228,
      "learning_rate": 7.696529459241324e-05,
      "loss": 1.173,
      "step": 5810
    },
    {
      "epoch": 0.4678456591639871,
      "grad_norm": 1.4296817779541016,
      "learning_rate": 7.692493946731235e-05,
      "loss": 1.19,
      "step": 5820
    },
    {
      "epoch": 0.46864951768488744,
      "grad_norm": 1.4732331037521362,
      "learning_rate": 7.688458434221146e-05,
      "loss": 1.1519,
      "step": 5830
    },
    {
      "epoch": 0.4694533762057878,
      "grad_norm": 1.2672715187072754,
      "learning_rate": 7.684422921711057e-05,
      "loss": 1.1821,
      "step": 5840
    },
    {
      "epoch": 0.4702572347266881,
      "grad_norm": 1.078221321105957,
      "learning_rate": 7.680387409200969e-05,
      "loss": 1.1839,
      "step": 5850
    },
    {
      "epoch": 0.47106109324758844,
      "grad_norm": 1.2685880661010742,
      "learning_rate": 7.67635189669088e-05,
      "loss": 1.2756,
      "step": 5860
    },
    {
      "epoch": 0.47186495176848875,
      "grad_norm": 1.14828360080719,
      "learning_rate": 7.672316384180791e-05,
      "loss": 1.122,
      "step": 5870
    },
    {
      "epoch": 0.47266881028938906,
      "grad_norm": 1.5411911010742188,
      "learning_rate": 7.668280871670703e-05,
      "loss": 1.3135,
      "step": 5880
    },
    {
      "epoch": 0.4734726688102894,
      "grad_norm": 1.9456324577331543,
      "learning_rate": 7.664245359160613e-05,
      "loss": 1.3535,
      "step": 5890
    },
    {
      "epoch": 0.4742765273311897,
      "grad_norm": 1.542708396911621,
      "learning_rate": 7.660209846650525e-05,
      "loss": 1.3669,
      "step": 5900
    },
    {
      "epoch": 0.47508038585209006,
      "grad_norm": 1.6236876249313354,
      "learning_rate": 7.656174334140435e-05,
      "loss": 1.128,
      "step": 5910
    },
    {
      "epoch": 0.4758842443729904,
      "grad_norm": 1.0894272327423096,
      "learning_rate": 7.652138821630347e-05,
      "loss": 1.2741,
      "step": 5920
    },
    {
      "epoch": 0.4766881028938907,
      "grad_norm": 1.1134109497070312,
      "learning_rate": 7.648103309120259e-05,
      "loss": 1.2204,
      "step": 5930
    },
    {
      "epoch": 0.477491961414791,
      "grad_norm": 1.1840413808822632,
      "learning_rate": 7.644067796610171e-05,
      "loss": 1.12,
      "step": 5940
    },
    {
      "epoch": 0.4782958199356913,
      "grad_norm": 2.185009241104126,
      "learning_rate": 7.640032284100081e-05,
      "loss": 1.1607,
      "step": 5950
    },
    {
      "epoch": 0.4790996784565916,
      "grad_norm": 1.7973313331604004,
      "learning_rate": 7.635996771589993e-05,
      "loss": 1.1847,
      "step": 5960
    },
    {
      "epoch": 0.47990353697749194,
      "grad_norm": 1.4497991800308228,
      "learning_rate": 7.631961259079904e-05,
      "loss": 1.1776,
      "step": 5970
    },
    {
      "epoch": 0.4807073954983923,
      "grad_norm": 1.4984551668167114,
      "learning_rate": 7.627925746569815e-05,
      "loss": 1.2923,
      "step": 5980
    },
    {
      "epoch": 0.4815112540192926,
      "grad_norm": 1.1014914512634277,
      "learning_rate": 7.623890234059726e-05,
      "loss": 1.0993,
      "step": 5990
    },
    {
      "epoch": 0.48231511254019294,
      "grad_norm": 1.5178377628326416,
      "learning_rate": 7.619854721549638e-05,
      "loss": 1.161,
      "step": 6000
    },
    {
      "epoch": 0.48231511254019294,
      "eval_yahma/alpaca-cleaned_loss": 1.2380210161209106,
      "eval_yahma/alpaca-cleaned_runtime": 115.6688,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.291,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.161,
      "step": 6000
    },
    {
      "epoch": 0.48311897106109325,
      "grad_norm": 1.408656120300293,
      "learning_rate": 7.615819209039548e-05,
      "loss": 1.2989,
      "step": 6010
    },
    {
      "epoch": 0.48392282958199356,
      "grad_norm": 1.2989753484725952,
      "learning_rate": 7.61178369652946e-05,
      "loss": 1.2565,
      "step": 6020
    },
    {
      "epoch": 0.4847266881028939,
      "grad_norm": 1.5819889307022095,
      "learning_rate": 7.60774818401937e-05,
      "loss": 1.1385,
      "step": 6030
    },
    {
      "epoch": 0.4855305466237942,
      "grad_norm": 1.1295589208602905,
      "learning_rate": 7.603712671509282e-05,
      "loss": 1.203,
      "step": 6040
    },
    {
      "epoch": 0.48633440514469456,
      "grad_norm": 1.2585210800170898,
      "learning_rate": 7.599677158999193e-05,
      "loss": 1.3078,
      "step": 6050
    },
    {
      "epoch": 0.4871382636655949,
      "grad_norm": 1.024665355682373,
      "learning_rate": 7.595641646489105e-05,
      "loss": 1.2181,
      "step": 6060
    },
    {
      "epoch": 0.4879421221864952,
      "grad_norm": 1.289249062538147,
      "learning_rate": 7.591606133979015e-05,
      "loss": 1.3394,
      "step": 6070
    },
    {
      "epoch": 0.4887459807073955,
      "grad_norm": 3.0227270126342773,
      "learning_rate": 7.587570621468927e-05,
      "loss": 1.1185,
      "step": 6080
    },
    {
      "epoch": 0.4895498392282958,
      "grad_norm": 1.325408935546875,
      "learning_rate": 7.583535108958839e-05,
      "loss": 1.2544,
      "step": 6090
    },
    {
      "epoch": 0.4903536977491961,
      "grad_norm": 1.8913363218307495,
      "learning_rate": 7.579499596448749e-05,
      "loss": 1.2673,
      "step": 6100
    },
    {
      "epoch": 0.49115755627009644,
      "grad_norm": 1.3949637413024902,
      "learning_rate": 7.575464083938661e-05,
      "loss": 1.1629,
      "step": 6110
    },
    {
      "epoch": 0.4919614147909968,
      "grad_norm": 1.7149893045425415,
      "learning_rate": 7.571428571428571e-05,
      "loss": 1.3161,
      "step": 6120
    },
    {
      "epoch": 0.4927652733118971,
      "grad_norm": 1.4487000703811646,
      "learning_rate": 7.567393058918483e-05,
      "loss": 1.1248,
      "step": 6130
    },
    {
      "epoch": 0.49356913183279744,
      "grad_norm": 1.4671707153320312,
      "learning_rate": 7.563357546408394e-05,
      "loss": 1.1904,
      "step": 6140
    },
    {
      "epoch": 0.49437299035369775,
      "grad_norm": 1.4180289506912231,
      "learning_rate": 7.559322033898305e-05,
      "loss": 1.1742,
      "step": 6150
    },
    {
      "epoch": 0.49517684887459806,
      "grad_norm": 1.335613489151001,
      "learning_rate": 7.555286521388216e-05,
      "loss": 1.19,
      "step": 6160
    },
    {
      "epoch": 0.4959807073954984,
      "grad_norm": 1.493781566619873,
      "learning_rate": 7.551251008878128e-05,
      "loss": 1.249,
      "step": 6170
    },
    {
      "epoch": 0.4967845659163987,
      "grad_norm": 1.7311235666275024,
      "learning_rate": 7.547215496368038e-05,
      "loss": 1.2581,
      "step": 6180
    },
    {
      "epoch": 0.49758842443729906,
      "grad_norm": 1.3849860429763794,
      "learning_rate": 7.54317998385795e-05,
      "loss": 1.2069,
      "step": 6190
    },
    {
      "epoch": 0.4983922829581994,
      "grad_norm": 1.0231637954711914,
      "learning_rate": 7.539144471347862e-05,
      "loss": 1.1676,
      "step": 6200
    },
    {
      "epoch": 0.4991961414790997,
      "grad_norm": 1.1450506448745728,
      "learning_rate": 7.535108958837774e-05,
      "loss": 1.2508,
      "step": 6210
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.203847885131836,
      "learning_rate": 7.531073446327684e-05,
      "loss": 1.3101,
      "step": 6220
    },
    {
      "epoch": 0.5008038585209004,
      "grad_norm": 1.2034550905227661,
      "learning_rate": 7.527037933817596e-05,
      "loss": 1.2576,
      "step": 6230
    },
    {
      "epoch": 0.5016077170418006,
      "grad_norm": 1.3764240741729736,
      "learning_rate": 7.523002421307506e-05,
      "loss": 1.1927,
      "step": 6240
    },
    {
      "epoch": 0.502411575562701,
      "grad_norm": 1.907971978187561,
      "learning_rate": 7.518966908797418e-05,
      "loss": 1.2045,
      "step": 6250
    },
    {
      "epoch": 0.5032154340836013,
      "grad_norm": 1.417354702949524,
      "learning_rate": 7.514931396287329e-05,
      "loss": 1.2857,
      "step": 6260
    },
    {
      "epoch": 0.5040192926045016,
      "grad_norm": 1.4104013442993164,
      "learning_rate": 7.51089588377724e-05,
      "loss": 1.2178,
      "step": 6270
    },
    {
      "epoch": 0.5048231511254019,
      "grad_norm": 1.3243697881698608,
      "learning_rate": 7.506860371267151e-05,
      "loss": 1.2072,
      "step": 6280
    },
    {
      "epoch": 0.5056270096463023,
      "grad_norm": 1.7258855104446411,
      "learning_rate": 7.502824858757063e-05,
      "loss": 1.2003,
      "step": 6290
    },
    {
      "epoch": 0.5064308681672026,
      "grad_norm": 1.7383719682693481,
      "learning_rate": 7.498789346246973e-05,
      "loss": 1.21,
      "step": 6300
    },
    {
      "epoch": 0.5072347266881029,
      "grad_norm": 1.1666938066482544,
      "learning_rate": 7.494753833736885e-05,
      "loss": 1.2231,
      "step": 6310
    },
    {
      "epoch": 0.5080385852090032,
      "grad_norm": 1.3047698736190796,
      "learning_rate": 7.491121872477805e-05,
      "loss": 1.025,
      "step": 6320
    },
    {
      "epoch": 0.5088424437299035,
      "grad_norm": 2.041975259780884,
      "learning_rate": 7.487086359967717e-05,
      "loss": 1.1173,
      "step": 6330
    },
    {
      "epoch": 0.5096463022508039,
      "grad_norm": 1.8618550300598145,
      "learning_rate": 7.483050847457627e-05,
      "loss": 1.1592,
      "step": 6340
    },
    {
      "epoch": 0.5104501607717041,
      "grad_norm": 1.1699954271316528,
      "learning_rate": 7.479015334947539e-05,
      "loss": 1.0528,
      "step": 6350
    },
    {
      "epoch": 0.5112540192926045,
      "grad_norm": 1.319175362586975,
      "learning_rate": 7.47497982243745e-05,
      "loss": 1.199,
      "step": 6360
    },
    {
      "epoch": 0.5120578778135049,
      "grad_norm": 1.1470156908035278,
      "learning_rate": 7.470944309927361e-05,
      "loss": 1.0411,
      "step": 6370
    },
    {
      "epoch": 0.5128617363344051,
      "grad_norm": 1.7862601280212402,
      "learning_rate": 7.466908797417272e-05,
      "loss": 1.2587,
      "step": 6380
    },
    {
      "epoch": 0.5136655948553055,
      "grad_norm": 1.924744963645935,
      "learning_rate": 7.462873284907184e-05,
      "loss": 1.2351,
      "step": 6390
    },
    {
      "epoch": 0.5144694533762058,
      "grad_norm": 1.4155441522598267,
      "learning_rate": 7.458837772397094e-05,
      "loss": 1.1327,
      "step": 6400
    },
    {
      "epoch": 0.5152733118971061,
      "grad_norm": 1.2903022766113281,
      "learning_rate": 7.455205811138016e-05,
      "loss": 1.0905,
      "step": 6410
    },
    {
      "epoch": 0.5160771704180064,
      "grad_norm": 1.745773434638977,
      "learning_rate": 7.451170298627926e-05,
      "loss": 1.1903,
      "step": 6420
    },
    {
      "epoch": 0.5168810289389068,
      "grad_norm": 1.2349950075149536,
      "learning_rate": 7.447134786117838e-05,
      "loss": 1.1813,
      "step": 6430
    },
    {
      "epoch": 0.5176848874598071,
      "grad_norm": 1.8500481843948364,
      "learning_rate": 7.443099273607748e-05,
      "loss": 1.1865,
      "step": 6440
    },
    {
      "epoch": 0.5184887459807074,
      "grad_norm": 1.355523705482483,
      "learning_rate": 7.43906376109766e-05,
      "loss": 1.1611,
      "step": 6450
    },
    {
      "epoch": 0.5192926045016077,
      "grad_norm": 1.3113207817077637,
      "learning_rate": 7.43502824858757e-05,
      "loss": 1.1685,
      "step": 6460
    },
    {
      "epoch": 0.520096463022508,
      "grad_norm": 1.3658199310302734,
      "learning_rate": 7.430992736077482e-05,
      "loss": 1.2853,
      "step": 6470
    },
    {
      "epoch": 0.5209003215434084,
      "grad_norm": 1.1712766885757446,
      "learning_rate": 7.426957223567393e-05,
      "loss": 1.2977,
      "step": 6480
    },
    {
      "epoch": 0.5217041800643086,
      "grad_norm": 1.9549909830093384,
      "learning_rate": 7.422921711057305e-05,
      "loss": 1.2796,
      "step": 6490
    },
    {
      "epoch": 0.522508038585209,
      "grad_norm": 2.5356125831604004,
      "learning_rate": 7.418886198547215e-05,
      "loss": 1.143,
      "step": 6500
    },
    {
      "epoch": 0.5233118971061094,
      "grad_norm": 1.2105944156646729,
      "learning_rate": 7.414850686037127e-05,
      "loss": 1.3304,
      "step": 6510
    },
    {
      "epoch": 0.5241157556270096,
      "grad_norm": 0.9766327738761902,
      "learning_rate": 7.410815173527039e-05,
      "loss": 1.2569,
      "step": 6520
    },
    {
      "epoch": 0.52491961414791,
      "grad_norm": 1.3324644565582275,
      "learning_rate": 7.406779661016949e-05,
      "loss": 1.1962,
      "step": 6530
    },
    {
      "epoch": 0.5257234726688103,
      "grad_norm": 1.4825365543365479,
      "learning_rate": 7.402744148506861e-05,
      "loss": 1.182,
      "step": 6540
    },
    {
      "epoch": 0.5265273311897106,
      "grad_norm": 1.0984454154968262,
      "learning_rate": 7.398708635996772e-05,
      "loss": 1.325,
      "step": 6550
    },
    {
      "epoch": 0.5273311897106109,
      "grad_norm": 1.4700428247451782,
      "learning_rate": 7.394673123486683e-05,
      "loss": 1.1464,
      "step": 6560
    },
    {
      "epoch": 0.5281350482315113,
      "grad_norm": 1.6935198307037354,
      "learning_rate": 7.390637610976594e-05,
      "loss": 1.2638,
      "step": 6570
    },
    {
      "epoch": 0.5289389067524116,
      "grad_norm": 1.5501216650009155,
      "learning_rate": 7.386602098466506e-05,
      "loss": 1.1705,
      "step": 6580
    },
    {
      "epoch": 0.5297427652733119,
      "grad_norm": 1.5348395109176636,
      "learning_rate": 7.382566585956416e-05,
      "loss": 1.249,
      "step": 6590
    },
    {
      "epoch": 0.5305466237942122,
      "grad_norm": 1.548414945602417,
      "learning_rate": 7.378531073446328e-05,
      "loss": 1.2001,
      "step": 6600
    },
    {
      "epoch": 0.5313504823151125,
      "grad_norm": 1.1692911386489868,
      "learning_rate": 7.374495560936238e-05,
      "loss": 1.0845,
      "step": 6610
    },
    {
      "epoch": 0.5321543408360129,
      "grad_norm": 2.0657691955566406,
      "learning_rate": 7.370460048426152e-05,
      "loss": 1.4335,
      "step": 6620
    },
    {
      "epoch": 0.5329581993569131,
      "grad_norm": 1.2757883071899414,
      "learning_rate": 7.366424535916062e-05,
      "loss": 1.2333,
      "step": 6630
    },
    {
      "epoch": 0.5337620578778135,
      "grad_norm": 1.6110565662384033,
      "learning_rate": 7.362389023405974e-05,
      "loss": 1.1881,
      "step": 6640
    },
    {
      "epoch": 0.5345659163987139,
      "grad_norm": 1.3758124113082886,
      "learning_rate": 7.358353510895884e-05,
      "loss": 1.2932,
      "step": 6650
    },
    {
      "epoch": 0.5353697749196141,
      "grad_norm": 1.5866690874099731,
      "learning_rate": 7.354317998385796e-05,
      "loss": 1.2757,
      "step": 6660
    },
    {
      "epoch": 0.5361736334405145,
      "grad_norm": 1.281878113746643,
      "learning_rate": 7.350282485875707e-05,
      "loss": 1.2987,
      "step": 6670
    },
    {
      "epoch": 0.5369774919614148,
      "grad_norm": 1.4316363334655762,
      "learning_rate": 7.346246973365618e-05,
      "loss": 1.2868,
      "step": 6680
    },
    {
      "epoch": 0.5377813504823151,
      "grad_norm": 1.2454899549484253,
      "learning_rate": 7.342211460855529e-05,
      "loss": 1.2473,
      "step": 6690
    },
    {
      "epoch": 0.5385852090032154,
      "grad_norm": 1.4715195894241333,
      "learning_rate": 7.33817594834544e-05,
      "loss": 1.2471,
      "step": 6700
    },
    {
      "epoch": 0.5393890675241158,
      "grad_norm": 1.0020066499710083,
      "learning_rate": 7.334140435835351e-05,
      "loss": 1.2116,
      "step": 6710
    },
    {
      "epoch": 0.5401929260450161,
      "grad_norm": 1.8996168375015259,
      "learning_rate": 7.330104923325263e-05,
      "loss": 1.1744,
      "step": 6720
    },
    {
      "epoch": 0.5409967845659164,
      "grad_norm": 1.3301576375961304,
      "learning_rate": 7.326069410815173e-05,
      "loss": 1.2056,
      "step": 6730
    },
    {
      "epoch": 0.5418006430868167,
      "grad_norm": 1.354276180267334,
      "learning_rate": 7.322033898305085e-05,
      "loss": 1.2008,
      "step": 6740
    },
    {
      "epoch": 0.542604501607717,
      "grad_norm": 1.4722867012023926,
      "learning_rate": 7.317998385794997e-05,
      "loss": 1.2248,
      "step": 6750
    },
    {
      "epoch": 0.5434083601286174,
      "grad_norm": 1.3259929418563843,
      "learning_rate": 7.313962873284907e-05,
      "loss": 1.2193,
      "step": 6760
    },
    {
      "epoch": 0.5442122186495176,
      "grad_norm": 1.7194771766662598,
      "learning_rate": 7.309927360774819e-05,
      "loss": 1.2812,
      "step": 6770
    },
    {
      "epoch": 0.545016077170418,
      "grad_norm": 1.4885749816894531,
      "learning_rate": 7.30589184826473e-05,
      "loss": 1.0297,
      "step": 6780
    },
    {
      "epoch": 0.5458199356913184,
      "grad_norm": 1.3488175868988037,
      "learning_rate": 7.301856335754641e-05,
      "loss": 1.2759,
      "step": 6790
    },
    {
      "epoch": 0.5466237942122186,
      "grad_norm": 1.410971760749817,
      "learning_rate": 7.297820823244552e-05,
      "loss": 1.1545,
      "step": 6800
    },
    {
      "epoch": 0.547427652733119,
      "grad_norm": 1.3579992055892944,
      "learning_rate": 7.293785310734464e-05,
      "loss": 1.219,
      "step": 6810
    },
    {
      "epoch": 0.5482315112540193,
      "grad_norm": 1.5444881916046143,
      "learning_rate": 7.289749798224374e-05,
      "loss": 1.1847,
      "step": 6820
    },
    {
      "epoch": 0.5490353697749196,
      "grad_norm": 1.577883005142212,
      "learning_rate": 7.285714285714286e-05,
      "loss": 1.103,
      "step": 6830
    },
    {
      "epoch": 0.5498392282958199,
      "grad_norm": 1.9665368795394897,
      "learning_rate": 7.281678773204197e-05,
      "loss": 1.1444,
      "step": 6840
    },
    {
      "epoch": 0.5506430868167203,
      "grad_norm": 1.849618673324585,
      "learning_rate": 7.277643260694108e-05,
      "loss": 1.3576,
      "step": 6850
    },
    {
      "epoch": 0.5514469453376206,
      "grad_norm": 1.6548022031784058,
      "learning_rate": 7.273607748184019e-05,
      "loss": 1.2216,
      "step": 6860
    },
    {
      "epoch": 0.5522508038585209,
      "grad_norm": 1.2869874238967896,
      "learning_rate": 7.26957223567393e-05,
      "loss": 1.2912,
      "step": 6870
    },
    {
      "epoch": 0.5530546623794212,
      "grad_norm": 2.3842945098876953,
      "learning_rate": 7.265536723163842e-05,
      "loss": 1.2517,
      "step": 6880
    },
    {
      "epoch": 0.5538585209003215,
      "grad_norm": 1.285078525543213,
      "learning_rate": 7.261501210653754e-05,
      "loss": 1.27,
      "step": 6890
    },
    {
      "epoch": 0.5546623794212219,
      "grad_norm": 1.1571310758590698,
      "learning_rate": 7.257465698143665e-05,
      "loss": 1.2052,
      "step": 6900
    },
    {
      "epoch": 0.5554662379421221,
      "grad_norm": 3.9599387645721436,
      "learning_rate": 7.253430185633576e-05,
      "loss": 1.1625,
      "step": 6910
    },
    {
      "epoch": 0.5562700964630225,
      "grad_norm": 1.3315590620040894,
      "learning_rate": 7.249394673123487e-05,
      "loss": 1.129,
      "step": 6920
    },
    {
      "epoch": 0.5570739549839229,
      "grad_norm": 1.445112705230713,
      "learning_rate": 7.245359160613399e-05,
      "loss": 1.2613,
      "step": 6930
    },
    {
      "epoch": 0.5578778135048231,
      "grad_norm": 1.5418325662612915,
      "learning_rate": 7.241323648103309e-05,
      "loss": 1.17,
      "step": 6940
    },
    {
      "epoch": 0.5586816720257235,
      "grad_norm": 1.247528314590454,
      "learning_rate": 7.237288135593221e-05,
      "loss": 1.2158,
      "step": 6950
    },
    {
      "epoch": 0.5594855305466238,
      "grad_norm": 2.0513546466827393,
      "learning_rate": 7.233252623083131e-05,
      "loss": 1.1665,
      "step": 6960
    },
    {
      "epoch": 0.5602893890675241,
      "grad_norm": 1.661195158958435,
      "learning_rate": 7.229217110573043e-05,
      "loss": 1.2148,
      "step": 6970
    },
    {
      "epoch": 0.5610932475884244,
      "grad_norm": 1.1958410739898682,
      "learning_rate": 7.225181598062955e-05,
      "loss": 1.175,
      "step": 6980
    },
    {
      "epoch": 0.5618971061093248,
      "grad_norm": 1.3750405311584473,
      "learning_rate": 7.221146085552866e-05,
      "loss": 1.1609,
      "step": 6990
    },
    {
      "epoch": 0.5627009646302251,
      "grad_norm": 1.5093348026275635,
      "learning_rate": 7.217110573042777e-05,
      "loss": 1.1352,
      "step": 7000
    },
    {
      "epoch": 0.5635048231511254,
      "grad_norm": 2.5393998622894287,
      "learning_rate": 7.213075060532688e-05,
      "loss": 1.1646,
      "step": 7010
    },
    {
      "epoch": 0.5643086816720257,
      "grad_norm": 2.845985174179077,
      "learning_rate": 7.2090395480226e-05,
      "loss": 1.1089,
      "step": 7020
    },
    {
      "epoch": 0.565112540192926,
      "grad_norm": 1.2273420095443726,
      "learning_rate": 7.20500403551251e-05,
      "loss": 1.3287,
      "step": 7030
    },
    {
      "epoch": 0.5659163987138264,
      "grad_norm": 1.1808136701583862,
      "learning_rate": 7.200968523002422e-05,
      "loss": 1.2131,
      "step": 7040
    },
    {
      "epoch": 0.5667202572347267,
      "grad_norm": 1.4661297798156738,
      "learning_rate": 7.196933010492332e-05,
      "loss": 1.2178,
      "step": 7050
    },
    {
      "epoch": 0.567524115755627,
      "grad_norm": 1.872639775276184,
      "learning_rate": 7.192897497982244e-05,
      "loss": 1.1569,
      "step": 7060
    },
    {
      "epoch": 0.5683279742765274,
      "grad_norm": 1.1631591320037842,
      "learning_rate": 7.188861985472155e-05,
      "loss": 1.2776,
      "step": 7070
    },
    {
      "epoch": 0.5691318327974276,
      "grad_norm": 1.7431219816207886,
      "learning_rate": 7.184826472962066e-05,
      "loss": 1.116,
      "step": 7080
    },
    {
      "epoch": 0.569935691318328,
      "grad_norm": 1.3899308443069458,
      "learning_rate": 7.180790960451977e-05,
      "loss": 1.2137,
      "step": 7090
    },
    {
      "epoch": 0.5707395498392283,
      "grad_norm": 1.6811970472335815,
      "learning_rate": 7.176755447941889e-05,
      "loss": 1.2248,
      "step": 7100
    },
    {
      "epoch": 0.5715434083601286,
      "grad_norm": 1.0294607877731323,
      "learning_rate": 7.172719935431799e-05,
      "loss": 1.2298,
      "step": 7110
    },
    {
      "epoch": 0.572347266881029,
      "grad_norm": 1.3149453401565552,
      "learning_rate": 7.168684422921711e-05,
      "loss": 1.162,
      "step": 7120
    },
    {
      "epoch": 0.5731511254019293,
      "grad_norm": 1.3619112968444824,
      "learning_rate": 7.164648910411621e-05,
      "loss": 1.1935,
      "step": 7130
    },
    {
      "epoch": 0.5739549839228296,
      "grad_norm": 1.2184911966323853,
      "learning_rate": 7.160613397901533e-05,
      "loss": 1.0804,
      "step": 7140
    },
    {
      "epoch": 0.5747588424437299,
      "grad_norm": 1.170629858970642,
      "learning_rate": 7.156577885391445e-05,
      "loss": 1.1793,
      "step": 7150
    },
    {
      "epoch": 0.5755627009646302,
      "grad_norm": 1.2444965839385986,
      "learning_rate": 7.152542372881357e-05,
      "loss": 1.2071,
      "step": 7160
    },
    {
      "epoch": 0.5763665594855305,
      "grad_norm": 1.3081021308898926,
      "learning_rate": 7.148506860371267e-05,
      "loss": 1.1656,
      "step": 7170
    },
    {
      "epoch": 0.5771704180064309,
      "grad_norm": 2.752758264541626,
      "learning_rate": 7.144471347861179e-05,
      "loss": 1.2164,
      "step": 7180
    },
    {
      "epoch": 0.5779742765273312,
      "grad_norm": 1.5479896068572998,
      "learning_rate": 7.14043583535109e-05,
      "loss": 1.1613,
      "step": 7190
    },
    {
      "epoch": 0.5787781350482315,
      "grad_norm": 1.40146803855896,
      "learning_rate": 7.136400322841001e-05,
      "loss": 1.1026,
      "step": 7200
    },
    {
      "epoch": 0.5795819935691319,
      "grad_norm": 1.0006448030471802,
      "learning_rate": 7.132364810330913e-05,
      "loss": 1.1749,
      "step": 7210
    },
    {
      "epoch": 0.5803858520900321,
      "grad_norm": 1.4498153924942017,
      "learning_rate": 7.128329297820824e-05,
      "loss": 1.065,
      "step": 7220
    },
    {
      "epoch": 0.5811897106109325,
      "grad_norm": 1.3882859945297241,
      "learning_rate": 7.124293785310736e-05,
      "loss": 1.248,
      "step": 7230
    },
    {
      "epoch": 0.5819935691318328,
      "grad_norm": 0.9907482266426086,
      "learning_rate": 7.120258272800646e-05,
      "loss": 1.1794,
      "step": 7240
    },
    {
      "epoch": 0.5827974276527331,
      "grad_norm": 2.1010050773620605,
      "learning_rate": 7.116222760290558e-05,
      "loss": 1.1451,
      "step": 7250
    },
    {
      "epoch": 0.5836012861736335,
      "grad_norm": 1.1258735656738281,
      "learning_rate": 7.112187247780468e-05,
      "loss": 1.2021,
      "step": 7260
    },
    {
      "epoch": 0.5844051446945338,
      "grad_norm": 1.8240938186645508,
      "learning_rate": 7.10815173527038e-05,
      "loss": 1.2344,
      "step": 7270
    },
    {
      "epoch": 0.5852090032154341,
      "grad_norm": 2.2731316089630127,
      "learning_rate": 7.10411622276029e-05,
      "loss": 1.1984,
      "step": 7280
    },
    {
      "epoch": 0.5860128617363344,
      "grad_norm": 1.6902538537979126,
      "learning_rate": 7.100080710250202e-05,
      "loss": 1.2406,
      "step": 7290
    },
    {
      "epoch": 0.5868167202572347,
      "grad_norm": 1.1986221075057983,
      "learning_rate": 7.096045197740113e-05,
      "loss": 1.2479,
      "step": 7300
    },
    {
      "epoch": 0.587620578778135,
      "grad_norm": 1.3120813369750977,
      "learning_rate": 7.092009685230025e-05,
      "loss": 1.2603,
      "step": 7310
    },
    {
      "epoch": 0.5884244372990354,
      "grad_norm": 1.1902378797531128,
      "learning_rate": 7.087974172719935e-05,
      "loss": 1.1116,
      "step": 7320
    },
    {
      "epoch": 0.5892282958199357,
      "grad_norm": 1.3847696781158447,
      "learning_rate": 7.083938660209847e-05,
      "loss": 1.2127,
      "step": 7330
    },
    {
      "epoch": 0.590032154340836,
      "grad_norm": 2.2988333702087402,
      "learning_rate": 7.079903147699757e-05,
      "loss": 1.1229,
      "step": 7340
    },
    {
      "epoch": 0.5908360128617364,
      "grad_norm": 1.5824936628341675,
      "learning_rate": 7.075867635189669e-05,
      "loss": 1.333,
      "step": 7350
    },
    {
      "epoch": 0.5916398713826366,
      "grad_norm": 1.1853426694869995,
      "learning_rate": 7.07183212267958e-05,
      "loss": 1.1416,
      "step": 7360
    },
    {
      "epoch": 0.592443729903537,
      "grad_norm": 1.1010386943817139,
      "learning_rate": 7.067796610169491e-05,
      "loss": 1.3182,
      "step": 7370
    },
    {
      "epoch": 0.5932475884244373,
      "grad_norm": 1.2553712129592896,
      "learning_rate": 7.063761097659403e-05,
      "loss": 1.2814,
      "step": 7380
    },
    {
      "epoch": 0.5940514469453376,
      "grad_norm": 1.5219231843948364,
      "learning_rate": 7.059725585149314e-05,
      "loss": 1.2996,
      "step": 7390
    },
    {
      "epoch": 0.594855305466238,
      "grad_norm": 1.8472049236297607,
      "learning_rate": 7.055690072639226e-05,
      "loss": 1.1994,
      "step": 7400
    },
    {
      "epoch": 0.5956591639871383,
      "grad_norm": 1.3701444864273071,
      "learning_rate": 7.051654560129137e-05,
      "loss": 1.1645,
      "step": 7410
    },
    {
      "epoch": 0.5964630225080386,
      "grad_norm": 1.3976223468780518,
      "learning_rate": 7.047619047619048e-05,
      "loss": 1.1904,
      "step": 7420
    },
    {
      "epoch": 0.5972668810289389,
      "grad_norm": 2.294698476791382,
      "learning_rate": 7.04358353510896e-05,
      "loss": 1.2541,
      "step": 7430
    },
    {
      "epoch": 0.5980707395498392,
      "grad_norm": 1.3611372709274292,
      "learning_rate": 7.039548022598871e-05,
      "loss": 1.1069,
      "step": 7440
    },
    {
      "epoch": 0.5988745980707395,
      "grad_norm": 1.5398073196411133,
      "learning_rate": 7.035512510088782e-05,
      "loss": 1.1596,
      "step": 7450
    },
    {
      "epoch": 0.5996784565916399,
      "grad_norm": 1.7317296266555786,
      "learning_rate": 7.031476997578694e-05,
      "loss": 1.241,
      "step": 7460
    },
    {
      "epoch": 0.6004823151125402,
      "grad_norm": 1.5638251304626465,
      "learning_rate": 7.027441485068604e-05,
      "loss": 1.2383,
      "step": 7470
    },
    {
      "epoch": 0.6012861736334405,
      "grad_norm": 1.892224669456482,
      "learning_rate": 7.023405972558516e-05,
      "loss": 1.025,
      "step": 7480
    },
    {
      "epoch": 0.6020900321543409,
      "grad_norm": 1.6151889562606812,
      "learning_rate": 7.019370460048426e-05,
      "loss": 1.1827,
      "step": 7490
    },
    {
      "epoch": 0.6028938906752411,
      "grad_norm": 1.3697775602340698,
      "learning_rate": 7.015334947538338e-05,
      "loss": 1.2432,
      "step": 7500
    },
    {
      "epoch": 0.6036977491961415,
      "grad_norm": 1.0278207063674927,
      "learning_rate": 7.011299435028249e-05,
      "loss": 1.3371,
      "step": 7510
    },
    {
      "epoch": 0.6045016077170418,
      "grad_norm": 1.348399043083191,
      "learning_rate": 7.00726392251816e-05,
      "loss": 1.0785,
      "step": 7520
    },
    {
      "epoch": 0.6053054662379421,
      "grad_norm": 1.1832696199417114,
      "learning_rate": 7.003228410008071e-05,
      "loss": 1.1552,
      "step": 7530
    },
    {
      "epoch": 0.6061093247588425,
      "grad_norm": 1.6915167570114136,
      "learning_rate": 6.999192897497983e-05,
      "loss": 1.2163,
      "step": 7540
    },
    {
      "epoch": 0.6069131832797428,
      "grad_norm": 1.2538834810256958,
      "learning_rate": 6.995157384987893e-05,
      "loss": 1.2105,
      "step": 7550
    },
    {
      "epoch": 0.6077170418006431,
      "grad_norm": 1.0548170804977417,
      "learning_rate": 6.991121872477805e-05,
      "loss": 1.2295,
      "step": 7560
    },
    {
      "epoch": 0.6085209003215434,
      "grad_norm": 1.6791824102401733,
      "learning_rate": 6.987086359967716e-05,
      "loss": 1.1932,
      "step": 7570
    },
    {
      "epoch": 0.6093247588424437,
      "grad_norm": 1.7227991819381714,
      "learning_rate": 6.983050847457627e-05,
      "loss": 1.2275,
      "step": 7580
    },
    {
      "epoch": 0.610128617363344,
      "grad_norm": 1.4850959777832031,
      "learning_rate": 6.979015334947538e-05,
      "loss": 1.1649,
      "step": 7590
    },
    {
      "epoch": 0.6109324758842444,
      "grad_norm": 1.2839738130569458,
      "learning_rate": 6.97497982243745e-05,
      "loss": 1.2562,
      "step": 7600
    },
    {
      "epoch": 0.6117363344051447,
      "grad_norm": 1.845155119895935,
      "learning_rate": 6.970944309927361e-05,
      "loss": 1.297,
      "step": 7610
    },
    {
      "epoch": 0.612540192926045,
      "grad_norm": 1.21802818775177,
      "learning_rate": 6.966908797417272e-05,
      "loss": 1.2947,
      "step": 7620
    },
    {
      "epoch": 0.6133440514469454,
      "grad_norm": 1.228559136390686,
      "learning_rate": 6.962873284907184e-05,
      "loss": 1.3151,
      "step": 7630
    },
    {
      "epoch": 0.6141479099678456,
      "grad_norm": 1.5754231214523315,
      "learning_rate": 6.958837772397094e-05,
      "loss": 1.1846,
      "step": 7640
    },
    {
      "epoch": 0.614951768488746,
      "grad_norm": 1.7374244928359985,
      "learning_rate": 6.954802259887006e-05,
      "loss": 1.2737,
      "step": 7650
    },
    {
      "epoch": 0.6157556270096463,
      "grad_norm": 2.792844533920288,
      "learning_rate": 6.950766747376916e-05,
      "loss": 1.1331,
      "step": 7660
    },
    {
      "epoch": 0.6165594855305466,
      "grad_norm": 1.3875277042388916,
      "learning_rate": 6.94673123486683e-05,
      "loss": 1.2218,
      "step": 7670
    },
    {
      "epoch": 0.617363344051447,
      "grad_norm": 1.1979440450668335,
      "learning_rate": 6.94269572235674e-05,
      "loss": 1.1648,
      "step": 7680
    },
    {
      "epoch": 0.6181672025723473,
      "grad_norm": 1.5823277235031128,
      "learning_rate": 6.938660209846652e-05,
      "loss": 1.249,
      "step": 7690
    },
    {
      "epoch": 0.6189710610932476,
      "grad_norm": 1.26264226436615,
      "learning_rate": 6.934624697336562e-05,
      "loss": 1.2724,
      "step": 7700
    },
    {
      "epoch": 0.6197749196141479,
      "grad_norm": 1.2197229862213135,
      "learning_rate": 6.930589184826474e-05,
      "loss": 1.2324,
      "step": 7710
    },
    {
      "epoch": 0.6205787781350482,
      "grad_norm": 1.3479337692260742,
      "learning_rate": 6.926553672316385e-05,
      "loss": 1.1723,
      "step": 7720
    },
    {
      "epoch": 0.6213826366559485,
      "grad_norm": 4.150685787200928,
      "learning_rate": 6.922518159806296e-05,
      "loss": 1.1546,
      "step": 7730
    },
    {
      "epoch": 0.6221864951768489,
      "grad_norm": 1.5650343894958496,
      "learning_rate": 6.918482647296207e-05,
      "loss": 1.1733,
      "step": 7740
    },
    {
      "epoch": 0.6229903536977492,
      "grad_norm": 1.296133279800415,
      "learning_rate": 6.914447134786119e-05,
      "loss": 1.0567,
      "step": 7750
    },
    {
      "epoch": 0.6237942122186495,
      "grad_norm": 1.382730484008789,
      "learning_rate": 6.910411622276029e-05,
      "loss": 1.3209,
      "step": 7760
    },
    {
      "epoch": 0.6245980707395499,
      "grad_norm": 1.6270509958267212,
      "learning_rate": 6.906376109765941e-05,
      "loss": 1.1816,
      "step": 7770
    },
    {
      "epoch": 0.6254019292604501,
      "grad_norm": 1.0578793287277222,
      "learning_rate": 6.902340597255851e-05,
      "loss": 1.15,
      "step": 7780
    },
    {
      "epoch": 0.6262057877813505,
      "grad_norm": 1.2071346044540405,
      "learning_rate": 6.898305084745763e-05,
      "loss": 1.2732,
      "step": 7790
    },
    {
      "epoch": 0.6270096463022508,
      "grad_norm": 1.1046452522277832,
      "learning_rate": 6.894269572235674e-05,
      "loss": 1.0907,
      "step": 7800
    },
    {
      "epoch": 0.6278135048231511,
      "grad_norm": 2.797956943511963,
      "learning_rate": 6.890234059725586e-05,
      "loss": 1.2019,
      "step": 7810
    },
    {
      "epoch": 0.6286173633440515,
      "grad_norm": 1.2115646600723267,
      "learning_rate": 6.886198547215496e-05,
      "loss": 1.2811,
      "step": 7820
    },
    {
      "epoch": 0.6294212218649518,
      "grad_norm": 1.8098793029785156,
      "learning_rate": 6.882163034705408e-05,
      "loss": 1.125,
      "step": 7830
    },
    {
      "epoch": 0.6302250803858521,
      "grad_norm": 1.6568626165390015,
      "learning_rate": 6.87812752219532e-05,
      "loss": 1.1808,
      "step": 7840
    },
    {
      "epoch": 0.6310289389067524,
      "grad_norm": 1.643615961074829,
      "learning_rate": 6.87409200968523e-05,
      "loss": 1.209,
      "step": 7850
    },
    {
      "epoch": 0.6318327974276527,
      "grad_norm": 1.4601590633392334,
      "learning_rate": 6.870056497175142e-05,
      "loss": 1.1949,
      "step": 7860
    },
    {
      "epoch": 0.632636655948553,
      "grad_norm": 1.7648741006851196,
      "learning_rate": 6.866020984665052e-05,
      "loss": 1.0492,
      "step": 7870
    },
    {
      "epoch": 0.6334405144694534,
      "grad_norm": 1.511143684387207,
      "learning_rate": 6.861985472154964e-05,
      "loss": 1.3411,
      "step": 7880
    },
    {
      "epoch": 0.6342443729903537,
      "grad_norm": 1.235887050628662,
      "learning_rate": 6.857949959644875e-05,
      "loss": 1.1432,
      "step": 7890
    },
    {
      "epoch": 0.635048231511254,
      "grad_norm": 1.6829694509506226,
      "learning_rate": 6.853914447134786e-05,
      "loss": 1.1803,
      "step": 7900
    },
    {
      "epoch": 0.6358520900321544,
      "grad_norm": 1.0991063117980957,
      "learning_rate": 6.849878934624697e-05,
      "loss": 1.2734,
      "step": 7910
    },
    {
      "epoch": 0.6366559485530546,
      "grad_norm": 2.292754888534546,
      "learning_rate": 6.845843422114609e-05,
      "loss": 1.152,
      "step": 7920
    },
    {
      "epoch": 0.637459807073955,
      "grad_norm": 1.5538569688796997,
      "learning_rate": 6.841807909604519e-05,
      "loss": 1.1571,
      "step": 7930
    },
    {
      "epoch": 0.6382636655948553,
      "grad_norm": 1.1748164892196655,
      "learning_rate": 6.837772397094432e-05,
      "loss": 1.1991,
      "step": 7940
    },
    {
      "epoch": 0.6390675241157556,
      "grad_norm": 1.2783567905426025,
      "learning_rate": 6.833736884584343e-05,
      "loss": 1.2955,
      "step": 7950
    },
    {
      "epoch": 0.639871382636656,
      "grad_norm": 1.7576483488082886,
      "learning_rate": 6.829701372074255e-05,
      "loss": 1.2463,
      "step": 7960
    },
    {
      "epoch": 0.6406752411575563,
      "grad_norm": 1.5240103006362915,
      "learning_rate": 6.825665859564165e-05,
      "loss": 1.1883,
      "step": 7970
    },
    {
      "epoch": 0.6414790996784566,
      "grad_norm": 1.1293576955795288,
      "learning_rate": 6.821630347054077e-05,
      "loss": 1.1206,
      "step": 7980
    },
    {
      "epoch": 0.6422829581993569,
      "grad_norm": 1.4663335084915161,
      "learning_rate": 6.817594834543987e-05,
      "loss": 1.2008,
      "step": 7990
    },
    {
      "epoch": 0.6430868167202572,
      "grad_norm": 2.1061480045318604,
      "learning_rate": 6.813559322033899e-05,
      "loss": 1.2704,
      "step": 8000
    },
    {
      "epoch": 0.6430868167202572,
      "eval_yahma/alpaca-cleaned_loss": 1.225741982460022,
      "eval_yahma/alpaca-cleaned_runtime": 115.6052,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.3,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.163,
      "step": 8000
    },
    {
      "epoch": 0.6438906752411575,
      "grad_norm": 1.7242372035980225,
      "learning_rate": 6.80952380952381e-05,
      "loss": 1.1511,
      "step": 8010
    },
    {
      "epoch": 0.6446945337620579,
      "grad_norm": 2.951997995376587,
      "learning_rate": 6.805488297013721e-05,
      "loss": 1.2597,
      "step": 8020
    },
    {
      "epoch": 0.6454983922829582,
      "grad_norm": 1.1992913484573364,
      "learning_rate": 6.801452784503632e-05,
      "loss": 1.116,
      "step": 8030
    },
    {
      "epoch": 0.6463022508038585,
      "grad_norm": 1.4141520261764526,
      "learning_rate": 6.797417271993544e-05,
      "loss": 1.2101,
      "step": 8040
    },
    {
      "epoch": 0.6471061093247589,
      "grad_norm": 1.491201400756836,
      "learning_rate": 6.793381759483454e-05,
      "loss": 1.2755,
      "step": 8050
    },
    {
      "epoch": 0.6479099678456591,
      "grad_norm": 1.392905592918396,
      "learning_rate": 6.789346246973366e-05,
      "loss": 1.13,
      "step": 8060
    },
    {
      "epoch": 0.6487138263665595,
      "grad_norm": 2.571122169494629,
      "learning_rate": 6.785310734463278e-05,
      "loss": 1.2448,
      "step": 8070
    },
    {
      "epoch": 0.6495176848874598,
      "grad_norm": 1.7001698017120361,
      "learning_rate": 6.781275221953188e-05,
      "loss": 1.1545,
      "step": 8080
    },
    {
      "epoch": 0.6503215434083601,
      "grad_norm": 1.1046559810638428,
      "learning_rate": 6.7772397094431e-05,
      "loss": 1.221,
      "step": 8090
    },
    {
      "epoch": 0.6511254019292605,
      "grad_norm": 1.3756605386734009,
      "learning_rate": 6.77320419693301e-05,
      "loss": 1.2753,
      "step": 8100
    },
    {
      "epoch": 0.6519292604501608,
      "grad_norm": 1.8165565729141235,
      "learning_rate": 6.769168684422922e-05,
      "loss": 1.2406,
      "step": 8110
    },
    {
      "epoch": 0.6527331189710611,
      "grad_norm": 1.0720582008361816,
      "learning_rate": 6.765133171912833e-05,
      "loss": 1.2176,
      "step": 8120
    },
    {
      "epoch": 0.6535369774919614,
      "grad_norm": 2.1698720455169678,
      "learning_rate": 6.761097659402745e-05,
      "loss": 1.225,
      "step": 8130
    },
    {
      "epoch": 0.6543408360128617,
      "grad_norm": 1.36322820186615,
      "learning_rate": 6.757062146892655e-05,
      "loss": 1.2478,
      "step": 8140
    },
    {
      "epoch": 0.655144694533762,
      "grad_norm": 2.453728199005127,
      "learning_rate": 6.753026634382567e-05,
      "loss": 1.1111,
      "step": 8150
    },
    {
      "epoch": 0.6559485530546624,
      "grad_norm": 1.62954843044281,
      "learning_rate": 6.748991121872477e-05,
      "loss": 1.1836,
      "step": 8160
    },
    {
      "epoch": 0.6567524115755627,
      "grad_norm": 1.3338814973831177,
      "learning_rate": 6.744955609362389e-05,
      "loss": 1.108,
      "step": 8170
    },
    {
      "epoch": 0.657556270096463,
      "grad_norm": 2.812258243560791,
      "learning_rate": 6.7409200968523e-05,
      "loss": 1.2557,
      "step": 8180
    },
    {
      "epoch": 0.6583601286173634,
      "grad_norm": 1.158793330192566,
      "learning_rate": 6.736884584342211e-05,
      "loss": 1.1807,
      "step": 8190
    },
    {
      "epoch": 0.6591639871382636,
      "grad_norm": 4.249461650848389,
      "learning_rate": 6.732849071832123e-05,
      "loss": 1.2792,
      "step": 8200
    },
    {
      "epoch": 0.659967845659164,
      "grad_norm": 1.1763710975646973,
      "learning_rate": 6.728813559322035e-05,
      "loss": 1.2391,
      "step": 8210
    },
    {
      "epoch": 0.6607717041800643,
      "grad_norm": 0.897494912147522,
      "learning_rate": 6.724778046811946e-05,
      "loss": 1.2074,
      "step": 8220
    },
    {
      "epoch": 0.6615755627009646,
      "grad_norm": 1.7237837314605713,
      "learning_rate": 6.720742534301857e-05,
      "loss": 1.3164,
      "step": 8230
    },
    {
      "epoch": 0.662379421221865,
      "grad_norm": 1.3664112091064453,
      "learning_rate": 6.716707021791768e-05,
      "loss": 1.2813,
      "step": 8240
    },
    {
      "epoch": 0.6631832797427653,
      "grad_norm": 1.3216561079025269,
      "learning_rate": 6.71267150928168e-05,
      "loss": 1.1742,
      "step": 8250
    },
    {
      "epoch": 0.6639871382636656,
      "grad_norm": 1.8324904441833496,
      "learning_rate": 6.70863599677159e-05,
      "loss": 1.2871,
      "step": 8260
    },
    {
      "epoch": 0.6647909967845659,
      "grad_norm": 1.4093453884124756,
      "learning_rate": 6.704600484261502e-05,
      "loss": 1.1952,
      "step": 8270
    },
    {
      "epoch": 0.6655948553054662,
      "grad_norm": 1.3671215772628784,
      "learning_rate": 6.700564971751412e-05,
      "loss": 1.1578,
      "step": 8280
    },
    {
      "epoch": 0.6663987138263665,
      "grad_norm": 1.5380281209945679,
      "learning_rate": 6.696529459241324e-05,
      "loss": 1.1117,
      "step": 8290
    },
    {
      "epoch": 0.6672025723472669,
      "grad_norm": 1.2650505304336548,
      "learning_rate": 6.692493946731236e-05,
      "loss": 1.2053,
      "step": 8300
    },
    {
      "epoch": 0.6680064308681672,
      "grad_norm": 1.4757028818130493,
      "learning_rate": 6.688458434221146e-05,
      "loss": 1.2185,
      "step": 8310
    },
    {
      "epoch": 0.6688102893890675,
      "grad_norm": 1.5317448377609253,
      "learning_rate": 6.684422921711058e-05,
      "loss": 1.1689,
      "step": 8320
    },
    {
      "epoch": 0.6696141479099679,
      "grad_norm": 1.8137046098709106,
      "learning_rate": 6.680387409200969e-05,
      "loss": 1.3851,
      "step": 8330
    },
    {
      "epoch": 0.6704180064308681,
      "grad_norm": 1.6186026334762573,
      "learning_rate": 6.67635189669088e-05,
      "loss": 1.1795,
      "step": 8340
    },
    {
      "epoch": 0.6712218649517685,
      "grad_norm": 3.300424814224243,
      "learning_rate": 6.672316384180791e-05,
      "loss": 1.1575,
      "step": 8350
    },
    {
      "epoch": 0.6720257234726688,
      "grad_norm": 1.2810907363891602,
      "learning_rate": 6.668280871670703e-05,
      "loss": 1.1909,
      "step": 8360
    },
    {
      "epoch": 0.6728295819935691,
      "grad_norm": 1.7278164625167847,
      "learning_rate": 6.664245359160613e-05,
      "loss": 1.14,
      "step": 8370
    },
    {
      "epoch": 0.6736334405144695,
      "grad_norm": 1.4007691144943237,
      "learning_rate": 6.660209846650525e-05,
      "loss": 1.1765,
      "step": 8380
    },
    {
      "epoch": 0.6744372990353698,
      "grad_norm": 1.2572942972183228,
      "learning_rate": 6.656174334140436e-05,
      "loss": 1.2409,
      "step": 8390
    },
    {
      "epoch": 0.6752411575562701,
      "grad_norm": 1.5756325721740723,
      "learning_rate": 6.652138821630347e-05,
      "loss": 1.2796,
      "step": 8400
    },
    {
      "epoch": 0.6760450160771704,
      "grad_norm": 1.283668041229248,
      "learning_rate": 6.648103309120258e-05,
      "loss": 1.0367,
      "step": 8410
    },
    {
      "epoch": 0.6768488745980707,
      "grad_norm": 1.803141474723816,
      "learning_rate": 6.64406779661017e-05,
      "loss": 1.2924,
      "step": 8420
    },
    {
      "epoch": 0.677652733118971,
      "grad_norm": 1.47214937210083,
      "learning_rate": 6.64003228410008e-05,
      "loss": 1.1776,
      "step": 8430
    },
    {
      "epoch": 0.6784565916398714,
      "grad_norm": 1.3774545192718506,
      "learning_rate": 6.635996771589992e-05,
      "loss": 1.2473,
      "step": 8440
    },
    {
      "epoch": 0.6792604501607717,
      "grad_norm": 1.7005946636199951,
      "learning_rate": 6.631961259079902e-05,
      "loss": 1.1774,
      "step": 8450
    },
    {
      "epoch": 0.680064308681672,
      "grad_norm": 5.010222911834717,
      "learning_rate": 6.627925746569814e-05,
      "loss": 1.1893,
      "step": 8460
    },
    {
      "epoch": 0.6808681672025724,
      "grad_norm": 1.1495386362075806,
      "learning_rate": 6.623890234059726e-05,
      "loss": 1.1586,
      "step": 8470
    },
    {
      "epoch": 0.6816720257234726,
      "grad_norm": 1.4174809455871582,
      "learning_rate": 6.619854721549638e-05,
      "loss": 1.1392,
      "step": 8480
    },
    {
      "epoch": 0.682475884244373,
      "grad_norm": 1.0881890058517456,
      "learning_rate": 6.615819209039548e-05,
      "loss": 1.1908,
      "step": 8490
    },
    {
      "epoch": 0.6832797427652733,
      "grad_norm": 1.5683577060699463,
      "learning_rate": 6.61178369652946e-05,
      "loss": 1.1267,
      "step": 8500
    },
    {
      "epoch": 0.6840836012861736,
      "grad_norm": 1.0879807472229004,
      "learning_rate": 6.60774818401937e-05,
      "loss": 1.1413,
      "step": 8510
    },
    {
      "epoch": 0.684887459807074,
      "grad_norm": 1.0915292501449585,
      "learning_rate": 6.603712671509282e-05,
      "loss": 1.2611,
      "step": 8520
    },
    {
      "epoch": 0.6856913183279743,
      "grad_norm": 1.571441888809204,
      "learning_rate": 6.599677158999193e-05,
      "loss": 1.1953,
      "step": 8530
    },
    {
      "epoch": 0.6864951768488746,
      "grad_norm": 1.2628365755081177,
      "learning_rate": 6.595641646489105e-05,
      "loss": 1.2556,
      "step": 8540
    },
    {
      "epoch": 0.6872990353697749,
      "grad_norm": 1.3509798049926758,
      "learning_rate": 6.591606133979016e-05,
      "loss": 1.1023,
      "step": 8550
    },
    {
      "epoch": 0.6881028938906752,
      "grad_norm": 1.2220946550369263,
      "learning_rate": 6.587570621468927e-05,
      "loss": 1.216,
      "step": 8560
    },
    {
      "epoch": 0.6889067524115756,
      "grad_norm": 1.320622444152832,
      "learning_rate": 6.583535108958839e-05,
      "loss": 1.2987,
      "step": 8570
    },
    {
      "epoch": 0.6897106109324759,
      "grad_norm": 1.3505820035934448,
      "learning_rate": 6.579499596448749e-05,
      "loss": 1.3089,
      "step": 8580
    },
    {
      "epoch": 0.6905144694533762,
      "grad_norm": 1.2899729013442993,
      "learning_rate": 6.575464083938661e-05,
      "loss": 1.1303,
      "step": 8590
    },
    {
      "epoch": 0.6913183279742765,
      "grad_norm": 1.7675771713256836,
      "learning_rate": 6.571428571428571e-05,
      "loss": 1.2207,
      "step": 8600
    },
    {
      "epoch": 0.6921221864951769,
      "grad_norm": 1.1679964065551758,
      "learning_rate": 6.567393058918483e-05,
      "loss": 1.1867,
      "step": 8610
    },
    {
      "epoch": 0.6929260450160771,
      "grad_norm": 1.245856523513794,
      "learning_rate": 6.563357546408394e-05,
      "loss": 1.1644,
      "step": 8620
    },
    {
      "epoch": 0.6937299035369775,
      "grad_norm": 1.3648661375045776,
      "learning_rate": 6.559322033898306e-05,
      "loss": 1.2478,
      "step": 8630
    },
    {
      "epoch": 0.6945337620578779,
      "grad_norm": 1.5502279996871948,
      "learning_rate": 6.555286521388216e-05,
      "loss": 1.0907,
      "step": 8640
    },
    {
      "epoch": 0.6953376205787781,
      "grad_norm": 1.7564395666122437,
      "learning_rate": 6.551251008878128e-05,
      "loss": 1.1526,
      "step": 8650
    },
    {
      "epoch": 0.6961414790996785,
      "grad_norm": 1.3868829011917114,
      "learning_rate": 6.547215496368038e-05,
      "loss": 1.2426,
      "step": 8660
    },
    {
      "epoch": 0.6969453376205788,
      "grad_norm": 1.6506773233413696,
      "learning_rate": 6.54317998385795e-05,
      "loss": 1.0556,
      "step": 8670
    },
    {
      "epoch": 0.6977491961414791,
      "grad_norm": 1.3206250667572021,
      "learning_rate": 6.53914447134786e-05,
      "loss": 1.3248,
      "step": 8680
    },
    {
      "epoch": 0.6985530546623794,
      "grad_norm": 1.1856706142425537,
      "learning_rate": 6.535108958837772e-05,
      "loss": 1.3029,
      "step": 8690
    },
    {
      "epoch": 0.6993569131832797,
      "grad_norm": 1.089762806892395,
      "learning_rate": 6.531073446327683e-05,
      "loss": 1.1649,
      "step": 8700
    },
    {
      "epoch": 0.7001607717041801,
      "grad_norm": 1.6999385356903076,
      "learning_rate": 6.527037933817595e-05,
      "loss": 1.221,
      "step": 8710
    },
    {
      "epoch": 0.7009646302250804,
      "grad_norm": 1.2625404596328735,
      "learning_rate": 6.523002421307506e-05,
      "loss": 1.0953,
      "step": 8720
    },
    {
      "epoch": 0.7017684887459807,
      "grad_norm": 1.6326946020126343,
      "learning_rate": 6.518966908797418e-05,
      "loss": 1.3168,
      "step": 8730
    },
    {
      "epoch": 0.702572347266881,
      "grad_norm": 1.0178214311599731,
      "learning_rate": 6.514931396287329e-05,
      "loss": 1.1271,
      "step": 8740
    },
    {
      "epoch": 0.7033762057877814,
      "grad_norm": 1.3457207679748535,
      "learning_rate": 6.51089588377724e-05,
      "loss": 1.2128,
      "step": 8750
    },
    {
      "epoch": 0.7041800643086816,
      "grad_norm": 1.4977234601974487,
      "learning_rate": 6.506860371267151e-05,
      "loss": 1.241,
      "step": 8760
    },
    {
      "epoch": 0.704983922829582,
      "grad_norm": 1.3846150636672974,
      "learning_rate": 6.502824858757063e-05,
      "loss": 1.1745,
      "step": 8770
    },
    {
      "epoch": 0.7057877813504824,
      "grad_norm": 1.4363330602645874,
      "learning_rate": 6.498789346246975e-05,
      "loss": 1.1242,
      "step": 8780
    },
    {
      "epoch": 0.7065916398713826,
      "grad_norm": 1.6958163976669312,
      "learning_rate": 6.494753833736885e-05,
      "loss": 1.2168,
      "step": 8790
    },
    {
      "epoch": 0.707395498392283,
      "grad_norm": 0.9998372793197632,
      "learning_rate": 6.490718321226797e-05,
      "loss": 1.0476,
      "step": 8800
    },
    {
      "epoch": 0.7081993569131833,
      "grad_norm": 1.3063734769821167,
      "learning_rate": 6.486682808716707e-05,
      "loss": 1.231,
      "step": 8810
    },
    {
      "epoch": 0.7090032154340836,
      "grad_norm": 1.0710753202438354,
      "learning_rate": 6.482647296206619e-05,
      "loss": 1.1191,
      "step": 8820
    },
    {
      "epoch": 0.7098070739549839,
      "grad_norm": 1.6944465637207031,
      "learning_rate": 6.47861178369653e-05,
      "loss": 1.1878,
      "step": 8830
    },
    {
      "epoch": 0.7106109324758842,
      "grad_norm": 1.3219144344329834,
      "learning_rate": 6.474576271186441e-05,
      "loss": 1.2053,
      "step": 8840
    },
    {
      "epoch": 0.7114147909967846,
      "grad_norm": 1.5892850160598755,
      "learning_rate": 6.470540758676352e-05,
      "loss": 1.0587,
      "step": 8850
    },
    {
      "epoch": 0.7122186495176849,
      "grad_norm": 3.019225597381592,
      "learning_rate": 6.466505246166264e-05,
      "loss": 1.1815,
      "step": 8860
    },
    {
      "epoch": 0.7130225080385852,
      "grad_norm": 1.493353009223938,
      "learning_rate": 6.462469733656174e-05,
      "loss": 1.1207,
      "step": 8870
    },
    {
      "epoch": 0.7138263665594855,
      "grad_norm": 2.1397647857666016,
      "learning_rate": 6.458837772397096e-05,
      "loss": 1.1683,
      "step": 8880
    },
    {
      "epoch": 0.7146302250803859,
      "grad_norm": 1.1362327337265015,
      "learning_rate": 6.454802259887006e-05,
      "loss": 1.0961,
      "step": 8890
    },
    {
      "epoch": 0.7154340836012861,
      "grad_norm": 1.0656194686889648,
      "learning_rate": 6.450766747376918e-05,
      "loss": 1.1447,
      "step": 8900
    },
    {
      "epoch": 0.7162379421221865,
      "grad_norm": 1.3920767307281494,
      "learning_rate": 6.446731234866828e-05,
      "loss": 1.2971,
      "step": 8910
    },
    {
      "epoch": 0.7170418006430869,
      "grad_norm": 1.0478270053863525,
      "learning_rate": 6.44269572235674e-05,
      "loss": 1.23,
      "step": 8920
    },
    {
      "epoch": 0.7178456591639871,
      "grad_norm": 1.6214714050292969,
      "learning_rate": 6.43866020984665e-05,
      "loss": 1.2869,
      "step": 8930
    },
    {
      "epoch": 0.7186495176848875,
      "grad_norm": 1.8323612213134766,
      "learning_rate": 6.434624697336562e-05,
      "loss": 1.1301,
      "step": 8940
    },
    {
      "epoch": 0.7194533762057878,
      "grad_norm": 1.6926652193069458,
      "learning_rate": 6.430589184826473e-05,
      "loss": 1.3043,
      "step": 8950
    },
    {
      "epoch": 0.7202572347266881,
      "grad_norm": 1.4071663618087769,
      "learning_rate": 6.426553672316385e-05,
      "loss": 1.0648,
      "step": 8960
    },
    {
      "epoch": 0.7210610932475884,
      "grad_norm": 1.264920949935913,
      "learning_rate": 6.422518159806295e-05,
      "loss": 1.1961,
      "step": 8970
    },
    {
      "epoch": 0.7218649517684887,
      "grad_norm": 1.331862211227417,
      "learning_rate": 6.418482647296207e-05,
      "loss": 1.3281,
      "step": 8980
    },
    {
      "epoch": 0.7226688102893891,
      "grad_norm": 1.2815852165222168,
      "learning_rate": 6.414447134786117e-05,
      "loss": 1.2381,
      "step": 8990
    },
    {
      "epoch": 0.7234726688102894,
      "grad_norm": 1.4118419885635376,
      "learning_rate": 6.410411622276029e-05,
      "loss": 1.1161,
      "step": 9000
    },
    {
      "epoch": 0.7242765273311897,
      "grad_norm": 1.3302923440933228,
      "learning_rate": 6.40637610976594e-05,
      "loss": 1.1986,
      "step": 9010
    },
    {
      "epoch": 0.72508038585209,
      "grad_norm": 1.8220189809799194,
      "learning_rate": 6.402340597255852e-05,
      "loss": 1.1209,
      "step": 9020
    },
    {
      "epoch": 0.7258842443729904,
      "grad_norm": 1.5864709615707397,
      "learning_rate": 6.398305084745762e-05,
      "loss": 1.2848,
      "step": 9030
    },
    {
      "epoch": 0.7266881028938906,
      "grad_norm": 2.4576737880706787,
      "learning_rate": 6.394269572235674e-05,
      "loss": 1.2107,
      "step": 9040
    },
    {
      "epoch": 0.727491961414791,
      "grad_norm": 1.4367574453353882,
      "learning_rate": 6.390234059725586e-05,
      "loss": 1.3456,
      "step": 9050
    },
    {
      "epoch": 0.7282958199356914,
      "grad_norm": 1.2669857740402222,
      "learning_rate": 6.386198547215496e-05,
      "loss": 1.099,
      "step": 9060
    },
    {
      "epoch": 0.7290996784565916,
      "grad_norm": 1.5575499534606934,
      "learning_rate": 6.382163034705408e-05,
      "loss": 1.1937,
      "step": 9070
    },
    {
      "epoch": 0.729903536977492,
      "grad_norm": 1.5874534845352173,
      "learning_rate": 6.37812752219532e-05,
      "loss": 1.2806,
      "step": 9080
    },
    {
      "epoch": 0.7307073954983923,
      "grad_norm": 1.2446188926696777,
      "learning_rate": 6.37409200968523e-05,
      "loss": 1.1843,
      "step": 9090
    },
    {
      "epoch": 0.7315112540192926,
      "grad_norm": 1.694628119468689,
      "learning_rate": 6.370056497175142e-05,
      "loss": 1.2325,
      "step": 9100
    },
    {
      "epoch": 0.7323151125401929,
      "grad_norm": 1.0469274520874023,
      "learning_rate": 6.366020984665054e-05,
      "loss": 1.2086,
      "step": 9110
    },
    {
      "epoch": 0.7331189710610932,
      "grad_norm": 1.4143184423446655,
      "learning_rate": 6.361985472154964e-05,
      "loss": 1.1298,
      "step": 9120
    },
    {
      "epoch": 0.7339228295819936,
      "grad_norm": 1.917021632194519,
      "learning_rate": 6.357949959644876e-05,
      "loss": 1.1863,
      "step": 9130
    },
    {
      "epoch": 0.7347266881028939,
      "grad_norm": 1.2662550210952759,
      "learning_rate": 6.353914447134787e-05,
      "loss": 1.0804,
      "step": 9140
    },
    {
      "epoch": 0.7355305466237942,
      "grad_norm": 1.4026659727096558,
      "learning_rate": 6.349878934624698e-05,
      "loss": 1.1445,
      "step": 9150
    },
    {
      "epoch": 0.7363344051446945,
      "grad_norm": 1.2977662086486816,
      "learning_rate": 6.345843422114609e-05,
      "loss": 1.1224,
      "step": 9160
    },
    {
      "epoch": 0.7371382636655949,
      "grad_norm": 1.3581264019012451,
      "learning_rate": 6.34180790960452e-05,
      "loss": 1.2324,
      "step": 9170
    },
    {
      "epoch": 0.7379421221864951,
      "grad_norm": 1.3826358318328857,
      "learning_rate": 6.337772397094431e-05,
      "loss": 1.2467,
      "step": 9180
    },
    {
      "epoch": 0.7387459807073955,
      "grad_norm": 1.1410869359970093,
      "learning_rate": 6.333736884584343e-05,
      "loss": 1.1995,
      "step": 9190
    },
    {
      "epoch": 0.7395498392282959,
      "grad_norm": 1.2657575607299805,
      "learning_rate": 6.329701372074253e-05,
      "loss": 1.1736,
      "step": 9200
    },
    {
      "epoch": 0.7403536977491961,
      "grad_norm": 1.88206148147583,
      "learning_rate": 6.325665859564165e-05,
      "loss": 1.1617,
      "step": 9210
    },
    {
      "epoch": 0.7411575562700965,
      "grad_norm": 2.55361270904541,
      "learning_rate": 6.321630347054076e-05,
      "loss": 1.1128,
      "step": 9220
    },
    {
      "epoch": 0.7419614147909968,
      "grad_norm": 1.296675205230713,
      "learning_rate": 6.317594834543987e-05,
      "loss": 1.1873,
      "step": 9230
    },
    {
      "epoch": 0.7427652733118971,
      "grad_norm": 1.9324688911437988,
      "learning_rate": 6.313559322033898e-05,
      "loss": 1.2678,
      "step": 9240
    },
    {
      "epoch": 0.7435691318327974,
      "grad_norm": 1.5639374256134033,
      "learning_rate": 6.30952380952381e-05,
      "loss": 1.2444,
      "step": 9250
    },
    {
      "epoch": 0.7443729903536977,
      "grad_norm": 1.1058989763259888,
      "learning_rate": 6.30548829701372e-05,
      "loss": 1.1626,
      "step": 9260
    },
    {
      "epoch": 0.7451768488745981,
      "grad_norm": 1.621466040611267,
      "learning_rate": 6.301452784503632e-05,
      "loss": 1.1706,
      "step": 9270
    },
    {
      "epoch": 0.7459807073954984,
      "grad_norm": 1.8088675737380981,
      "learning_rate": 6.297417271993544e-05,
      "loss": 1.111,
      "step": 9280
    },
    {
      "epoch": 0.7467845659163987,
      "grad_norm": 1.7105693817138672,
      "learning_rate": 6.293381759483454e-05,
      "loss": 1.1828,
      "step": 9290
    },
    {
      "epoch": 0.747588424437299,
      "grad_norm": 1.2621886730194092,
      "learning_rate": 6.289346246973366e-05,
      "loss": 1.3506,
      "step": 9300
    },
    {
      "epoch": 0.7483922829581994,
      "grad_norm": 1.4790949821472168,
      "learning_rate": 6.285310734463276e-05,
      "loss": 1.2797,
      "step": 9310
    },
    {
      "epoch": 0.7491961414790996,
      "grad_norm": 1.2256627082824707,
      "learning_rate": 6.281275221953188e-05,
      "loss": 1.2512,
      "step": 9320
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9516915678977966,
      "learning_rate": 6.277239709443099e-05,
      "loss": 1.2157,
      "step": 9330
    },
    {
      "epoch": 0.7508038585209004,
      "grad_norm": 1.2063665390014648,
      "learning_rate": 6.273204196933012e-05,
      "loss": 1.1853,
      "step": 9340
    },
    {
      "epoch": 0.7516077170418006,
      "grad_norm": 1.9841398000717163,
      "learning_rate": 6.269168684422922e-05,
      "loss": 1.1145,
      "step": 9350
    },
    {
      "epoch": 0.752411575562701,
      "grad_norm": 1.2014284133911133,
      "learning_rate": 6.265133171912834e-05,
      "loss": 1.1489,
      "step": 9360
    },
    {
      "epoch": 0.7532154340836013,
      "grad_norm": 1.0883020162582397,
      "learning_rate": 6.261097659402745e-05,
      "loss": 1.3133,
      "step": 9370
    },
    {
      "epoch": 0.7540192926045016,
      "grad_norm": 1.5855486392974854,
      "learning_rate": 6.257062146892656e-05,
      "loss": 1.2134,
      "step": 9380
    },
    {
      "epoch": 0.7548231511254019,
      "grad_norm": 1.7968937158584595,
      "learning_rate": 6.253026634382567e-05,
      "loss": 1.1743,
      "step": 9390
    },
    {
      "epoch": 0.7556270096463023,
      "grad_norm": 1.5748929977416992,
      "learning_rate": 6.248991121872479e-05,
      "loss": 1.1228,
      "step": 9400
    },
    {
      "epoch": 0.7564308681672026,
      "grad_norm": 1.267173409461975,
      "learning_rate": 6.244955609362389e-05,
      "loss": 1.0907,
      "step": 9410
    },
    {
      "epoch": 0.7572347266881029,
      "grad_norm": 1.3452446460723877,
      "learning_rate": 6.240920096852301e-05,
      "loss": 1.2708,
      "step": 9420
    },
    {
      "epoch": 0.7580385852090032,
      "grad_norm": 1.4234520196914673,
      "learning_rate": 6.236884584342211e-05,
      "loss": 1.2494,
      "step": 9430
    },
    {
      "epoch": 0.7588424437299035,
      "grad_norm": 1.4506953954696655,
      "learning_rate": 6.232849071832123e-05,
      "loss": 1.1533,
      "step": 9440
    },
    {
      "epoch": 0.7596463022508039,
      "grad_norm": 2.0665547847747803,
      "learning_rate": 6.228813559322034e-05,
      "loss": 1.2174,
      "step": 9450
    },
    {
      "epoch": 0.7604501607717041,
      "grad_norm": 1.514133095741272,
      "learning_rate": 6.224778046811946e-05,
      "loss": 1.171,
      "step": 9460
    },
    {
      "epoch": 0.7612540192926045,
      "grad_norm": 1.0817458629608154,
      "learning_rate": 6.220742534301856e-05,
      "loss": 1.2038,
      "step": 9470
    },
    {
      "epoch": 0.7620578778135049,
      "grad_norm": 1.4435175657272339,
      "learning_rate": 6.216707021791768e-05,
      "loss": 1.1971,
      "step": 9480
    },
    {
      "epoch": 0.7628617363344051,
      "grad_norm": 1.4563673734664917,
      "learning_rate": 6.212671509281678e-05,
      "loss": 1.1997,
      "step": 9490
    },
    {
      "epoch": 0.7636655948553055,
      "grad_norm": 1.271362543106079,
      "learning_rate": 6.20863599677159e-05,
      "loss": 1.3304,
      "step": 9500
    },
    {
      "epoch": 0.7644694533762058,
      "grad_norm": 1.5026984214782715,
      "learning_rate": 6.204600484261502e-05,
      "loss": 1.2449,
      "step": 9510
    },
    {
      "epoch": 0.7652733118971061,
      "grad_norm": 1.5852859020233154,
      "learning_rate": 6.200564971751412e-05,
      "loss": 1.191,
      "step": 9520
    },
    {
      "epoch": 0.7660771704180064,
      "grad_norm": 1.6117398738861084,
      "learning_rate": 6.196529459241324e-05,
      "loss": 1.1819,
      "step": 9530
    },
    {
      "epoch": 0.7668810289389068,
      "grad_norm": 1.41196608543396,
      "learning_rate": 6.192493946731235e-05,
      "loss": 1.3235,
      "step": 9540
    },
    {
      "epoch": 0.7676848874598071,
      "grad_norm": 1.1791915893554688,
      "learning_rate": 6.188458434221146e-05,
      "loss": 1.1063,
      "step": 9550
    },
    {
      "epoch": 0.7684887459807074,
      "grad_norm": 1.18216073513031,
      "learning_rate": 6.184422921711057e-05,
      "loss": 1.1579,
      "step": 9560
    },
    {
      "epoch": 0.7692926045016077,
      "grad_norm": 1.2205034494400024,
      "learning_rate": 6.180387409200969e-05,
      "loss": 1.2236,
      "step": 9570
    },
    {
      "epoch": 0.770096463022508,
      "grad_norm": 1.3474924564361572,
      "learning_rate": 6.176351896690879e-05,
      "loss": 1.131,
      "step": 9580
    },
    {
      "epoch": 0.7709003215434084,
      "grad_norm": 1.1282883882522583,
      "learning_rate": 6.172316384180791e-05,
      "loss": 1.1688,
      "step": 9590
    },
    {
      "epoch": 0.7717041800643086,
      "grad_norm": 1.1234114170074463,
      "learning_rate": 6.168280871670703e-05,
      "loss": 1.2443,
      "step": 9600
    },
    {
      "epoch": 0.772508038585209,
      "grad_norm": 2.7613494396209717,
      "learning_rate": 6.164245359160615e-05,
      "loss": 1.1255,
      "step": 9610
    },
    {
      "epoch": 0.7733118971061094,
      "grad_norm": 1.71947181224823,
      "learning_rate": 6.160209846650525e-05,
      "loss": 1.2513,
      "step": 9620
    },
    {
      "epoch": 0.7741157556270096,
      "grad_norm": 1.1563067436218262,
      "learning_rate": 6.156174334140437e-05,
      "loss": 1.005,
      "step": 9630
    },
    {
      "epoch": 0.77491961414791,
      "grad_norm": 1.963090419769287,
      "learning_rate": 6.152138821630347e-05,
      "loss": 1.2146,
      "step": 9640
    },
    {
      "epoch": 0.7757234726688103,
      "grad_norm": 1.3083224296569824,
      "learning_rate": 6.148103309120259e-05,
      "loss": 1.1852,
      "step": 9650
    },
    {
      "epoch": 0.7765273311897106,
      "grad_norm": 1.2836787700653076,
      "learning_rate": 6.14406779661017e-05,
      "loss": 1.1658,
      "step": 9660
    },
    {
      "epoch": 0.7773311897106109,
      "grad_norm": 1.1122138500213623,
      "learning_rate": 6.140032284100081e-05,
      "loss": 1.1442,
      "step": 9670
    },
    {
      "epoch": 0.7781350482315113,
      "grad_norm": 2.9130606651306152,
      "learning_rate": 6.135996771589992e-05,
      "loss": 1.3943,
      "step": 9680
    },
    {
      "epoch": 0.7789389067524116,
      "grad_norm": 1.2485105991363525,
      "learning_rate": 6.131961259079904e-05,
      "loss": 1.2203,
      "step": 9690
    },
    {
      "epoch": 0.7797427652733119,
      "grad_norm": 1.7842824459075928,
      "learning_rate": 6.127925746569814e-05,
      "loss": 1.2153,
      "step": 9700
    },
    {
      "epoch": 0.7805466237942122,
      "grad_norm": 1.3240492343902588,
      "learning_rate": 6.123890234059726e-05,
      "loss": 1.2034,
      "step": 9710
    },
    {
      "epoch": 0.7813504823151125,
      "grad_norm": 1.3932291269302368,
      "learning_rate": 6.119854721549636e-05,
      "loss": 1.1873,
      "step": 9720
    },
    {
      "epoch": 0.7821543408360129,
      "grad_norm": 1.2369239330291748,
      "learning_rate": 6.115819209039548e-05,
      "loss": 1.1213,
      "step": 9730
    },
    {
      "epoch": 0.7829581993569131,
      "grad_norm": 1.24186110496521,
      "learning_rate": 6.11178369652946e-05,
      "loss": 1.1573,
      "step": 9740
    },
    {
      "epoch": 0.7837620578778135,
      "grad_norm": 1.5317646265029907,
      "learning_rate": 6.10774818401937e-05,
      "loss": 1.1962,
      "step": 9750
    },
    {
      "epoch": 0.7845659163987139,
      "grad_norm": 1.554369568824768,
      "learning_rate": 6.103712671509282e-05,
      "loss": 1.304,
      "step": 9760
    },
    {
      "epoch": 0.7853697749196141,
      "grad_norm": 2.1366379261016846,
      "learning_rate": 6.099677158999193e-05,
      "loss": 1.1438,
      "step": 9770
    },
    {
      "epoch": 0.7861736334405145,
      "grad_norm": 1.6524451971054077,
      "learning_rate": 6.095641646489104e-05,
      "loss": 1.12,
      "step": 9780
    },
    {
      "epoch": 0.7869774919614148,
      "grad_norm": 1.5162711143493652,
      "learning_rate": 6.091606133979015e-05,
      "loss": 1.1718,
      "step": 9790
    },
    {
      "epoch": 0.7877813504823151,
      "grad_norm": 1.8277530670166016,
      "learning_rate": 6.087570621468926e-05,
      "loss": 1.1945,
      "step": 9800
    },
    {
      "epoch": 0.7885852090032154,
      "grad_norm": 1.7623543739318848,
      "learning_rate": 6.0835351089588374e-05,
      "loss": 1.1627,
      "step": 9810
    },
    {
      "epoch": 0.7893890675241158,
      "grad_norm": 1.6323825120925903,
      "learning_rate": 6.079499596448749e-05,
      "loss": 1.3111,
      "step": 9820
    },
    {
      "epoch": 0.7901929260450161,
      "grad_norm": 1.194577932357788,
      "learning_rate": 6.0754640839386603e-05,
      "loss": 1.2049,
      "step": 9830
    },
    {
      "epoch": 0.7909967845659164,
      "grad_norm": 1.5735859870910645,
      "learning_rate": 6.0714285714285715e-05,
      "loss": 1.3183,
      "step": 9840
    },
    {
      "epoch": 0.7918006430868167,
      "grad_norm": 1.4071894884109497,
      "learning_rate": 6.0673930589184826e-05,
      "loss": 1.2357,
      "step": 9850
    },
    {
      "epoch": 0.792604501607717,
      "grad_norm": 1.5908355712890625,
      "learning_rate": 6.0633575464083944e-05,
      "loss": 1.1461,
      "step": 9860
    },
    {
      "epoch": 0.7934083601286174,
      "grad_norm": 1.2151633501052856,
      "learning_rate": 6.0593220338983056e-05,
      "loss": 1.1617,
      "step": 9870
    },
    {
      "epoch": 0.7942122186495176,
      "grad_norm": 1.1097683906555176,
      "learning_rate": 6.0552865213882174e-05,
      "loss": 1.2553,
      "step": 9880
    },
    {
      "epoch": 0.795016077170418,
      "grad_norm": 1.1444097757339478,
      "learning_rate": 6.0512510088781285e-05,
      "loss": 1.2079,
      "step": 9890
    },
    {
      "epoch": 0.7958199356913184,
      "grad_norm": 2.0320544242858887,
      "learning_rate": 6.0472154963680397e-05,
      "loss": 1.2089,
      "step": 9900
    },
    {
      "epoch": 0.7966237942122186,
      "grad_norm": 1.5209218263626099,
      "learning_rate": 6.043179983857951e-05,
      "loss": 1.1291,
      "step": 9910
    },
    {
      "epoch": 0.797427652733119,
      "grad_norm": 1.0873534679412842,
      "learning_rate": 6.039144471347862e-05,
      "loss": 1.221,
      "step": 9920
    },
    {
      "epoch": 0.7982315112540193,
      "grad_norm": 1.0633383989334106,
      "learning_rate": 6.035108958837773e-05,
      "loss": 1.1958,
      "step": 9930
    },
    {
      "epoch": 0.7990353697749196,
      "grad_norm": 2.4503116607666016,
      "learning_rate": 6.031073446327684e-05,
      "loss": 1.1433,
      "step": 9940
    },
    {
      "epoch": 0.7998392282958199,
      "grad_norm": 1.9616880416870117,
      "learning_rate": 6.027037933817595e-05,
      "loss": 1.2961,
      "step": 9950
    },
    {
      "epoch": 0.8006430868167203,
      "grad_norm": 1.9644263982772827,
      "learning_rate": 6.0230024213075065e-05,
      "loss": 1.0541,
      "step": 9960
    },
    {
      "epoch": 0.8014469453376206,
      "grad_norm": 1.420846700668335,
      "learning_rate": 6.0189669087974176e-05,
      "loss": 1.3086,
      "step": 9970
    },
    {
      "epoch": 0.8022508038585209,
      "grad_norm": 2.0573222637176514,
      "learning_rate": 6.014931396287329e-05,
      "loss": 1.0556,
      "step": 9980
    },
    {
      "epoch": 0.8030546623794212,
      "grad_norm": 1.1859737634658813,
      "learning_rate": 6.01089588377724e-05,
      "loss": 1.1973,
      "step": 9990
    },
    {
      "epoch": 0.8038585209003215,
      "grad_norm": 1.1852400302886963,
      "learning_rate": 6.006860371267151e-05,
      "loss": 1.1116,
      "step": 10000
    },
    {
      "epoch": 0.8038585209003215,
      "eval_yahma/alpaca-cleaned_loss": 1.2155007123947144,
      "eval_yahma/alpaca-cleaned_runtime": 115.7956,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.272,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.159,
      "step": 10000
    },
    {
      "epoch": 0.8046623794212219,
      "grad_norm": 1.667161464691162,
      "learning_rate": 6.002824858757062e-05,
      "loss": 1.2352,
      "step": 10010
    },
    {
      "epoch": 0.8054662379421221,
      "grad_norm": 2.6935198307037354,
      "learning_rate": 5.998789346246973e-05,
      "loss": 1.1627,
      "step": 10020
    },
    {
      "epoch": 0.8062700964630225,
      "grad_norm": 1.9311403036117554,
      "learning_rate": 5.9947538337368844e-05,
      "loss": 1.0085,
      "step": 10030
    },
    {
      "epoch": 0.8070739549839229,
      "grad_norm": 2.0662097930908203,
      "learning_rate": 5.9907183212267956e-05,
      "loss": 1.1416,
      "step": 10040
    },
    {
      "epoch": 0.8078778135048231,
      "grad_norm": 2.6683616638183594,
      "learning_rate": 5.9866828087167074e-05,
      "loss": 1.0442,
      "step": 10050
    },
    {
      "epoch": 0.8086816720257235,
      "grad_norm": 1.4304804801940918,
      "learning_rate": 5.9826472962066185e-05,
      "loss": 1.3182,
      "step": 10060
    },
    {
      "epoch": 0.8094855305466238,
      "grad_norm": 2.3583433628082275,
      "learning_rate": 5.9786117836965296e-05,
      "loss": 1.2935,
      "step": 10070
    },
    {
      "epoch": 0.8102893890675241,
      "grad_norm": 1.488024115562439,
      "learning_rate": 5.974576271186441e-05,
      "loss": 1.081,
      "step": 10080
    },
    {
      "epoch": 0.8110932475884244,
      "grad_norm": 1.401058316230774,
      "learning_rate": 5.970540758676352e-05,
      "loss": 1.1965,
      "step": 10090
    },
    {
      "epoch": 0.8118971061093248,
      "grad_norm": 1.3213167190551758,
      "learning_rate": 5.966505246166263e-05,
      "loss": 1.1513,
      "step": 10100
    },
    {
      "epoch": 0.8127009646302251,
      "grad_norm": 1.2974404096603394,
      "learning_rate": 5.962469733656174e-05,
      "loss": 1.2244,
      "step": 10110
    },
    {
      "epoch": 0.8135048231511254,
      "grad_norm": 1.1411107778549194,
      "learning_rate": 5.958434221146085e-05,
      "loss": 1.1874,
      "step": 10120
    },
    {
      "epoch": 0.8143086816720257,
      "grad_norm": 1.4510279893875122,
      "learning_rate": 5.954398708635998e-05,
      "loss": 1.1776,
      "step": 10130
    },
    {
      "epoch": 0.815112540192926,
      "grad_norm": 1.4337852001190186,
      "learning_rate": 5.950363196125909e-05,
      "loss": 1.3446,
      "step": 10140
    },
    {
      "epoch": 0.8159163987138264,
      "grad_norm": 1.5213372707366943,
      "learning_rate": 5.94632768361582e-05,
      "loss": 1.1798,
      "step": 10150
    },
    {
      "epoch": 0.8167202572347267,
      "grad_norm": 1.4217019081115723,
      "learning_rate": 5.942292171105731e-05,
      "loss": 1.0763,
      "step": 10160
    },
    {
      "epoch": 0.817524115755627,
      "grad_norm": 1.2413939237594604,
      "learning_rate": 5.9382566585956424e-05,
      "loss": 1.0959,
      "step": 10170
    },
    {
      "epoch": 0.8183279742765274,
      "grad_norm": 1.3456542491912842,
      "learning_rate": 5.9342211460855535e-05,
      "loss": 1.1784,
      "step": 10180
    },
    {
      "epoch": 0.8191318327974276,
      "grad_norm": 1.8825595378875732,
      "learning_rate": 5.9301856335754646e-05,
      "loss": 1.1552,
      "step": 10190
    },
    {
      "epoch": 0.819935691318328,
      "grad_norm": 1.545015811920166,
      "learning_rate": 5.926150121065376e-05,
      "loss": 1.1653,
      "step": 10200
    },
    {
      "epoch": 0.8207395498392283,
      "grad_norm": 2.0575296878814697,
      "learning_rate": 5.922114608555287e-05,
      "loss": 1.0649,
      "step": 10210
    },
    {
      "epoch": 0.8215434083601286,
      "grad_norm": 1.4069037437438965,
      "learning_rate": 5.918079096045198e-05,
      "loss": 1.159,
      "step": 10220
    },
    {
      "epoch": 0.822347266881029,
      "grad_norm": 1.1657689809799194,
      "learning_rate": 5.914043583535109e-05,
      "loss": 1.1816,
      "step": 10230
    },
    {
      "epoch": 0.8231511254019293,
      "grad_norm": 1.254135251045227,
      "learning_rate": 5.91000807102502e-05,
      "loss": 1.2129,
      "step": 10240
    },
    {
      "epoch": 0.8239549839228296,
      "grad_norm": 2.080061912536621,
      "learning_rate": 5.9059725585149315e-05,
      "loss": 1.2498,
      "step": 10250
    },
    {
      "epoch": 0.8247588424437299,
      "grad_norm": 1.7076865434646606,
      "learning_rate": 5.9019370460048426e-05,
      "loss": 1.0514,
      "step": 10260
    },
    {
      "epoch": 0.8255627009646302,
      "grad_norm": 1.2622432708740234,
      "learning_rate": 5.897901533494754e-05,
      "loss": 1.2691,
      "step": 10270
    },
    {
      "epoch": 0.8263665594855305,
      "grad_norm": 1.3416787385940552,
      "learning_rate": 5.8938660209846655e-05,
      "loss": 1.1225,
      "step": 10280
    },
    {
      "epoch": 0.8271704180064309,
      "grad_norm": 1.5244756937026978,
      "learning_rate": 5.889830508474577e-05,
      "loss": 1.2898,
      "step": 10290
    },
    {
      "epoch": 0.8279742765273312,
      "grad_norm": 1.0199939012527466,
      "learning_rate": 5.885794995964488e-05,
      "loss": 1.1597,
      "step": 10300
    },
    {
      "epoch": 0.8287781350482315,
      "grad_norm": 1.2063865661621094,
      "learning_rate": 5.881759483454399e-05,
      "loss": 1.3285,
      "step": 10310
    },
    {
      "epoch": 0.8295819935691319,
      "grad_norm": 1.3245232105255127,
      "learning_rate": 5.87772397094431e-05,
      "loss": 1.0705,
      "step": 10320
    },
    {
      "epoch": 0.8303858520900321,
      "grad_norm": 1.2130075693130493,
      "learning_rate": 5.873688458434221e-05,
      "loss": 1.2279,
      "step": 10330
    },
    {
      "epoch": 0.8311897106109325,
      "grad_norm": 1.4783278703689575,
      "learning_rate": 5.8696529459241324e-05,
      "loss": 1.1647,
      "step": 10340
    },
    {
      "epoch": 0.8319935691318328,
      "grad_norm": 1.352412462234497,
      "learning_rate": 5.8656174334140435e-05,
      "loss": 1.194,
      "step": 10350
    },
    {
      "epoch": 0.8327974276527331,
      "grad_norm": 1.2860634326934814,
      "learning_rate": 5.8615819209039546e-05,
      "loss": 1.2023,
      "step": 10360
    },
    {
      "epoch": 0.8336012861736335,
      "grad_norm": 1.5555285215377808,
      "learning_rate": 5.857546408393866e-05,
      "loss": 1.1868,
      "step": 10370
    },
    {
      "epoch": 0.8344051446945338,
      "grad_norm": 1.3393827676773071,
      "learning_rate": 5.853510895883777e-05,
      "loss": 1.2442,
      "step": 10380
    },
    {
      "epoch": 0.8352090032154341,
      "grad_norm": 1.7181309461593628,
      "learning_rate": 5.8494753833736894e-05,
      "loss": 1.0575,
      "step": 10390
    },
    {
      "epoch": 0.8360128617363344,
      "grad_norm": 1.2780259847640991,
      "learning_rate": 5.8454398708636005e-05,
      "loss": 1.1243,
      "step": 10400
    },
    {
      "epoch": 0.8368167202572347,
      "grad_norm": 2.2146565914154053,
      "learning_rate": 5.841404358353512e-05,
      "loss": 1.1439,
      "step": 10410
    },
    {
      "epoch": 0.837620578778135,
      "grad_norm": 2.1241023540496826,
      "learning_rate": 5.837368845843423e-05,
      "loss": 1.2429,
      "step": 10420
    },
    {
      "epoch": 0.8384244372990354,
      "grad_norm": 1.5142546892166138,
      "learning_rate": 5.833333333333334e-05,
      "loss": 1.1969,
      "step": 10430
    },
    {
      "epoch": 0.8392282958199357,
      "grad_norm": 1.250694751739502,
      "learning_rate": 5.829297820823245e-05,
      "loss": 1.2337,
      "step": 10440
    },
    {
      "epoch": 0.840032154340836,
      "grad_norm": 1.0889441967010498,
      "learning_rate": 5.825262308313156e-05,
      "loss": 1.1911,
      "step": 10450
    },
    {
      "epoch": 0.8408360128617364,
      "grad_norm": 1.120822787284851,
      "learning_rate": 5.8212267958030674e-05,
      "loss": 1.211,
      "step": 10460
    },
    {
      "epoch": 0.8416398713826366,
      "grad_norm": 1.4169942140579224,
      "learning_rate": 5.8171912832929785e-05,
      "loss": 1.1747,
      "step": 10470
    },
    {
      "epoch": 0.842443729903537,
      "grad_norm": 1.7394747734069824,
      "learning_rate": 5.8131557707828896e-05,
      "loss": 1.1991,
      "step": 10480
    },
    {
      "epoch": 0.8432475884244373,
      "grad_norm": 1.3221678733825684,
      "learning_rate": 5.809120258272801e-05,
      "loss": 1.1979,
      "step": 10490
    },
    {
      "epoch": 0.8440514469453376,
      "grad_norm": 1.1339091062545776,
      "learning_rate": 5.805084745762712e-05,
      "loss": 1.2497,
      "step": 10500
    },
    {
      "epoch": 0.844855305466238,
      "grad_norm": 1.2805355787277222,
      "learning_rate": 5.801049233252624e-05,
      "loss": 1.2663,
      "step": 10510
    },
    {
      "epoch": 0.8456591639871383,
      "grad_norm": 2.003892660140991,
      "learning_rate": 5.797013720742535e-05,
      "loss": 1.2009,
      "step": 10520
    },
    {
      "epoch": 0.8464630225080386,
      "grad_norm": 1.854628562927246,
      "learning_rate": 5.792978208232446e-05,
      "loss": 1.0923,
      "step": 10530
    },
    {
      "epoch": 0.8472668810289389,
      "grad_norm": 1.1635708808898926,
      "learning_rate": 5.788942695722357e-05,
      "loss": 1.0859,
      "step": 10540
    },
    {
      "epoch": 0.8480707395498392,
      "grad_norm": 1.0551413297653198,
      "learning_rate": 5.784907183212268e-05,
      "loss": 1.2706,
      "step": 10550
    },
    {
      "epoch": 0.8488745980707395,
      "grad_norm": 1.675723671913147,
      "learning_rate": 5.7808716707021794e-05,
      "loss": 1.3267,
      "step": 10560
    },
    {
      "epoch": 0.8496784565916399,
      "grad_norm": 1.2302464246749878,
      "learning_rate": 5.7768361581920905e-05,
      "loss": 1.1912,
      "step": 10570
    },
    {
      "epoch": 0.8504823151125402,
      "grad_norm": 1.1186902523040771,
      "learning_rate": 5.772800645682002e-05,
      "loss": 1.1371,
      "step": 10580
    },
    {
      "epoch": 0.8512861736334405,
      "grad_norm": 1.3289992809295654,
      "learning_rate": 5.768765133171913e-05,
      "loss": 1.2077,
      "step": 10590
    },
    {
      "epoch": 0.8520900321543409,
      "grad_norm": 1.3730969429016113,
      "learning_rate": 5.764729620661824e-05,
      "loss": 1.1428,
      "step": 10600
    },
    {
      "epoch": 0.8528938906752411,
      "grad_norm": 1.6732438802719116,
      "learning_rate": 5.760694108151735e-05,
      "loss": 1.0955,
      "step": 10610
    },
    {
      "epoch": 0.8536977491961415,
      "grad_norm": 1.1242672204971313,
      "learning_rate": 5.756658595641646e-05,
      "loss": 1.2269,
      "step": 10620
    },
    {
      "epoch": 0.8545016077170418,
      "grad_norm": 1.3563451766967773,
      "learning_rate": 5.7526230831315574e-05,
      "loss": 1.051,
      "step": 10630
    },
    {
      "epoch": 0.8553054662379421,
      "grad_norm": 1.6396784782409668,
      "learning_rate": 5.7485875706214685e-05,
      "loss": 1.2013,
      "step": 10640
    },
    {
      "epoch": 0.8561093247588425,
      "grad_norm": 1.704081416130066,
      "learning_rate": 5.744552058111381e-05,
      "loss": 1.113,
      "step": 10650
    },
    {
      "epoch": 0.8569131832797428,
      "grad_norm": 1.072421908378601,
      "learning_rate": 5.740516545601292e-05,
      "loss": 1.2712,
      "step": 10660
    },
    {
      "epoch": 0.8577170418006431,
      "grad_norm": 1.3971763849258423,
      "learning_rate": 5.736481033091203e-05,
      "loss": 1.0984,
      "step": 10670
    },
    {
      "epoch": 0.8585209003215434,
      "grad_norm": 1.2550153732299805,
      "learning_rate": 5.7324455205811144e-05,
      "loss": 1.1208,
      "step": 10680
    },
    {
      "epoch": 0.8593247588424437,
      "grad_norm": 1.531835913658142,
      "learning_rate": 5.7284100080710255e-05,
      "loss": 1.3186,
      "step": 10690
    },
    {
      "epoch": 0.860128617363344,
      "grad_norm": 1.407632827758789,
      "learning_rate": 5.724374495560937e-05,
      "loss": 1.0284,
      "step": 10700
    },
    {
      "epoch": 0.8609324758842444,
      "grad_norm": 1.1127479076385498,
      "learning_rate": 5.720338983050848e-05,
      "loss": 1.2213,
      "step": 10710
    },
    {
      "epoch": 0.8617363344051447,
      "grad_norm": 1.3501721620559692,
      "learning_rate": 5.716303470540759e-05,
      "loss": 1.1947,
      "step": 10720
    },
    {
      "epoch": 0.862540192926045,
      "grad_norm": 1.3318605422973633,
      "learning_rate": 5.71226795803067e-05,
      "loss": 1.1896,
      "step": 10730
    },
    {
      "epoch": 0.8633440514469454,
      "grad_norm": 1.1702275276184082,
      "learning_rate": 5.708232445520582e-05,
      "loss": 1.1036,
      "step": 10740
    },
    {
      "epoch": 0.8641479099678456,
      "grad_norm": 1.3750840425491333,
      "learning_rate": 5.704196933010493e-05,
      "loss": 1.1948,
      "step": 10750
    },
    {
      "epoch": 0.864951768488746,
      "grad_norm": 1.5514822006225586,
      "learning_rate": 5.700161420500404e-05,
      "loss": 1.0426,
      "step": 10760
    },
    {
      "epoch": 0.8657556270096463,
      "grad_norm": 1.1812459230422974,
      "learning_rate": 5.696125907990315e-05,
      "loss": 1.0519,
      "step": 10770
    },
    {
      "epoch": 0.8665594855305466,
      "grad_norm": 1.336189866065979,
      "learning_rate": 5.6920903954802264e-05,
      "loss": 1.3293,
      "step": 10780
    },
    {
      "epoch": 0.867363344051447,
      "grad_norm": 1.533086895942688,
      "learning_rate": 5.6880548829701376e-05,
      "loss": 1.0463,
      "step": 10790
    },
    {
      "epoch": 0.8681672025723473,
      "grad_norm": 1.7019423246383667,
      "learning_rate": 5.684019370460049e-05,
      "loss": 1.1906,
      "step": 10800
    },
    {
      "epoch": 0.8689710610932476,
      "grad_norm": 1.6200381517410278,
      "learning_rate": 5.67998385794996e-05,
      "loss": 1.1637,
      "step": 10810
    },
    {
      "epoch": 0.8697749196141479,
      "grad_norm": 1.5051641464233398,
      "learning_rate": 5.675948345439871e-05,
      "loss": 1.2058,
      "step": 10820
    },
    {
      "epoch": 0.8705787781350482,
      "grad_norm": 1.2912993431091309,
      "learning_rate": 5.671912832929782e-05,
      "loss": 1.2305,
      "step": 10830
    },
    {
      "epoch": 0.8713826366559485,
      "grad_norm": 1.2812238931655884,
      "learning_rate": 5.667877320419693e-05,
      "loss": 1.1353,
      "step": 10840
    },
    {
      "epoch": 0.8721864951768489,
      "grad_norm": 2.6196494102478027,
      "learning_rate": 5.6638418079096044e-05,
      "loss": 1.2835,
      "step": 10850
    },
    {
      "epoch": 0.8729903536977492,
      "grad_norm": 1.3909611701965332,
      "learning_rate": 5.6598062953995155e-05,
      "loss": 1.1568,
      "step": 10860
    },
    {
      "epoch": 0.8737942122186495,
      "grad_norm": 1.1649755239486694,
      "learning_rate": 5.655770782889427e-05,
      "loss": 1.1305,
      "step": 10870
    },
    {
      "epoch": 0.8745980707395499,
      "grad_norm": 1.537785291671753,
      "learning_rate": 5.651735270379338e-05,
      "loss": 1.2737,
      "step": 10880
    },
    {
      "epoch": 0.8754019292604501,
      "grad_norm": 1.1643513441085815,
      "learning_rate": 5.647699757869249e-05,
      "loss": 1.2144,
      "step": 10890
    },
    {
      "epoch": 0.8762057877813505,
      "grad_norm": 1.3522480726242065,
      "learning_rate": 5.64366424535916e-05,
      "loss": 1.2893,
      "step": 10900
    },
    {
      "epoch": 0.8770096463022508,
      "grad_norm": 1.7712152004241943,
      "learning_rate": 5.639628732849072e-05,
      "loss": 1.1344,
      "step": 10910
    },
    {
      "epoch": 0.8778135048231511,
      "grad_norm": 1.3849141597747803,
      "learning_rate": 5.635593220338984e-05,
      "loss": 1.2632,
      "step": 10920
    },
    {
      "epoch": 0.8786173633440515,
      "grad_norm": 1.3868497610092163,
      "learning_rate": 5.631557707828895e-05,
      "loss": 1.2128,
      "step": 10930
    },
    {
      "epoch": 0.8794212218649518,
      "grad_norm": 1.1606500148773193,
      "learning_rate": 5.627522195318806e-05,
      "loss": 1.1074,
      "step": 10940
    },
    {
      "epoch": 0.8802250803858521,
      "grad_norm": 1.3538761138916016,
      "learning_rate": 5.623486682808717e-05,
      "loss": 1.209,
      "step": 10950
    },
    {
      "epoch": 0.8810289389067524,
      "grad_norm": 1.223466157913208,
      "learning_rate": 5.619451170298628e-05,
      "loss": 1.1811,
      "step": 10960
    },
    {
      "epoch": 0.8818327974276527,
      "grad_norm": 1.899377465248108,
      "learning_rate": 5.61541565778854e-05,
      "loss": 1.0492,
      "step": 10970
    },
    {
      "epoch": 0.882636655948553,
      "grad_norm": 2.095160722732544,
      "learning_rate": 5.611380145278451e-05,
      "loss": 1.2333,
      "step": 10980
    },
    {
      "epoch": 0.8834405144694534,
      "grad_norm": 1.1777327060699463,
      "learning_rate": 5.607344632768362e-05,
      "loss": 1.192,
      "step": 10990
    },
    {
      "epoch": 0.8842443729903537,
      "grad_norm": 1.3834155797958374,
      "learning_rate": 5.6033091202582735e-05,
      "loss": 1.1716,
      "step": 11000
    },
    {
      "epoch": 0.885048231511254,
      "grad_norm": 3.961866855621338,
      "learning_rate": 5.5992736077481846e-05,
      "loss": 1.2473,
      "step": 11010
    },
    {
      "epoch": 0.8858520900321544,
      "grad_norm": 1.7471206188201904,
      "learning_rate": 5.595238095238096e-05,
      "loss": 1.2372,
      "step": 11020
    },
    {
      "epoch": 0.8866559485530546,
      "grad_norm": 1.2151943445205688,
      "learning_rate": 5.591202582728007e-05,
      "loss": 1.2137,
      "step": 11030
    },
    {
      "epoch": 0.887459807073955,
      "grad_norm": 2.229262351989746,
      "learning_rate": 5.587167070217918e-05,
      "loss": 1.338,
      "step": 11040
    },
    {
      "epoch": 0.8882636655948553,
      "grad_norm": 1.3745701313018799,
      "learning_rate": 5.583131557707829e-05,
      "loss": 1.1346,
      "step": 11050
    },
    {
      "epoch": 0.8890675241157556,
      "grad_norm": 1.6774941682815552,
      "learning_rate": 5.57909604519774e-05,
      "loss": 1.2882,
      "step": 11060
    },
    {
      "epoch": 0.889871382636656,
      "grad_norm": 1.7017117738723755,
      "learning_rate": 5.5750605326876514e-05,
      "loss": 1.1427,
      "step": 11070
    },
    {
      "epoch": 0.8906752411575563,
      "grad_norm": 1.3809226751327515,
      "learning_rate": 5.5710250201775626e-05,
      "loss": 1.2229,
      "step": 11080
    },
    {
      "epoch": 0.8914790996784566,
      "grad_norm": 1.2096409797668457,
      "learning_rate": 5.566989507667474e-05,
      "loss": 1.1795,
      "step": 11090
    },
    {
      "epoch": 0.8922829581993569,
      "grad_norm": 1.653446078300476,
      "learning_rate": 5.562953995157385e-05,
      "loss": 1.2252,
      "step": 11100
    },
    {
      "epoch": 0.8930868167202572,
      "grad_norm": 1.1945871114730835,
      "learning_rate": 5.558918482647296e-05,
      "loss": 1.1876,
      "step": 11110
    },
    {
      "epoch": 0.8938906752411575,
      "grad_norm": 1.0288665294647217,
      "learning_rate": 5.554882970137207e-05,
      "loss": 1.1355,
      "step": 11120
    },
    {
      "epoch": 0.8946945337620579,
      "grad_norm": 1.1207334995269775,
      "learning_rate": 5.550847457627118e-05,
      "loss": 1.1891,
      "step": 11130
    },
    {
      "epoch": 0.8954983922829582,
      "grad_norm": 1.56515371799469,
      "learning_rate": 5.54681194511703e-05,
      "loss": 1.233,
      "step": 11140
    },
    {
      "epoch": 0.8963022508038585,
      "grad_norm": 2.0786585807800293,
      "learning_rate": 5.542776432606941e-05,
      "loss": 1.1558,
      "step": 11150
    },
    {
      "epoch": 0.8971061093247589,
      "grad_norm": 1.1563533544540405,
      "learning_rate": 5.538740920096852e-05,
      "loss": 1.1564,
      "step": 11160
    },
    {
      "epoch": 0.8979099678456591,
      "grad_norm": 1.2741695642471313,
      "learning_rate": 5.5347054075867635e-05,
      "loss": 1.0598,
      "step": 11170
    },
    {
      "epoch": 0.8987138263665595,
      "grad_norm": 1.7394829988479614,
      "learning_rate": 5.530669895076675e-05,
      "loss": 1.0792,
      "step": 11180
    },
    {
      "epoch": 0.8995176848874598,
      "grad_norm": 2.2357265949249268,
      "learning_rate": 5.5266343825665864e-05,
      "loss": 1.1272,
      "step": 11190
    },
    {
      "epoch": 0.9003215434083601,
      "grad_norm": 1.740126371383667,
      "learning_rate": 5.5225988700564976e-05,
      "loss": 1.2022,
      "step": 11200
    },
    {
      "epoch": 0.9011254019292605,
      "grad_norm": 1.7292535305023193,
      "learning_rate": 5.5185633575464094e-05,
      "loss": 1.1981,
      "step": 11210
    },
    {
      "epoch": 0.9019292604501608,
      "grad_norm": 1.5324060916900635,
      "learning_rate": 5.5145278450363205e-05,
      "loss": 1.1449,
      "step": 11220
    },
    {
      "epoch": 0.9027331189710611,
      "grad_norm": 1.1555876731872559,
      "learning_rate": 5.5104923325262316e-05,
      "loss": 1.1939,
      "step": 11230
    },
    {
      "epoch": 0.9035369774919614,
      "grad_norm": 1.3041651248931885,
      "learning_rate": 5.506456820016143e-05,
      "loss": 1.2318,
      "step": 11240
    },
    {
      "epoch": 0.9043408360128617,
      "grad_norm": 1.3641607761383057,
      "learning_rate": 5.502421307506054e-05,
      "loss": 1.1277,
      "step": 11250
    },
    {
      "epoch": 0.905144694533762,
      "grad_norm": 1.4824761152267456,
      "learning_rate": 5.498385794995965e-05,
      "loss": 1.2096,
      "step": 11260
    },
    {
      "epoch": 0.9059485530546624,
      "grad_norm": 1.7768710851669312,
      "learning_rate": 5.494350282485876e-05,
      "loss": 1.1968,
      "step": 11270
    },
    {
      "epoch": 0.9067524115755627,
      "grad_norm": 1.0921682119369507,
      "learning_rate": 5.490314769975787e-05,
      "loss": 1.1655,
      "step": 11280
    },
    {
      "epoch": 0.907556270096463,
      "grad_norm": 1.1076873540878296,
      "learning_rate": 5.4862792574656985e-05,
      "loss": 1.1469,
      "step": 11290
    },
    {
      "epoch": 0.9083601286173634,
      "grad_norm": 2.4272866249084473,
      "learning_rate": 5.4822437449556096e-05,
      "loss": 1.2036,
      "step": 11300
    },
    {
      "epoch": 0.9091639871382636,
      "grad_norm": 1.5198862552642822,
      "learning_rate": 5.478208232445521e-05,
      "loss": 1.1712,
      "step": 11310
    },
    {
      "epoch": 0.909967845659164,
      "grad_norm": 1.6450421810150146,
      "learning_rate": 5.474172719935432e-05,
      "loss": 1.1633,
      "step": 11320
    },
    {
      "epoch": 0.9107717041800643,
      "grad_norm": 1.6671468019485474,
      "learning_rate": 5.470137207425343e-05,
      "loss": 1.0944,
      "step": 11330
    },
    {
      "epoch": 0.9115755627009646,
      "grad_norm": 1.7807215452194214,
      "learning_rate": 5.466101694915254e-05,
      "loss": 1.096,
      "step": 11340
    },
    {
      "epoch": 0.912379421221865,
      "grad_norm": 1.8122543096542358,
      "learning_rate": 5.462066182405165e-05,
      "loss": 1.0954,
      "step": 11350
    },
    {
      "epoch": 0.9131832797427653,
      "grad_norm": 1.3115532398223877,
      "learning_rate": 5.4580306698950764e-05,
      "loss": 1.0989,
      "step": 11360
    },
    {
      "epoch": 0.9139871382636656,
      "grad_norm": 2.4599406719207764,
      "learning_rate": 5.4539951573849876e-05,
      "loss": 1.166,
      "step": 11370
    },
    {
      "epoch": 0.9147909967845659,
      "grad_norm": 1.021290898323059,
      "learning_rate": 5.4499596448748994e-05,
      "loss": 1.1451,
      "step": 11380
    },
    {
      "epoch": 0.9155948553054662,
      "grad_norm": 1.0792917013168335,
      "learning_rate": 5.4459241323648105e-05,
      "loss": 1.2435,
      "step": 11390
    },
    {
      "epoch": 0.9163987138263665,
      "grad_norm": 1.6936174631118774,
      "learning_rate": 5.4418886198547216e-05,
      "loss": 1.2064,
      "step": 11400
    },
    {
      "epoch": 0.9172025723472669,
      "grad_norm": 1.6657130718231201,
      "learning_rate": 5.437853107344633e-05,
      "loss": 1.079,
      "step": 11410
    },
    {
      "epoch": 0.9180064308681672,
      "grad_norm": 1.8588035106658936,
      "learning_rate": 5.433817594834544e-05,
      "loss": 1.1328,
      "step": 11420
    },
    {
      "epoch": 0.9188102893890675,
      "grad_norm": 1.3137052059173584,
      "learning_rate": 5.429782082324455e-05,
      "loss": 1.1395,
      "step": 11430
    },
    {
      "epoch": 0.9196141479099679,
      "grad_norm": 1.2235493659973145,
      "learning_rate": 5.425746569814366e-05,
      "loss": 1.1534,
      "step": 11440
    },
    {
      "epoch": 0.9204180064308681,
      "grad_norm": 1.411837100982666,
      "learning_rate": 5.421711057304279e-05,
      "loss": 1.3355,
      "step": 11450
    },
    {
      "epoch": 0.9212218649517685,
      "grad_norm": 1.2228326797485352,
      "learning_rate": 5.41767554479419e-05,
      "loss": 1.2599,
      "step": 11460
    },
    {
      "epoch": 0.9220257234726688,
      "grad_norm": 2.3366942405700684,
      "learning_rate": 5.413640032284101e-05,
      "loss": 1.2152,
      "step": 11470
    },
    {
      "epoch": 0.9228295819935691,
      "grad_norm": 1.593866229057312,
      "learning_rate": 5.409604519774012e-05,
      "loss": 1.1455,
      "step": 11480
    },
    {
      "epoch": 0.9236334405144695,
      "grad_norm": 1.1690460443496704,
      "learning_rate": 5.405569007263923e-05,
      "loss": 1.1246,
      "step": 11490
    },
    {
      "epoch": 0.9244372990353698,
      "grad_norm": 1.9292722940444946,
      "learning_rate": 5.4015334947538344e-05,
      "loss": 1.1446,
      "step": 11500
    },
    {
      "epoch": 0.9252411575562701,
      "grad_norm": 1.288496732711792,
      "learning_rate": 5.3974979822437455e-05,
      "loss": 1.1215,
      "step": 11510
    },
    {
      "epoch": 0.9260450160771704,
      "grad_norm": 1.6341341733932495,
      "learning_rate": 5.3934624697336566e-05,
      "loss": 1.0706,
      "step": 11520
    },
    {
      "epoch": 0.9268488745980707,
      "grad_norm": 1.1956830024719238,
      "learning_rate": 5.389426957223568e-05,
      "loss": 1.1117,
      "step": 11530
    },
    {
      "epoch": 0.927652733118971,
      "grad_norm": 2.764254093170166,
      "learning_rate": 5.385391444713479e-05,
      "loss": 1.2636,
      "step": 11540
    },
    {
      "epoch": 0.9284565916398714,
      "grad_norm": 1.0441055297851562,
      "learning_rate": 5.38135593220339e-05,
      "loss": 1.1597,
      "step": 11550
    },
    {
      "epoch": 0.9292604501607717,
      "grad_norm": 1.1911731958389282,
      "learning_rate": 5.377320419693301e-05,
      "loss": 0.9624,
      "step": 11560
    },
    {
      "epoch": 0.930064308681672,
      "grad_norm": 1.9981105327606201,
      "learning_rate": 5.373284907183212e-05,
      "loss": 1.0886,
      "step": 11570
    },
    {
      "epoch": 0.9308681672025724,
      "grad_norm": 1.4466192722320557,
      "learning_rate": 5.3692493946731235e-05,
      "loss": 1.1698,
      "step": 11580
    },
    {
      "epoch": 0.9316720257234726,
      "grad_norm": 1.43555748462677,
      "learning_rate": 5.3652138821630346e-05,
      "loss": 1.0717,
      "step": 11590
    },
    {
      "epoch": 0.932475884244373,
      "grad_norm": 1.0620185136795044,
      "learning_rate": 5.361178369652946e-05,
      "loss": 1.2611,
      "step": 11600
    },
    {
      "epoch": 0.9332797427652733,
      "grad_norm": 1.1827366352081299,
      "learning_rate": 5.3571428571428575e-05,
      "loss": 1.2629,
      "step": 11610
    },
    {
      "epoch": 0.9340836012861736,
      "grad_norm": 1.6303887367248535,
      "learning_rate": 5.353107344632769e-05,
      "loss": 1.1004,
      "step": 11620
    },
    {
      "epoch": 0.934887459807074,
      "grad_norm": 1.6994215250015259,
      "learning_rate": 5.34907183212268e-05,
      "loss": 1.1702,
      "step": 11630
    },
    {
      "epoch": 0.9356913183279743,
      "grad_norm": 2.8394522666931152,
      "learning_rate": 5.345036319612591e-05,
      "loss": 1.2469,
      "step": 11640
    },
    {
      "epoch": 0.9364951768488746,
      "grad_norm": 1.0047556161880493,
      "learning_rate": 5.341000807102502e-05,
      "loss": 1.1197,
      "step": 11650
    },
    {
      "epoch": 0.9372990353697749,
      "grad_norm": 1.9585012197494507,
      "learning_rate": 5.336965294592413e-05,
      "loss": 1.1801,
      "step": 11660
    },
    {
      "epoch": 0.9381028938906752,
      "grad_norm": 1.2510764598846436,
      "learning_rate": 5.3329297820823244e-05,
      "loss": 1.0855,
      "step": 11670
    },
    {
      "epoch": 0.9389067524115756,
      "grad_norm": 1.580871343612671,
      "learning_rate": 5.3288942695722355e-05,
      "loss": 1.1472,
      "step": 11680
    },
    {
      "epoch": 0.9397106109324759,
      "grad_norm": 1.1376216411590576,
      "learning_rate": 5.3248587570621466e-05,
      "loss": 1.1334,
      "step": 11690
    },
    {
      "epoch": 0.9405144694533762,
      "grad_norm": 1.5755469799041748,
      "learning_rate": 5.320823244552058e-05,
      "loss": 1.1797,
      "step": 11700
    },
    {
      "epoch": 0.9413183279742765,
      "grad_norm": 1.6171435117721558,
      "learning_rate": 5.31678773204197e-05,
      "loss": 1.2272,
      "step": 11710
    },
    {
      "epoch": 0.9421221864951769,
      "grad_norm": 1.393620252609253,
      "learning_rate": 5.3127522195318814e-05,
      "loss": 1.3367,
      "step": 11720
    },
    {
      "epoch": 0.9429260450160771,
      "grad_norm": 1.172006607055664,
      "learning_rate": 5.3087167070217925e-05,
      "loss": 1.0395,
      "step": 11730
    },
    {
      "epoch": 0.9437299035369775,
      "grad_norm": 1.4762243032455444,
      "learning_rate": 5.304681194511704e-05,
      "loss": 1.1633,
      "step": 11740
    },
    {
      "epoch": 0.9445337620578779,
      "grad_norm": 1.6818078756332397,
      "learning_rate": 5.300645682001615e-05,
      "loss": 1.1805,
      "step": 11750
    },
    {
      "epoch": 0.9453376205787781,
      "grad_norm": 1.3330456018447876,
      "learning_rate": 5.296610169491526e-05,
      "loss": 1.2,
      "step": 11760
    },
    {
      "epoch": 0.9461414790996785,
      "grad_norm": 1.6012849807739258,
      "learning_rate": 5.292574656981437e-05,
      "loss": 1.082,
      "step": 11770
    },
    {
      "epoch": 0.9469453376205788,
      "grad_norm": 1.347171425819397,
      "learning_rate": 5.288539144471348e-05,
      "loss": 1.3004,
      "step": 11780
    },
    {
      "epoch": 0.9477491961414791,
      "grad_norm": 2.2929813861846924,
      "learning_rate": 5.2845036319612594e-05,
      "loss": 1.1575,
      "step": 11790
    },
    {
      "epoch": 0.9485530546623794,
      "grad_norm": 1.209058165550232,
      "learning_rate": 5.2804681194511705e-05,
      "loss": 1.1463,
      "step": 11800
    },
    {
      "epoch": 0.9493569131832797,
      "grad_norm": 1.7847411632537842,
      "learning_rate": 5.2764326069410816e-05,
      "loss": 1.1799,
      "step": 11810
    },
    {
      "epoch": 0.9501607717041801,
      "grad_norm": 1.4044344425201416,
      "learning_rate": 5.272397094430993e-05,
      "loss": 1.1854,
      "step": 11820
    },
    {
      "epoch": 0.9509646302250804,
      "grad_norm": 1.4244343042373657,
      "learning_rate": 5.268361581920904e-05,
      "loss": 1.2356,
      "step": 11830
    },
    {
      "epoch": 0.9517684887459807,
      "grad_norm": 1.699362874031067,
      "learning_rate": 5.264326069410816e-05,
      "loss": 1.1849,
      "step": 11840
    },
    {
      "epoch": 0.952572347266881,
      "grad_norm": 3.909731864929199,
      "learning_rate": 5.260290556900727e-05,
      "loss": 1.0881,
      "step": 11850
    },
    {
      "epoch": 0.9533762057877814,
      "grad_norm": 1.665794849395752,
      "learning_rate": 5.256255044390638e-05,
      "loss": 1.1655,
      "step": 11860
    },
    {
      "epoch": 0.9541800643086816,
      "grad_norm": 2.0441787242889404,
      "learning_rate": 5.252219531880549e-05,
      "loss": 1.2538,
      "step": 11870
    },
    {
      "epoch": 0.954983922829582,
      "grad_norm": 1.2380964756011963,
      "learning_rate": 5.24818401937046e-05,
      "loss": 1.1699,
      "step": 11880
    },
    {
      "epoch": 0.9557877813504824,
      "grad_norm": 1.538638949394226,
      "learning_rate": 5.2441485068603714e-05,
      "loss": 1.3367,
      "step": 11890
    },
    {
      "epoch": 0.9565916398713826,
      "grad_norm": 1.4636310338974,
      "learning_rate": 5.2401129943502825e-05,
      "loss": 1.2077,
      "step": 11900
    },
    {
      "epoch": 0.957395498392283,
      "grad_norm": 1.2368042469024658,
      "learning_rate": 5.236077481840194e-05,
      "loss": 1.2506,
      "step": 11910
    },
    {
      "epoch": 0.9581993569131833,
      "grad_norm": 1.5975849628448486,
      "learning_rate": 5.232041969330105e-05,
      "loss": 1.2262,
      "step": 11920
    },
    {
      "epoch": 0.9590032154340836,
      "grad_norm": 1.2507879734039307,
      "learning_rate": 5.228006456820016e-05,
      "loss": 1.1564,
      "step": 11930
    },
    {
      "epoch": 0.9598070739549839,
      "grad_norm": 1.380014419555664,
      "learning_rate": 5.223970944309927e-05,
      "loss": 1.1145,
      "step": 11940
    },
    {
      "epoch": 0.9606109324758842,
      "grad_norm": 1.2493743896484375,
      "learning_rate": 5.219935431799838e-05,
      "loss": 1.1854,
      "step": 11950
    },
    {
      "epoch": 0.9614147909967846,
      "grad_norm": 1.350716233253479,
      "learning_rate": 5.2158999192897494e-05,
      "loss": 1.1534,
      "step": 11960
    },
    {
      "epoch": 0.9622186495176849,
      "grad_norm": NaN,
      "learning_rate": 5.21226795803067e-05,
      "loss": 1.2298,
      "step": 11970
    },
    {
      "epoch": 0.9630225080385852,
      "grad_norm": 1.5916483402252197,
      "learning_rate": 5.208232445520581e-05,
      "loss": 1.233,
      "step": 11980
    },
    {
      "epoch": 0.9638263665594855,
      "grad_norm": 1.3891382217407227,
      "learning_rate": 5.2041969330104924e-05,
      "loss": 1.33,
      "step": 11990
    },
    {
      "epoch": 0.9646302250803859,
      "grad_norm": 1.6548006534576416,
      "learning_rate": 5.2001614205004035e-05,
      "loss": 1.2027,
      "step": 12000
    },
    {
      "epoch": 0.9646302250803859,
      "eval_yahma/alpaca-cleaned_loss": 1.2050625085830688,
      "eval_yahma/alpaca-cleaned_runtime": 115.646,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.294,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.162,
      "step": 12000
    },
    {
      "epoch": 0.9654340836012861,
      "grad_norm": 1.2655686140060425,
      "learning_rate": 5.196125907990315e-05,
      "loss": 1.1269,
      "step": 12010
    },
    {
      "epoch": 0.9662379421221865,
      "grad_norm": 1.2995678186416626,
      "learning_rate": 5.192090395480226e-05,
      "loss": 1.1227,
      "step": 12020
    },
    {
      "epoch": 0.9670418006430869,
      "grad_norm": 1.0736733675003052,
      "learning_rate": 5.188054882970137e-05,
      "loss": 1.2294,
      "step": 12030
    },
    {
      "epoch": 0.9678456591639871,
      "grad_norm": 1.19216787815094,
      "learning_rate": 5.184019370460048e-05,
      "loss": 1.2031,
      "step": 12040
    },
    {
      "epoch": 0.9686495176848875,
      "grad_norm": 1.541259765625,
      "learning_rate": 5.1799838579499606e-05,
      "loss": 1.1046,
      "step": 12050
    },
    {
      "epoch": 0.9694533762057878,
      "grad_norm": 1.6200529336929321,
      "learning_rate": 5.175948345439872e-05,
      "loss": 1.1826,
      "step": 12060
    },
    {
      "epoch": 0.9702572347266881,
      "grad_norm": 1.4181767702102661,
      "learning_rate": 5.171912832929783e-05,
      "loss": 1.1438,
      "step": 12070
    },
    {
      "epoch": 0.9710610932475884,
      "grad_norm": 1.6790999174118042,
      "learning_rate": 5.167877320419694e-05,
      "loss": 1.1761,
      "step": 12080
    },
    {
      "epoch": 0.9718649517684887,
      "grad_norm": 1.2507883310317993,
      "learning_rate": 5.163841807909605e-05,
      "loss": 1.2146,
      "step": 12090
    },
    {
      "epoch": 0.9726688102893891,
      "grad_norm": 1.203534483909607,
      "learning_rate": 5.159806295399516e-05,
      "loss": 1.3176,
      "step": 12100
    },
    {
      "epoch": 0.9734726688102894,
      "grad_norm": 1.6226143836975098,
      "learning_rate": 5.1557707828894274e-05,
      "loss": 1.1501,
      "step": 12110
    },
    {
      "epoch": 0.9742765273311897,
      "grad_norm": 1.6095842123031616,
      "learning_rate": 5.1517352703793385e-05,
      "loss": 1.0896,
      "step": 12120
    },
    {
      "epoch": 0.97508038585209,
      "grad_norm": 1.9545879364013672,
      "learning_rate": 5.14769975786925e-05,
      "loss": 1.2088,
      "step": 12130
    },
    {
      "epoch": 0.9758842443729904,
      "grad_norm": 1.478491187095642,
      "learning_rate": 5.143664245359161e-05,
      "loss": 1.1762,
      "step": 12140
    },
    {
      "epoch": 0.9766881028938906,
      "grad_norm": 1.239725947380066,
      "learning_rate": 5.139628732849072e-05,
      "loss": 1.2697,
      "step": 12150
    },
    {
      "epoch": 0.977491961414791,
      "grad_norm": 1.663704514503479,
      "learning_rate": 5.135593220338983e-05,
      "loss": 1.2736,
      "step": 12160
    },
    {
      "epoch": 0.9782958199356914,
      "grad_norm": 1.4106169939041138,
      "learning_rate": 5.131557707828895e-05,
      "loss": 1.0439,
      "step": 12170
    },
    {
      "epoch": 0.9790996784565916,
      "grad_norm": 1.6039048433303833,
      "learning_rate": 5.127522195318806e-05,
      "loss": 1.2114,
      "step": 12180
    },
    {
      "epoch": 0.979903536977492,
      "grad_norm": 1.2211772203445435,
      "learning_rate": 5.123486682808717e-05,
      "loss": 1.246,
      "step": 12190
    },
    {
      "epoch": 0.9807073954983923,
      "grad_norm": 1.741607904434204,
      "learning_rate": 5.119451170298628e-05,
      "loss": 1.1022,
      "step": 12200
    },
    {
      "epoch": 0.9815112540192926,
      "grad_norm": 2.1637446880340576,
      "learning_rate": 5.1154156577885394e-05,
      "loss": 1.1241,
      "step": 12210
    },
    {
      "epoch": 0.9823151125401929,
      "grad_norm": 1.2326878309249878,
      "learning_rate": 5.1113801452784506e-05,
      "loss": 1.1947,
      "step": 12220
    },
    {
      "epoch": 0.9831189710610932,
      "grad_norm": 1.2424410581588745,
      "learning_rate": 5.107344632768362e-05,
      "loss": 1.1497,
      "step": 12230
    },
    {
      "epoch": 0.9839228295819936,
      "grad_norm": 1.4629040956497192,
      "learning_rate": 5.103309120258273e-05,
      "loss": 1.1187,
      "step": 12240
    },
    {
      "epoch": 0.9847266881028939,
      "grad_norm": 1.7737675905227661,
      "learning_rate": 5.099273607748184e-05,
      "loss": 1.1671,
      "step": 12250
    },
    {
      "epoch": 0.9855305466237942,
      "grad_norm": 2.268542766571045,
      "learning_rate": 5.095238095238095e-05,
      "loss": 1.1304,
      "step": 12260
    },
    {
      "epoch": 0.9863344051446945,
      "grad_norm": 1.690609097480774,
      "learning_rate": 5.091202582728006e-05,
      "loss": 1.1014,
      "step": 12270
    },
    {
      "epoch": 0.9871382636655949,
      "grad_norm": 1.7486388683319092,
      "learning_rate": 5.0871670702179174e-05,
      "loss": 1.1526,
      "step": 12280
    },
    {
      "epoch": 0.9879421221864951,
      "grad_norm": 1.747187614440918,
      "learning_rate": 5.0831315577078285e-05,
      "loss": 0.9875,
      "step": 12290
    },
    {
      "epoch": 0.9887459807073955,
      "grad_norm": 2.9285736083984375,
      "learning_rate": 5.0790960451977397e-05,
      "loss": 1.1122,
      "step": 12300
    },
    {
      "epoch": 0.9895498392282959,
      "grad_norm": 1.5583657026290894,
      "learning_rate": 5.075060532687651e-05,
      "loss": 1.1447,
      "step": 12310
    },
    {
      "epoch": 0.9903536977491961,
      "grad_norm": 1.8540047407150269,
      "learning_rate": 5.071025020177563e-05,
      "loss": 1.1729,
      "step": 12320
    },
    {
      "epoch": 0.9911575562700965,
      "grad_norm": 1.5402899980545044,
      "learning_rate": 5.0669895076674744e-05,
      "loss": 1.1523,
      "step": 12330
    },
    {
      "epoch": 0.9919614147909968,
      "grad_norm": 1.3348851203918457,
      "learning_rate": 5.0629539951573856e-05,
      "loss": 1.2239,
      "step": 12340
    },
    {
      "epoch": 0.9927652733118971,
      "grad_norm": 1.8120102882385254,
      "learning_rate": 5.058918482647297e-05,
      "loss": 1.2171,
      "step": 12350
    },
    {
      "epoch": 0.9935691318327974,
      "grad_norm": 1.2614649534225464,
      "learning_rate": 5.054882970137208e-05,
      "loss": 1.22,
      "step": 12360
    },
    {
      "epoch": 0.9943729903536977,
      "grad_norm": 1.311959981918335,
      "learning_rate": 5.050847457627119e-05,
      "loss": 1.0745,
      "step": 12370
    },
    {
      "epoch": 0.9951768488745981,
      "grad_norm": 1.1874438524246216,
      "learning_rate": 5.04681194511703e-05,
      "loss": 1.2944,
      "step": 12380
    },
    {
      "epoch": 0.9959807073954984,
      "grad_norm": 1.3048447370529175,
      "learning_rate": 5.042776432606941e-05,
      "loss": 1.1719,
      "step": 12390
    },
    {
      "epoch": 0.9967845659163987,
      "grad_norm": 1.0826609134674072,
      "learning_rate": 5.038740920096853e-05,
      "loss": 1.1412,
      "step": 12400
    },
    {
      "epoch": 0.997588424437299,
      "grad_norm": 1.49656343460083,
      "learning_rate": 5.034705407586764e-05,
      "loss": 1.1898,
      "step": 12410
    },
    {
      "epoch": 0.9983922829581994,
      "grad_norm": 1.256178617477417,
      "learning_rate": 5.030669895076675e-05,
      "loss": 1.1431,
      "step": 12420
    },
    {
      "epoch": 0.9991961414790996,
      "grad_norm": 1.5236985683441162,
      "learning_rate": 5.0266343825665865e-05,
      "loss": 1.1061,
      "step": 12430
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.45755934715271,
      "learning_rate": 5.0225988700564976e-05,
      "loss": 1.1797,
      "step": 12440
    },
    {
      "epoch": 1.0008038585209003,
      "grad_norm": 1.336484670639038,
      "learning_rate": 5.018563357546409e-05,
      "loss": 0.9872,
      "step": 12450
    },
    {
      "epoch": 1.0016077170418007,
      "grad_norm": 1.2816264629364014,
      "learning_rate": 5.01452784503632e-05,
      "loss": 1.0556,
      "step": 12460
    },
    {
      "epoch": 1.002411575562701,
      "grad_norm": 2.1287145614624023,
      "learning_rate": 5.010492332526231e-05,
      "loss": 1.1472,
      "step": 12470
    },
    {
      "epoch": 1.0032154340836013,
      "grad_norm": 2.0706117153167725,
      "learning_rate": 5.006456820016142e-05,
      "loss": 1.002,
      "step": 12480
    },
    {
      "epoch": 1.0040192926045015,
      "grad_norm": 1.6112709045410156,
      "learning_rate": 5.002421307506053e-05,
      "loss": 0.9798,
      "step": 12490
    },
    {
      "epoch": 1.004823151125402,
      "grad_norm": 2.151376247406006,
      "learning_rate": 4.9983857949959644e-05,
      "loss": 0.9878,
      "step": 12500
    },
    {
      "epoch": 1.0056270096463023,
      "grad_norm": 2.537773609161377,
      "learning_rate": 4.994350282485876e-05,
      "loss": 1.1223,
      "step": 12510
    },
    {
      "epoch": 1.0064308681672025,
      "grad_norm": 1.542268991470337,
      "learning_rate": 4.9903147699757874e-05,
      "loss": 1.1358,
      "step": 12520
    },
    {
      "epoch": 1.007234726688103,
      "grad_norm": 2.475477695465088,
      "learning_rate": 4.9862792574656985e-05,
      "loss": 1.0899,
      "step": 12530
    },
    {
      "epoch": 1.0080385852090032,
      "grad_norm": 1.7700741291046143,
      "learning_rate": 4.9822437449556096e-05,
      "loss": 1.1545,
      "step": 12540
    },
    {
      "epoch": 1.0088424437299035,
      "grad_norm": 1.5795843601226807,
      "learning_rate": 4.978208232445521e-05,
      "loss": 1.1137,
      "step": 12550
    },
    {
      "epoch": 1.0096463022508038,
      "grad_norm": 1.2464009523391724,
      "learning_rate": 4.974172719935432e-05,
      "loss": 1.0397,
      "step": 12560
    },
    {
      "epoch": 1.0104501607717042,
      "grad_norm": 1.3018213510513306,
      "learning_rate": 4.970137207425343e-05,
      "loss": 1.0142,
      "step": 12570
    },
    {
      "epoch": 1.0112540192926045,
      "grad_norm": 1.30471932888031,
      "learning_rate": 4.966101694915254e-05,
      "loss": 0.9567,
      "step": 12580
    },
    {
      "epoch": 1.0120578778135048,
      "grad_norm": 1.6906229257583618,
      "learning_rate": 4.962066182405165e-05,
      "loss": 0.8977,
      "step": 12590
    },
    {
      "epoch": 1.0128617363344052,
      "grad_norm": 1.269845724105835,
      "learning_rate": 4.958030669895077e-05,
      "loss": 1.0785,
      "step": 12600
    },
    {
      "epoch": 1.0136655948553055,
      "grad_norm": 1.7904671430587769,
      "learning_rate": 4.953995157384988e-05,
      "loss": 0.9146,
      "step": 12610
    },
    {
      "epoch": 1.0144694533762058,
      "grad_norm": 2.0014989376068115,
      "learning_rate": 4.9499596448748994e-05,
      "loss": 1.0232,
      "step": 12620
    },
    {
      "epoch": 1.015273311897106,
      "grad_norm": 1.4139068126678467,
      "learning_rate": 4.9459241323648106e-05,
      "loss": 1.0038,
      "step": 12630
    },
    {
      "epoch": 1.0160771704180065,
      "grad_norm": 1.4505324363708496,
      "learning_rate": 4.941888619854722e-05,
      "loss": 0.9644,
      "step": 12640
    },
    {
      "epoch": 1.0168810289389068,
      "grad_norm": 1.2977114915847778,
      "learning_rate": 4.9378531073446335e-05,
      "loss": 1.1053,
      "step": 12650
    },
    {
      "epoch": 1.017684887459807,
      "grad_norm": 1.6965181827545166,
      "learning_rate": 4.9338175948345446e-05,
      "loss": 1.0649,
      "step": 12660
    },
    {
      "epoch": 1.0184887459807075,
      "grad_norm": 1.7907180786132812,
      "learning_rate": 4.929782082324456e-05,
      "loss": 0.9434,
      "step": 12670
    },
    {
      "epoch": 1.0192926045016077,
      "grad_norm": 1.6487858295440674,
      "learning_rate": 4.925746569814367e-05,
      "loss": 1.0578,
      "step": 12680
    },
    {
      "epoch": 1.020096463022508,
      "grad_norm": 2.0607364177703857,
      "learning_rate": 4.921711057304278e-05,
      "loss": 0.9846,
      "step": 12690
    },
    {
      "epoch": 1.0209003215434083,
      "grad_norm": 1.3130086660385132,
      "learning_rate": 4.917675544794189e-05,
      "loss": 1.1002,
      "step": 12700
    },
    {
      "epoch": 1.0217041800643087,
      "grad_norm": 1.415685772895813,
      "learning_rate": 4.9136400322841e-05,
      "loss": 1.0871,
      "step": 12710
    },
    {
      "epoch": 1.022508038585209,
      "grad_norm": 1.438016653060913,
      "learning_rate": 4.9096045197740115e-05,
      "loss": 1.0543,
      "step": 12720
    },
    {
      "epoch": 1.0233118971061093,
      "grad_norm": 2.078965663909912,
      "learning_rate": 4.9055690072639226e-05,
      "loss": 0.9583,
      "step": 12730
    },
    {
      "epoch": 1.0241157556270097,
      "grad_norm": 1.4178944826126099,
      "learning_rate": 4.901533494753834e-05,
      "loss": 0.901,
      "step": 12740
    },
    {
      "epoch": 1.02491961414791,
      "grad_norm": 2.025350570678711,
      "learning_rate": 4.897497982243745e-05,
      "loss": 1.0372,
      "step": 12750
    },
    {
      "epoch": 1.0257234726688103,
      "grad_norm": 1.9596234560012817,
      "learning_rate": 4.893462469733656e-05,
      "loss": 1.048,
      "step": 12760
    },
    {
      "epoch": 1.0265273311897105,
      "grad_norm": 1.2961586713790894,
      "learning_rate": 4.889426957223567e-05,
      "loss": 1.0288,
      "step": 12770
    },
    {
      "epoch": 1.027331189710611,
      "grad_norm": 1.3893296718597412,
      "learning_rate": 4.885391444713479e-05,
      "loss": 1.0256,
      "step": 12780
    },
    {
      "epoch": 1.0281350482315113,
      "grad_norm": 1.41170072555542,
      "learning_rate": 4.88135593220339e-05,
      "loss": 0.9853,
      "step": 12790
    },
    {
      "epoch": 1.0289389067524115,
      "grad_norm": 1.3405449390411377,
      "learning_rate": 4.877320419693301e-05,
      "loss": 0.9838,
      "step": 12800
    },
    {
      "epoch": 1.029742765273312,
      "grad_norm": 1.413405179977417,
      "learning_rate": 4.8732849071832124e-05,
      "loss": 1.0329,
      "step": 12810
    },
    {
      "epoch": 1.0305466237942122,
      "grad_norm": 1.324311375617981,
      "learning_rate": 4.8692493946731235e-05,
      "loss": 1.0136,
      "step": 12820
    },
    {
      "epoch": 1.0313504823151125,
      "grad_norm": 1.9724149703979492,
      "learning_rate": 4.8652138821630346e-05,
      "loss": 0.9798,
      "step": 12830
    },
    {
      "epoch": 1.0321543408360128,
      "grad_norm": 2.3491575717926025,
      "learning_rate": 4.8611783696529465e-05,
      "loss": 0.9165,
      "step": 12840
    },
    {
      "epoch": 1.0329581993569132,
      "grad_norm": 1.1819404363632202,
      "learning_rate": 4.8571428571428576e-05,
      "loss": 1.0794,
      "step": 12850
    },
    {
      "epoch": 1.0337620578778135,
      "grad_norm": 1.3069233894348145,
      "learning_rate": 4.853107344632769e-05,
      "loss": 1.0239,
      "step": 12860
    },
    {
      "epoch": 1.0345659163987138,
      "grad_norm": 1.3474875688552856,
      "learning_rate": 4.84907183212268e-05,
      "loss": 1.0174,
      "step": 12870
    },
    {
      "epoch": 1.0353697749196142,
      "grad_norm": 2.287572145462036,
      "learning_rate": 4.845036319612591e-05,
      "loss": 0.8799,
      "step": 12880
    },
    {
      "epoch": 1.0361736334405145,
      "grad_norm": 1.2526694536209106,
      "learning_rate": 4.841000807102502e-05,
      "loss": 1.0667,
      "step": 12890
    },
    {
      "epoch": 1.0369774919614148,
      "grad_norm": 1.3657410144805908,
      "learning_rate": 4.836965294592413e-05,
      "loss": 1.1423,
      "step": 12900
    },
    {
      "epoch": 1.037781350482315,
      "grad_norm": 1.8211760520935059,
      "learning_rate": 4.832929782082325e-05,
      "loss": 1.0129,
      "step": 12910
    },
    {
      "epoch": 1.0385852090032155,
      "grad_norm": 1.362104058265686,
      "learning_rate": 4.828894269572236e-05,
      "loss": 0.9487,
      "step": 12920
    },
    {
      "epoch": 1.0393890675241158,
      "grad_norm": 1.7094135284423828,
      "learning_rate": 4.8248587570621474e-05,
      "loss": 1.0893,
      "step": 12930
    },
    {
      "epoch": 1.040192926045016,
      "grad_norm": 1.6447094678878784,
      "learning_rate": 4.8208232445520585e-05,
      "loss": 0.9119,
      "step": 12940
    },
    {
      "epoch": 1.0409967845659165,
      "grad_norm": 1.317784070968628,
      "learning_rate": 4.8167877320419696e-05,
      "loss": 1.0388,
      "step": 12950
    },
    {
      "epoch": 1.0418006430868167,
      "grad_norm": 3.135857105255127,
      "learning_rate": 4.812752219531881e-05,
      "loss": 0.9671,
      "step": 12960
    },
    {
      "epoch": 1.042604501607717,
      "grad_norm": 1.596756935119629,
      "learning_rate": 4.808716707021792e-05,
      "loss": 0.956,
      "step": 12970
    },
    {
      "epoch": 1.0434083601286173,
      "grad_norm": 1.973957896232605,
      "learning_rate": 4.804681194511703e-05,
      "loss": 1.1145,
      "step": 12980
    },
    {
      "epoch": 1.0442122186495177,
      "grad_norm": 1.675784707069397,
      "learning_rate": 4.800645682001614e-05,
      "loss": 0.9583,
      "step": 12990
    },
    {
      "epoch": 1.045016077170418,
      "grad_norm": 2.1188571453094482,
      "learning_rate": 4.796610169491525e-05,
      "loss": 1.2259,
      "step": 13000
    },
    {
      "epoch": 1.0458199356913183,
      "grad_norm": 2.490846872329712,
      "learning_rate": 4.7925746569814365e-05,
      "loss": 1.0112,
      "step": 13010
    },
    {
      "epoch": 1.0466237942122187,
      "grad_norm": 2.144216299057007,
      "learning_rate": 4.7885391444713476e-05,
      "loss": 0.9891,
      "step": 13020
    },
    {
      "epoch": 1.047427652733119,
      "grad_norm": 1.3466781377792358,
      "learning_rate": 4.784503631961259e-05,
      "loss": 1.0314,
      "step": 13030
    },
    {
      "epoch": 1.0482315112540193,
      "grad_norm": 1.44432532787323,
      "learning_rate": 4.7804681194511705e-05,
      "loss": 1.028,
      "step": 13040
    },
    {
      "epoch": 1.0490353697749195,
      "grad_norm": 1.6085883378982544,
      "learning_rate": 4.776432606941082e-05,
      "loss": 1.0979,
      "step": 13050
    },
    {
      "epoch": 1.04983922829582,
      "grad_norm": 2.7709875106811523,
      "learning_rate": 4.772397094430993e-05,
      "loss": 0.9387,
      "step": 13060
    },
    {
      "epoch": 1.0506430868167203,
      "grad_norm": 1.5809946060180664,
      "learning_rate": 4.7683615819209046e-05,
      "loss": 1.0414,
      "step": 13070
    },
    {
      "epoch": 1.0514469453376205,
      "grad_norm": 3.7606260776519775,
      "learning_rate": 4.764326069410816e-05,
      "loss": 1.1012,
      "step": 13080
    },
    {
      "epoch": 1.052250803858521,
      "grad_norm": 1.3225197792053223,
      "learning_rate": 4.760290556900727e-05,
      "loss": 0.9132,
      "step": 13090
    },
    {
      "epoch": 1.0530546623794212,
      "grad_norm": 1.6168928146362305,
      "learning_rate": 4.756255044390638e-05,
      "loss": 0.9281,
      "step": 13100
    },
    {
      "epoch": 1.0538585209003215,
      "grad_norm": 1.2707538604736328,
      "learning_rate": 4.752219531880549e-05,
      "loss": 1.0895,
      "step": 13110
    },
    {
      "epoch": 1.0546623794212218,
      "grad_norm": 1.8395226001739502,
      "learning_rate": 4.74818401937046e-05,
      "loss": 1.1246,
      "step": 13120
    },
    {
      "epoch": 1.0554662379421222,
      "grad_norm": 1.9540048837661743,
      "learning_rate": 4.7441485068603714e-05,
      "loss": 0.8961,
      "step": 13130
    },
    {
      "epoch": 1.0562700964630225,
      "grad_norm": 3.629779100418091,
      "learning_rate": 4.7401129943502826e-05,
      "loss": 0.9347,
      "step": 13140
    },
    {
      "epoch": 1.0570739549839228,
      "grad_norm": 1.8094236850738525,
      "learning_rate": 4.736077481840194e-05,
      "loss": 0.9579,
      "step": 13150
    },
    {
      "epoch": 1.0578778135048232,
      "grad_norm": 1.5016402006149292,
      "learning_rate": 4.732041969330105e-05,
      "loss": 0.9967,
      "step": 13160
    },
    {
      "epoch": 1.0586816720257235,
      "grad_norm": 3.5953845977783203,
      "learning_rate": 4.728006456820017e-05,
      "loss": 0.9922,
      "step": 13170
    },
    {
      "epoch": 1.0594855305466238,
      "grad_norm": 1.4096349477767944,
      "learning_rate": 4.723970944309928e-05,
      "loss": 1.0291,
      "step": 13180
    },
    {
      "epoch": 1.060289389067524,
      "grad_norm": 1.606210470199585,
      "learning_rate": 4.719935431799839e-05,
      "loss": 1.0506,
      "step": 13190
    },
    {
      "epoch": 1.0610932475884245,
      "grad_norm": 1.6466732025146484,
      "learning_rate": 4.71589991928975e-05,
      "loss": 1.0332,
      "step": 13200
    },
    {
      "epoch": 1.0618971061093248,
      "grad_norm": 1.3050963878631592,
      "learning_rate": 4.711864406779661e-05,
      "loss": 1.0517,
      "step": 13210
    },
    {
      "epoch": 1.062700964630225,
      "grad_norm": 1.3689507246017456,
      "learning_rate": 4.7078288942695723e-05,
      "loss": 1.0825,
      "step": 13220
    },
    {
      "epoch": 1.0635048231511255,
      "grad_norm": 1.8197163343429565,
      "learning_rate": 4.7037933817594835e-05,
      "loss": 1.0044,
      "step": 13230
    },
    {
      "epoch": 1.0643086816720257,
      "grad_norm": 1.6934906244277954,
      "learning_rate": 4.6997578692493946e-05,
      "loss": 0.9633,
      "step": 13240
    },
    {
      "epoch": 1.065112540192926,
      "grad_norm": 2.1707937717437744,
      "learning_rate": 4.695722356739306e-05,
      "loss": 1.1439,
      "step": 13250
    },
    {
      "epoch": 1.0659163987138263,
      "grad_norm": 1.2857168912887573,
      "learning_rate": 4.691686844229217e-05,
      "loss": 1.079,
      "step": 13260
    },
    {
      "epoch": 1.0667202572347267,
      "grad_norm": 1.3009968996047974,
      "learning_rate": 4.687651331719129e-05,
      "loss": 0.9304,
      "step": 13270
    },
    {
      "epoch": 1.067524115755627,
      "grad_norm": 2.0484039783477783,
      "learning_rate": 4.68361581920904e-05,
      "loss": 1.0075,
      "step": 13280
    },
    {
      "epoch": 1.0683279742765273,
      "grad_norm": 2.504826545715332,
      "learning_rate": 4.679580306698951e-05,
      "loss": 0.9569,
      "step": 13290
    },
    {
      "epoch": 1.0691318327974277,
      "grad_norm": 1.8763548135757446,
      "learning_rate": 4.675544794188862e-05,
      "loss": 1.0155,
      "step": 13300
    },
    {
      "epoch": 1.069935691318328,
      "grad_norm": 1.8800406455993652,
      "learning_rate": 4.671509281678774e-05,
      "loss": 0.9565,
      "step": 13310
    },
    {
      "epoch": 1.0707395498392283,
      "grad_norm": 1.3298194408416748,
      "learning_rate": 4.667473769168685e-05,
      "loss": 0.9434,
      "step": 13320
    },
    {
      "epoch": 1.0715434083601285,
      "grad_norm": 1.45872163772583,
      "learning_rate": 4.663438256658596e-05,
      "loss": 0.9882,
      "step": 13330
    },
    {
      "epoch": 1.072347266881029,
      "grad_norm": 2.167778730392456,
      "learning_rate": 4.6594027441485073e-05,
      "loss": 1.0759,
      "step": 13340
    },
    {
      "epoch": 1.0731511254019293,
      "grad_norm": 2.0066630840301514,
      "learning_rate": 4.6553672316384185e-05,
      "loss": 0.9653,
      "step": 13350
    },
    {
      "epoch": 1.0739549839228295,
      "grad_norm": 1.3754826784133911,
      "learning_rate": 4.6513317191283296e-05,
      "loss": 0.9961,
      "step": 13360
    },
    {
      "epoch": 1.07475884244373,
      "grad_norm": 1.6868056058883667,
      "learning_rate": 4.647296206618241e-05,
      "loss": 0.9634,
      "step": 13370
    },
    {
      "epoch": 1.0755627009646302,
      "grad_norm": 4.323541641235352,
      "learning_rate": 4.643260694108152e-05,
      "loss": 0.9814,
      "step": 13380
    },
    {
      "epoch": 1.0763665594855305,
      "grad_norm": 1.3567148447036743,
      "learning_rate": 4.639225181598063e-05,
      "loss": 1.0566,
      "step": 13390
    },
    {
      "epoch": 1.077170418006431,
      "grad_norm": 1.4276680946350098,
      "learning_rate": 4.635189669087974e-05,
      "loss": 1.0006,
      "step": 13400
    },
    {
      "epoch": 1.0779742765273312,
      "grad_norm": 1.8578076362609863,
      "learning_rate": 4.631154156577885e-05,
      "loss": 0.8861,
      "step": 13410
    },
    {
      "epoch": 1.0787781350482315,
      "grad_norm": 1.929571270942688,
      "learning_rate": 4.6271186440677964e-05,
      "loss": 0.9064,
      "step": 13420
    },
    {
      "epoch": 1.0795819935691318,
      "grad_norm": 1.605181336402893,
      "learning_rate": 4.6230831315577076e-05,
      "loss": 1.034,
      "step": 13430
    },
    {
      "epoch": 1.0803858520900322,
      "grad_norm": 1.9241151809692383,
      "learning_rate": 4.6190476190476194e-05,
      "loss": 1.1215,
      "step": 13440
    },
    {
      "epoch": 1.0811897106109325,
      "grad_norm": 3.210650682449341,
      "learning_rate": 4.6150121065375305e-05,
      "loss": 0.9467,
      "step": 13450
    },
    {
      "epoch": 1.0819935691318328,
      "grad_norm": 1.4049782752990723,
      "learning_rate": 4.6109765940274417e-05,
      "loss": 1.1126,
      "step": 13460
    },
    {
      "epoch": 1.082797427652733,
      "grad_norm": 1.3904056549072266,
      "learning_rate": 4.606941081517353e-05,
      "loss": 0.9465,
      "step": 13470
    },
    {
      "epoch": 1.0836012861736335,
      "grad_norm": 1.7483136653900146,
      "learning_rate": 4.602905569007264e-05,
      "loss": 1.0509,
      "step": 13480
    },
    {
      "epoch": 1.0844051446945338,
      "grad_norm": 1.6487077474594116,
      "learning_rate": 4.598870056497175e-05,
      "loss": 1.0358,
      "step": 13490
    },
    {
      "epoch": 1.085209003215434,
      "grad_norm": 1.433251976966858,
      "learning_rate": 4.594834543987087e-05,
      "loss": 0.9653,
      "step": 13500
    },
    {
      "epoch": 1.0860128617363345,
      "grad_norm": 1.551595687866211,
      "learning_rate": 4.590799031476998e-05,
      "loss": 1.0813,
      "step": 13510
    },
    {
      "epoch": 1.0868167202572347,
      "grad_norm": 1.8778648376464844,
      "learning_rate": 4.586763518966909e-05,
      "loss": 1.1041,
      "step": 13520
    },
    {
      "epoch": 1.087620578778135,
      "grad_norm": 1.5723669528961182,
      "learning_rate": 4.58272800645682e-05,
      "loss": 1.1798,
      "step": 13530
    },
    {
      "epoch": 1.0884244372990355,
      "grad_norm": 1.3904507160186768,
      "learning_rate": 4.5786924939467314e-05,
      "loss": 1.0971,
      "step": 13540
    },
    {
      "epoch": 1.0892282958199357,
      "grad_norm": 2.491220712661743,
      "learning_rate": 4.5746569814366426e-05,
      "loss": 1.0707,
      "step": 13550
    },
    {
      "epoch": 1.090032154340836,
      "grad_norm": 1.688567042350769,
      "learning_rate": 4.570621468926554e-05,
      "loss": 0.9696,
      "step": 13560
    },
    {
      "epoch": 1.0908360128617363,
      "grad_norm": 1.3612911701202393,
      "learning_rate": 4.5665859564164655e-05,
      "loss": 1.0198,
      "step": 13570
    },
    {
      "epoch": 1.0916398713826367,
      "grad_norm": 1.8947391510009766,
      "learning_rate": 4.5625504439063767e-05,
      "loss": 0.8653,
      "step": 13580
    },
    {
      "epoch": 1.092443729903537,
      "grad_norm": 1.4824692010879517,
      "learning_rate": 4.558514931396288e-05,
      "loss": 1.1218,
      "step": 13590
    },
    {
      "epoch": 1.0932475884244373,
      "grad_norm": 1.7231032848358154,
      "learning_rate": 4.554479418886199e-05,
      "loss": 1.0036,
      "step": 13600
    },
    {
      "epoch": 1.0940514469453375,
      "grad_norm": 1.3887447118759155,
      "learning_rate": 4.55044390637611e-05,
      "loss": 1.0402,
      "step": 13610
    },
    {
      "epoch": 1.094855305466238,
      "grad_norm": 1.8225650787353516,
      "learning_rate": 4.546408393866021e-05,
      "loss": 0.8615,
      "step": 13620
    },
    {
      "epoch": 1.0956591639871383,
      "grad_norm": 2.701054573059082,
      "learning_rate": 4.542372881355932e-05,
      "loss": 0.9572,
      "step": 13630
    },
    {
      "epoch": 1.0964630225080385,
      "grad_norm": 1.414825201034546,
      "learning_rate": 4.5383373688458435e-05,
      "loss": 1.0855,
      "step": 13640
    },
    {
      "epoch": 1.097266881028939,
      "grad_norm": 2.057631731033325,
      "learning_rate": 4.5343018563357546e-05,
      "loss": 1.0737,
      "step": 13650
    },
    {
      "epoch": 1.0980707395498392,
      "grad_norm": 2.5682103633880615,
      "learning_rate": 4.530266343825666e-05,
      "loss": 0.9428,
      "step": 13660
    },
    {
      "epoch": 1.0988745980707395,
      "grad_norm": 1.64361572265625,
      "learning_rate": 4.526230831315577e-05,
      "loss": 1.1099,
      "step": 13670
    },
    {
      "epoch": 1.09967845659164,
      "grad_norm": 1.4092881679534912,
      "learning_rate": 4.522195318805488e-05,
      "loss": 1.0326,
      "step": 13680
    },
    {
      "epoch": 1.1004823151125402,
      "grad_norm": 1.314833641052246,
      "learning_rate": 4.518159806295399e-05,
      "loss": 0.9389,
      "step": 13690
    },
    {
      "epoch": 1.1012861736334405,
      "grad_norm": 1.7281252145767212,
      "learning_rate": 4.514124293785311e-05,
      "loss": 1.096,
      "step": 13700
    },
    {
      "epoch": 1.1020900321543408,
      "grad_norm": 1.3276362419128418,
      "learning_rate": 4.510088781275222e-05,
      "loss": 0.9756,
      "step": 13710
    },
    {
      "epoch": 1.1028938906752412,
      "grad_norm": 1.3409945964813232,
      "learning_rate": 4.506053268765133e-05,
      "loss": 1.0168,
      "step": 13720
    },
    {
      "epoch": 1.1036977491961415,
      "grad_norm": 1.5758594274520874,
      "learning_rate": 4.502017756255045e-05,
      "loss": 0.9559,
      "step": 13730
    },
    {
      "epoch": 1.1045016077170418,
      "grad_norm": 1.2208436727523804,
      "learning_rate": 4.497982243744956e-05,
      "loss": 0.9951,
      "step": 13740
    },
    {
      "epoch": 1.105305466237942,
      "grad_norm": 2.6655726432800293,
      "learning_rate": 4.493946731234867e-05,
      "loss": 1.0113,
      "step": 13750
    },
    {
      "epoch": 1.1061093247588425,
      "grad_norm": 2.250619888305664,
      "learning_rate": 4.4899112187247785e-05,
      "loss": 0.8718,
      "step": 13760
    },
    {
      "epoch": 1.1069131832797428,
      "grad_norm": 1.9622561931610107,
      "learning_rate": 4.4858757062146896e-05,
      "loss": 1.0055,
      "step": 13770
    },
    {
      "epoch": 1.107717041800643,
      "grad_norm": 1.6765295267105103,
      "learning_rate": 4.48224374495561e-05,
      "loss": 1.0636,
      "step": 13780
    },
    {
      "epoch": 1.1085209003215435,
      "grad_norm": 1.735635757446289,
      "learning_rate": 4.478208232445521e-05,
      "loss": 0.9623,
      "step": 13790
    },
    {
      "epoch": 1.1093247588424437,
      "grad_norm": 1.2861146926879883,
      "learning_rate": 4.474172719935432e-05,
      "loss": 1.0217,
      "step": 13800
    },
    {
      "epoch": 1.110128617363344,
      "grad_norm": 1.6345241069793701,
      "learning_rate": 4.470137207425343e-05,
      "loss": 1.0118,
      "step": 13810
    },
    {
      "epoch": 1.1109324758842445,
      "grad_norm": 1.7163978815078735,
      "learning_rate": 4.466101694915254e-05,
      "loss": 1.0795,
      "step": 13820
    },
    {
      "epoch": 1.1117363344051447,
      "grad_norm": 1.987023115158081,
      "learning_rate": 4.462066182405166e-05,
      "loss": 0.9164,
      "step": 13830
    },
    {
      "epoch": 1.112540192926045,
      "grad_norm": 1.6738214492797852,
      "learning_rate": 4.458030669895077e-05,
      "loss": 0.9628,
      "step": 13840
    },
    {
      "epoch": 1.1133440514469453,
      "grad_norm": 1.9623785018920898,
      "learning_rate": 4.453995157384988e-05,
      "loss": 1.0293,
      "step": 13850
    },
    {
      "epoch": 1.1141479099678457,
      "grad_norm": 2.1075973510742188,
      "learning_rate": 4.4499596448748995e-05,
      "loss": 1.0122,
      "step": 13860
    },
    {
      "epoch": 1.114951768488746,
      "grad_norm": 1.572359323501587,
      "learning_rate": 4.4459241323648106e-05,
      "loss": 1.1115,
      "step": 13870
    },
    {
      "epoch": 1.1157556270096463,
      "grad_norm": 1.8940703868865967,
      "learning_rate": 4.441888619854722e-05,
      "loss": 1.0937,
      "step": 13880
    },
    {
      "epoch": 1.1165594855305465,
      "grad_norm": 1.657184362411499,
      "learning_rate": 4.437853107344633e-05,
      "loss": 0.8793,
      "step": 13890
    },
    {
      "epoch": 1.117363344051447,
      "grad_norm": 1.267751693725586,
      "learning_rate": 4.433817594834544e-05,
      "loss": 0.9959,
      "step": 13900
    },
    {
      "epoch": 1.1181672025723473,
      "grad_norm": 2.4804649353027344,
      "learning_rate": 4.429782082324456e-05,
      "loss": 0.9934,
      "step": 13910
    },
    {
      "epoch": 1.1189710610932475,
      "grad_norm": 1.7541329860687256,
      "learning_rate": 4.425746569814367e-05,
      "loss": 1.0165,
      "step": 13920
    },
    {
      "epoch": 1.119774919614148,
      "grad_norm": 1.3630015850067139,
      "learning_rate": 4.421711057304278e-05,
      "loss": 1.098,
      "step": 13930
    },
    {
      "epoch": 1.1205787781350482,
      "grad_norm": 1.548208236694336,
      "learning_rate": 4.417675544794189e-05,
      "loss": 1.1485,
      "step": 13940
    },
    {
      "epoch": 1.1213826366559485,
      "grad_norm": 2.2378597259521484,
      "learning_rate": 4.4136400322841004e-05,
      "loss": 0.9316,
      "step": 13950
    },
    {
      "epoch": 1.122186495176849,
      "grad_norm": 1.7924768924713135,
      "learning_rate": 4.4096045197740115e-05,
      "loss": 0.9349,
      "step": 13960
    },
    {
      "epoch": 1.1229903536977492,
      "grad_norm": 1.8515634536743164,
      "learning_rate": 4.4055690072639226e-05,
      "loss": 1.0372,
      "step": 13970
    },
    {
      "epoch": 1.1237942122186495,
      "grad_norm": 1.59554922580719,
      "learning_rate": 4.401533494753834e-05,
      "loss": 0.951,
      "step": 13980
    },
    {
      "epoch": 1.1245980707395498,
      "grad_norm": 1.879597783088684,
      "learning_rate": 4.397497982243745e-05,
      "loss": 1.0449,
      "step": 13990
    },
    {
      "epoch": 1.1254019292604502,
      "grad_norm": 2.153693199157715,
      "learning_rate": 4.393462469733656e-05,
      "loss": 1.0486,
      "step": 14000
    },
    {
      "epoch": 1.1254019292604502,
      "eval_yahma/alpaca-cleaned_loss": 1.2180111408233643,
      "eval_yahma/alpaca-cleaned_runtime": 115.6253,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.297,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.162,
      "step": 14000
    },
    {
      "epoch": 1.1262057877813505,
      "grad_norm": 1.7476816177368164,
      "learning_rate": 4.389426957223567e-05,
      "loss": 1.0463,
      "step": 14010
    },
    {
      "epoch": 1.1270096463022508,
      "grad_norm": 1.2039158344268799,
      "learning_rate": 4.385391444713478e-05,
      "loss": 0.8675,
      "step": 14020
    },
    {
      "epoch": 1.127813504823151,
      "grad_norm": 1.786009669303894,
      "learning_rate": 4.38135593220339e-05,
      "loss": 0.9991,
      "step": 14030
    },
    {
      "epoch": 1.1286173633440515,
      "grad_norm": 1.2885710000991821,
      "learning_rate": 4.377320419693301e-05,
      "loss": 1.1352,
      "step": 14040
    },
    {
      "epoch": 1.1294212218649518,
      "grad_norm": 2.1474835872650146,
      "learning_rate": 4.3732849071832124e-05,
      "loss": 0.9888,
      "step": 14050
    },
    {
      "epoch": 1.130225080385852,
      "grad_norm": 1.4155480861663818,
      "learning_rate": 4.369249394673124e-05,
      "loss": 1.0412,
      "step": 14060
    },
    {
      "epoch": 1.1310289389067525,
      "grad_norm": 1.7451374530792236,
      "learning_rate": 4.3652138821630354e-05,
      "loss": 1.0104,
      "step": 14070
    },
    {
      "epoch": 1.1318327974276527,
      "grad_norm": 2.5622613430023193,
      "learning_rate": 4.3611783696529465e-05,
      "loss": 1.0273,
      "step": 14080
    },
    {
      "epoch": 1.132636655948553,
      "grad_norm": 1.6344316005706787,
      "learning_rate": 4.3571428571428576e-05,
      "loss": 0.9315,
      "step": 14090
    },
    {
      "epoch": 1.1334405144694535,
      "grad_norm": 1.552697777748108,
      "learning_rate": 4.353107344632769e-05,
      "loss": 1.033,
      "step": 14100
    },
    {
      "epoch": 1.1342443729903537,
      "grad_norm": 2.1421780586242676,
      "learning_rate": 4.34907183212268e-05,
      "loss": 1.0476,
      "step": 14110
    },
    {
      "epoch": 1.135048231511254,
      "grad_norm": 1.256295084953308,
      "learning_rate": 4.345036319612591e-05,
      "loss": 0.973,
      "step": 14120
    },
    {
      "epoch": 1.1358520900321543,
      "grad_norm": 1.2301644086837769,
      "learning_rate": 4.341000807102502e-05,
      "loss": 1.0045,
      "step": 14130
    },
    {
      "epoch": 1.1366559485530547,
      "grad_norm": 2.543393135070801,
      "learning_rate": 4.336965294592413e-05,
      "loss": 0.9381,
      "step": 14140
    },
    {
      "epoch": 1.137459807073955,
      "grad_norm": 1.4487563371658325,
      "learning_rate": 4.3329297820823245e-05,
      "loss": 1.0133,
      "step": 14150
    },
    {
      "epoch": 1.1382636655948553,
      "grad_norm": 3.257087230682373,
      "learning_rate": 4.3288942695722356e-05,
      "loss": 1.0521,
      "step": 14160
    },
    {
      "epoch": 1.1390675241157555,
      "grad_norm": 3.0251243114471436,
      "learning_rate": 4.324858757062147e-05,
      "loss": 0.8705,
      "step": 14170
    },
    {
      "epoch": 1.139871382636656,
      "grad_norm": 1.651095986366272,
      "learning_rate": 4.3208232445520585e-05,
      "loss": 1.0427,
      "step": 14180
    },
    {
      "epoch": 1.1406752411575563,
      "grad_norm": 1.6251319646835327,
      "learning_rate": 4.31678773204197e-05,
      "loss": 1.0401,
      "step": 14190
    },
    {
      "epoch": 1.1414790996784565,
      "grad_norm": 2.8393075466156006,
      "learning_rate": 4.312752219531881e-05,
      "loss": 0.929,
      "step": 14200
    },
    {
      "epoch": 1.142282958199357,
      "grad_norm": 1.2079108953475952,
      "learning_rate": 4.308716707021792e-05,
      "loss": 1.0199,
      "step": 14210
    },
    {
      "epoch": 1.1430868167202572,
      "grad_norm": 1.8249192237854004,
      "learning_rate": 4.304681194511703e-05,
      "loss": 1.111,
      "step": 14220
    },
    {
      "epoch": 1.1438906752411575,
      "grad_norm": 1.9499272108078003,
      "learning_rate": 4.300645682001614e-05,
      "loss": 0.9521,
      "step": 14230
    },
    {
      "epoch": 1.144694533762058,
      "grad_norm": 1.579451084136963,
      "learning_rate": 4.2966101694915254e-05,
      "loss": 0.9798,
      "step": 14240
    },
    {
      "epoch": 1.1454983922829582,
      "grad_norm": 1.9789243936538696,
      "learning_rate": 4.2925746569814365e-05,
      "loss": 1.0356,
      "step": 14250
    },
    {
      "epoch": 1.1463022508038585,
      "grad_norm": 1.6762830018997192,
      "learning_rate": 4.288539144471348e-05,
      "loss": 0.9138,
      "step": 14260
    },
    {
      "epoch": 1.1471061093247588,
      "grad_norm": 1.5332313776016235,
      "learning_rate": 4.2845036319612594e-05,
      "loss": 0.9959,
      "step": 14270
    },
    {
      "epoch": 1.1479099678456592,
      "grad_norm": 2.1301565170288086,
      "learning_rate": 4.2804681194511706e-05,
      "loss": 0.9985,
      "step": 14280
    },
    {
      "epoch": 1.1487138263665595,
      "grad_norm": 1.6805988550186157,
      "learning_rate": 4.276432606941082e-05,
      "loss": 0.9996,
      "step": 14290
    },
    {
      "epoch": 1.1495176848874598,
      "grad_norm": 1.5411735773086548,
      "learning_rate": 4.272397094430993e-05,
      "loss": 0.9634,
      "step": 14300
    },
    {
      "epoch": 1.15032154340836,
      "grad_norm": 1.270493984222412,
      "learning_rate": 4.268361581920905e-05,
      "loss": 1.0496,
      "step": 14310
    },
    {
      "epoch": 1.1511254019292605,
      "grad_norm": 4.723905086517334,
      "learning_rate": 4.264326069410816e-05,
      "loss": 1.0178,
      "step": 14320
    },
    {
      "epoch": 1.1519292604501608,
      "grad_norm": 1.8460774421691895,
      "learning_rate": 4.260290556900727e-05,
      "loss": 0.8411,
      "step": 14330
    },
    {
      "epoch": 1.152733118971061,
      "grad_norm": 1.428191900253296,
      "learning_rate": 4.256255044390638e-05,
      "loss": 0.9971,
      "step": 14340
    },
    {
      "epoch": 1.1535369774919615,
      "grad_norm": 2.1410958766937256,
      "learning_rate": 4.252219531880549e-05,
      "loss": 1.0238,
      "step": 14350
    },
    {
      "epoch": 1.1543408360128617,
      "grad_norm": 2.1243741512298584,
      "learning_rate": 4.2481840193704604e-05,
      "loss": 1.0217,
      "step": 14360
    },
    {
      "epoch": 1.155144694533762,
      "grad_norm": 1.4340726137161255,
      "learning_rate": 4.2441485068603715e-05,
      "loss": 1.0448,
      "step": 14370
    },
    {
      "epoch": 1.1559485530546625,
      "grad_norm": 1.64406418800354,
      "learning_rate": 4.2401129943502826e-05,
      "loss": 1.095,
      "step": 14380
    },
    {
      "epoch": 1.1567524115755627,
      "grad_norm": 1.3270835876464844,
      "learning_rate": 4.236077481840194e-05,
      "loss": 1.0975,
      "step": 14390
    },
    {
      "epoch": 1.157556270096463,
      "grad_norm": 1.5200716257095337,
      "learning_rate": 4.232041969330105e-05,
      "loss": 1.1141,
      "step": 14400
    },
    {
      "epoch": 1.1583601286173633,
      "grad_norm": 1.362056016921997,
      "learning_rate": 4.228006456820016e-05,
      "loss": 1.0216,
      "step": 14410
    },
    {
      "epoch": 1.1591639871382637,
      "grad_norm": 1.8959547281265259,
      "learning_rate": 4.223970944309927e-05,
      "loss": 0.9394,
      "step": 14420
    },
    {
      "epoch": 1.159967845659164,
      "grad_norm": 1.3526105880737305,
      "learning_rate": 4.219935431799838e-05,
      "loss": 1.0009,
      "step": 14430
    },
    {
      "epoch": 1.1607717041800643,
      "grad_norm": 1.8486071825027466,
      "learning_rate": 4.21589991928975e-05,
      "loss": 1.051,
      "step": 14440
    },
    {
      "epoch": 1.1615755627009645,
      "grad_norm": 1.899803638458252,
      "learning_rate": 4.211864406779661e-05,
      "loss": 0.9083,
      "step": 14450
    },
    {
      "epoch": 1.162379421221865,
      "grad_norm": 1.7194709777832031,
      "learning_rate": 4.2078288942695724e-05,
      "loss": 1.0118,
      "step": 14460
    },
    {
      "epoch": 1.1631832797427653,
      "grad_norm": 1.7135143280029297,
      "learning_rate": 4.2037933817594835e-05,
      "loss": 1.0856,
      "step": 14470
    },
    {
      "epoch": 1.1639871382636655,
      "grad_norm": 1.6633415222167969,
      "learning_rate": 4.199757869249395e-05,
      "loss": 0.9976,
      "step": 14480
    },
    {
      "epoch": 1.164790996784566,
      "grad_norm": 1.6620136499404907,
      "learning_rate": 4.195722356739306e-05,
      "loss": 1.1468,
      "step": 14490
    },
    {
      "epoch": 1.1655948553054662,
      "grad_norm": 1.4163603782653809,
      "learning_rate": 4.1916868442292176e-05,
      "loss": 1.0357,
      "step": 14500
    },
    {
      "epoch": 1.1663987138263665,
      "grad_norm": 1.6289397478103638,
      "learning_rate": 4.187651331719129e-05,
      "loss": 1.0538,
      "step": 14510
    },
    {
      "epoch": 1.167202572347267,
      "grad_norm": 1.8465838432312012,
      "learning_rate": 4.18361581920904e-05,
      "loss": 0.9653,
      "step": 14520
    },
    {
      "epoch": 1.1680064308681672,
      "grad_norm": 1.3394323587417603,
      "learning_rate": 4.179580306698951e-05,
      "loss": 1.0897,
      "step": 14530
    },
    {
      "epoch": 1.1688102893890675,
      "grad_norm": 1.341867208480835,
      "learning_rate": 4.175544794188862e-05,
      "loss": 1.028,
      "step": 14540
    },
    {
      "epoch": 1.1696141479099678,
      "grad_norm": 1.6926863193511963,
      "learning_rate": 4.171509281678773e-05,
      "loss": 1.0484,
      "step": 14550
    },
    {
      "epoch": 1.1704180064308682,
      "grad_norm": 1.8348326683044434,
      "learning_rate": 4.1674737691686844e-05,
      "loss": 1.0871,
      "step": 14560
    },
    {
      "epoch": 1.1712218649517685,
      "grad_norm": 1.9606865644454956,
      "learning_rate": 4.163438256658596e-05,
      "loss": 1.0358,
      "step": 14570
    },
    {
      "epoch": 1.1720257234726688,
      "grad_norm": 1.9318088293075562,
      "learning_rate": 4.1594027441485074e-05,
      "loss": 0.9973,
      "step": 14580
    },
    {
      "epoch": 1.172829581993569,
      "grad_norm": 1.6759283542633057,
      "learning_rate": 4.1553672316384185e-05,
      "loss": 1.0548,
      "step": 14590
    },
    {
      "epoch": 1.1736334405144695,
      "grad_norm": 2.018908977508545,
      "learning_rate": 4.1513317191283297e-05,
      "loss": 1.0622,
      "step": 14600
    },
    {
      "epoch": 1.1744372990353698,
      "grad_norm": 2.886817455291748,
      "learning_rate": 4.147296206618241e-05,
      "loss": 0.9834,
      "step": 14610
    },
    {
      "epoch": 1.17524115755627,
      "grad_norm": 2.012728452682495,
      "learning_rate": 4.143260694108152e-05,
      "loss": 0.9802,
      "step": 14620
    },
    {
      "epoch": 1.1760450160771705,
      "grad_norm": 2.133406400680542,
      "learning_rate": 4.139225181598063e-05,
      "loss": 1.074,
      "step": 14630
    },
    {
      "epoch": 1.1768488745980707,
      "grad_norm": 2.4336249828338623,
      "learning_rate": 4.135189669087974e-05,
      "loss": 0.9832,
      "step": 14640
    },
    {
      "epoch": 1.177652733118971,
      "grad_norm": 1.6935405731201172,
      "learning_rate": 4.1311541565778853e-05,
      "loss": 1.012,
      "step": 14650
    },
    {
      "epoch": 1.1784565916398715,
      "grad_norm": 1.684714674949646,
      "learning_rate": 4.1271186440677965e-05,
      "loss": 1.0092,
      "step": 14660
    },
    {
      "epoch": 1.1792604501607717,
      "grad_norm": 1.4697527885437012,
      "learning_rate": 4.1230831315577076e-05,
      "loss": 1.0159,
      "step": 14670
    },
    {
      "epoch": 1.180064308681672,
      "grad_norm": 1.7050909996032715,
      "learning_rate": 4.119047619047619e-05,
      "loss": 1.0019,
      "step": 14680
    },
    {
      "epoch": 1.1808681672025723,
      "grad_norm": 1.9618253707885742,
      "learning_rate": 4.11501210653753e-05,
      "loss": 0.9164,
      "step": 14690
    },
    {
      "epoch": 1.1816720257234727,
      "grad_norm": 1.96260666847229,
      "learning_rate": 4.110976594027442e-05,
      "loss": 0.9642,
      "step": 14700
    },
    {
      "epoch": 1.182475884244373,
      "grad_norm": 1.8400181531906128,
      "learning_rate": 4.106941081517353e-05,
      "loss": 0.9593,
      "step": 14710
    },
    {
      "epoch": 1.1832797427652733,
      "grad_norm": 3.5856144428253174,
      "learning_rate": 4.102905569007264e-05,
      "loss": 1.3208,
      "step": 14720
    },
    {
      "epoch": 1.1840836012861735,
      "grad_norm": 1.7220356464385986,
      "learning_rate": 4.098870056497176e-05,
      "loss": 1.0172,
      "step": 14730
    },
    {
      "epoch": 1.184887459807074,
      "grad_norm": 1.6482996940612793,
      "learning_rate": 4.094834543987087e-05,
      "loss": 0.9748,
      "step": 14740
    },
    {
      "epoch": 1.1856913183279743,
      "grad_norm": 1.687031865119934,
      "learning_rate": 4.090799031476998e-05,
      "loss": 1.0431,
      "step": 14750
    },
    {
      "epoch": 1.1864951768488745,
      "grad_norm": 1.7135734558105469,
      "learning_rate": 4.086763518966909e-05,
      "loss": 1.0557,
      "step": 14760
    },
    {
      "epoch": 1.187299035369775,
      "grad_norm": 1.9167877435684204,
      "learning_rate": 4.08272800645682e-05,
      "loss": 1.0152,
      "step": 14770
    },
    {
      "epoch": 1.1881028938906752,
      "grad_norm": 1.734106421470642,
      "learning_rate": 4.0786924939467315e-05,
      "loss": 1.0033,
      "step": 14780
    },
    {
      "epoch": 1.1889067524115755,
      "grad_norm": 1.9168775081634521,
      "learning_rate": 4.0746569814366426e-05,
      "loss": 0.9823,
      "step": 14790
    },
    {
      "epoch": 1.189710610932476,
      "grad_norm": 3.9640393257141113,
      "learning_rate": 4.070621468926554e-05,
      "loss": 1.0677,
      "step": 14800
    },
    {
      "epoch": 1.1905144694533762,
      "grad_norm": 1.4360038042068481,
      "learning_rate": 4.066585956416465e-05,
      "loss": 1.0105,
      "step": 14810
    },
    {
      "epoch": 1.1913183279742765,
      "grad_norm": 4.0904541015625,
      "learning_rate": 4.062550443906376e-05,
      "loss": 0.9963,
      "step": 14820
    },
    {
      "epoch": 1.1921221864951768,
      "grad_norm": 1.6084517240524292,
      "learning_rate": 4.058514931396287e-05,
      "loss": 0.9998,
      "step": 14830
    },
    {
      "epoch": 1.1929260450160772,
      "grad_norm": 1.730948805809021,
      "learning_rate": 4.054479418886199e-05,
      "loss": 0.9964,
      "step": 14840
    },
    {
      "epoch": 1.1937299035369775,
      "grad_norm": 1.4679205417633057,
      "learning_rate": 4.05044390637611e-05,
      "loss": 1.1229,
      "step": 14850
    },
    {
      "epoch": 1.1945337620578778,
      "grad_norm": 2.152657985687256,
      "learning_rate": 4.046408393866021e-05,
      "loss": 1.0017,
      "step": 14860
    },
    {
      "epoch": 1.195337620578778,
      "grad_norm": 1.3565618991851807,
      "learning_rate": 4.0423728813559324e-05,
      "loss": 1.0653,
      "step": 14870
    },
    {
      "epoch": 1.1961414790996785,
      "grad_norm": 1.5673973560333252,
      "learning_rate": 4.0383373688458435e-05,
      "loss": 0.8449,
      "step": 14880
    },
    {
      "epoch": 1.1969453376205788,
      "grad_norm": 4.41684627532959,
      "learning_rate": 4.0343018563357547e-05,
      "loss": 1.0748,
      "step": 14890
    },
    {
      "epoch": 1.197749196141479,
      "grad_norm": 1.369780421257019,
      "learning_rate": 4.030266343825666e-05,
      "loss": 0.9427,
      "step": 14900
    },
    {
      "epoch": 1.1985530546623795,
      "grad_norm": 1.9163693189620972,
      "learning_rate": 4.026230831315577e-05,
      "loss": 1.0125,
      "step": 14910
    },
    {
      "epoch": 1.1993569131832797,
      "grad_norm": 2.2586166858673096,
      "learning_rate": 4.022195318805488e-05,
      "loss": 1.0345,
      "step": 14920
    },
    {
      "epoch": 1.20016077170418,
      "grad_norm": 1.5923124551773071,
      "learning_rate": 4.0181598062954e-05,
      "loss": 1.0082,
      "step": 14930
    },
    {
      "epoch": 1.2009646302250805,
      "grad_norm": 1.4561861753463745,
      "learning_rate": 4.014124293785311e-05,
      "loss": 0.939,
      "step": 14940
    },
    {
      "epoch": 1.2017684887459807,
      "grad_norm": 1.5259567499160767,
      "learning_rate": 4.010088781275222e-05,
      "loss": 0.9689,
      "step": 14950
    },
    {
      "epoch": 1.202572347266881,
      "grad_norm": 1.632391333580017,
      "learning_rate": 4.006053268765133e-05,
      "loss": 0.9775,
      "step": 14960
    },
    {
      "epoch": 1.2033762057877813,
      "grad_norm": 1.7772167921066284,
      "learning_rate": 4.002017756255045e-05,
      "loss": 1.1127,
      "step": 14970
    },
    {
      "epoch": 1.2041800643086817,
      "grad_norm": 1.8630040884017944,
      "learning_rate": 3.997982243744956e-05,
      "loss": 0.9544,
      "step": 14980
    },
    {
      "epoch": 1.204983922829582,
      "grad_norm": 3.0156354904174805,
      "learning_rate": 3.9939467312348674e-05,
      "loss": 1.1024,
      "step": 14990
    },
    {
      "epoch": 1.2057877813504823,
      "grad_norm": 1.5983487367630005,
      "learning_rate": 3.9899112187247785e-05,
      "loss": 0.9616,
      "step": 15000
    },
    {
      "epoch": 1.2065916398713825,
      "grad_norm": 1.3613848686218262,
      "learning_rate": 3.9858757062146896e-05,
      "loss": 1.0519,
      "step": 15010
    },
    {
      "epoch": 1.207395498392283,
      "grad_norm": 2.2423689365386963,
      "learning_rate": 3.981840193704601e-05,
      "loss": 1.053,
      "step": 15020
    },
    {
      "epoch": 1.2081993569131833,
      "grad_norm": 2.0904273986816406,
      "learning_rate": 3.977804681194512e-05,
      "loss": 0.9744,
      "step": 15030
    },
    {
      "epoch": 1.2090032154340835,
      "grad_norm": 1.6902353763580322,
      "learning_rate": 3.973769168684423e-05,
      "loss": 1.0055,
      "step": 15040
    },
    {
      "epoch": 1.209807073954984,
      "grad_norm": 1.4305819272994995,
      "learning_rate": 3.969733656174334e-05,
      "loss": 1.0954,
      "step": 15050
    },
    {
      "epoch": 1.2106109324758842,
      "grad_norm": 1.6137382984161377,
      "learning_rate": 3.965698143664245e-05,
      "loss": 0.8971,
      "step": 15060
    },
    {
      "epoch": 1.2114147909967845,
      "grad_norm": 1.9220008850097656,
      "learning_rate": 3.9616626311541565e-05,
      "loss": 0.9589,
      "step": 15070
    },
    {
      "epoch": 1.212218649517685,
      "grad_norm": 2.8765358924865723,
      "learning_rate": 3.9576271186440676e-05,
      "loss": 1.061,
      "step": 15080
    },
    {
      "epoch": 1.2130225080385852,
      "grad_norm": 2.1828126907348633,
      "learning_rate": 3.953591606133979e-05,
      "loss": 1.001,
      "step": 15090
    },
    {
      "epoch": 1.2138263665594855,
      "grad_norm": 1.8121625185012817,
      "learning_rate": 3.9495560936238905e-05,
      "loss": 1.0547,
      "step": 15100
    },
    {
      "epoch": 1.2146302250803858,
      "grad_norm": 1.592613697052002,
      "learning_rate": 3.945520581113802e-05,
      "loss": 0.9974,
      "step": 15110
    },
    {
      "epoch": 1.2154340836012862,
      "grad_norm": 2.5363457202911377,
      "learning_rate": 3.941485068603713e-05,
      "loss": 1.1056,
      "step": 15120
    },
    {
      "epoch": 1.2162379421221865,
      "grad_norm": 1.8469159603118896,
      "learning_rate": 3.937449556093624e-05,
      "loss": 0.9196,
      "step": 15130
    },
    {
      "epoch": 1.2170418006430868,
      "grad_norm": 1.3776907920837402,
      "learning_rate": 3.933414043583535e-05,
      "loss": 1.0031,
      "step": 15140
    },
    {
      "epoch": 1.217845659163987,
      "grad_norm": 2.2064387798309326,
      "learning_rate": 3.929378531073446e-05,
      "loss": 0.9778,
      "step": 15150
    },
    {
      "epoch": 1.2186495176848875,
      "grad_norm": 1.5302129983901978,
      "learning_rate": 3.925343018563358e-05,
      "loss": 1.1049,
      "step": 15160
    },
    {
      "epoch": 1.2194533762057878,
      "grad_norm": 1.896903395652771,
      "learning_rate": 3.921307506053269e-05,
      "loss": 0.9388,
      "step": 15170
    },
    {
      "epoch": 1.220257234726688,
      "grad_norm": 1.9947621822357178,
      "learning_rate": 3.91727199354318e-05,
      "loss": 0.9813,
      "step": 15180
    },
    {
      "epoch": 1.2210610932475885,
      "grad_norm": 2.0882503986358643,
      "learning_rate": 3.9132364810330915e-05,
      "loss": 1.0022,
      "step": 15190
    },
    {
      "epoch": 1.2218649517684887,
      "grad_norm": 1.3755741119384766,
      "learning_rate": 3.9092009685230026e-05,
      "loss": 0.9404,
      "step": 15200
    },
    {
      "epoch": 1.222668810289389,
      "grad_norm": 1.8343044519424438,
      "learning_rate": 3.905165456012914e-05,
      "loss": 1.1459,
      "step": 15210
    },
    {
      "epoch": 1.2234726688102895,
      "grad_norm": 2.009847402572632,
      "learning_rate": 3.901129943502825e-05,
      "loss": 0.9887,
      "step": 15220
    },
    {
      "epoch": 1.2242765273311897,
      "grad_norm": 1.5839662551879883,
      "learning_rate": 3.897094430992737e-05,
      "loss": 0.9295,
      "step": 15230
    },
    {
      "epoch": 1.22508038585209,
      "grad_norm": 1.9991822242736816,
      "learning_rate": 3.893058918482648e-05,
      "loss": 0.9761,
      "step": 15240
    },
    {
      "epoch": 1.2258842443729903,
      "grad_norm": 1.4759560823440552,
      "learning_rate": 3.889023405972559e-05,
      "loss": 1.024,
      "step": 15250
    },
    {
      "epoch": 1.2266881028938907,
      "grad_norm": 2.5121824741363525,
      "learning_rate": 3.88498789346247e-05,
      "loss": 0.9983,
      "step": 15260
    },
    {
      "epoch": 1.227491961414791,
      "grad_norm": 1.6680318117141724,
      "learning_rate": 3.880952380952381e-05,
      "loss": 0.9924,
      "step": 15270
    },
    {
      "epoch": 1.2282958199356913,
      "grad_norm": 1.9648467302322388,
      "learning_rate": 3.8769168684422924e-05,
      "loss": 1.0334,
      "step": 15280
    },
    {
      "epoch": 1.2290996784565915,
      "grad_norm": 1.6655272245407104,
      "learning_rate": 3.8728813559322035e-05,
      "loss": 0.8878,
      "step": 15290
    },
    {
      "epoch": 1.229903536977492,
      "grad_norm": 1.5394530296325684,
      "learning_rate": 3.8688458434221146e-05,
      "loss": 1.0047,
      "step": 15300
    },
    {
      "epoch": 1.2307073954983923,
      "grad_norm": 1.3396995067596436,
      "learning_rate": 3.864810330912026e-05,
      "loss": 0.9884,
      "step": 15310
    },
    {
      "epoch": 1.2315112540192925,
      "grad_norm": 2.521289110183716,
      "learning_rate": 3.860774818401937e-05,
      "loss": 0.9421,
      "step": 15320
    },
    {
      "epoch": 1.232315112540193,
      "grad_norm": 1.703194260597229,
      "learning_rate": 3.856739305891848e-05,
      "loss": 1.0233,
      "step": 15330
    },
    {
      "epoch": 1.2331189710610932,
      "grad_norm": 1.4893176555633545,
      "learning_rate": 3.852703793381759e-05,
      "loss": 0.9465,
      "step": 15340
    },
    {
      "epoch": 1.2339228295819935,
      "grad_norm": 1.442811369895935,
      "learning_rate": 3.84866828087167e-05,
      "loss": 0.9332,
      "step": 15350
    },
    {
      "epoch": 1.234726688102894,
      "grad_norm": 1.6394675970077515,
      "learning_rate": 3.844632768361582e-05,
      "loss": 0.9862,
      "step": 15360
    },
    {
      "epoch": 1.2355305466237942,
      "grad_norm": 1.969940423965454,
      "learning_rate": 3.840597255851493e-05,
      "loss": 1.0272,
      "step": 15370
    },
    {
      "epoch": 1.2363344051446945,
      "grad_norm": 2.1524131298065186,
      "learning_rate": 3.8365617433414044e-05,
      "loss": 1.0043,
      "step": 15380
    },
    {
      "epoch": 1.2371382636655948,
      "grad_norm": 1.8605786561965942,
      "learning_rate": 3.832526230831316e-05,
      "loss": 1.0146,
      "step": 15390
    },
    {
      "epoch": 1.2379421221864952,
      "grad_norm": 1.7119444608688354,
      "learning_rate": 3.8284907183212274e-05,
      "loss": 1.0073,
      "step": 15400
    },
    {
      "epoch": 1.2387459807073955,
      "grad_norm": 1.8046406507492065,
      "learning_rate": 3.8244552058111385e-05,
      "loss": 1.0533,
      "step": 15410
    },
    {
      "epoch": 1.2395498392282958,
      "grad_norm": 1.8285448551177979,
      "learning_rate": 3.8204196933010496e-05,
      "loss": 1.0184,
      "step": 15420
    },
    {
      "epoch": 1.2403536977491962,
      "grad_norm": 1.8175604343414307,
      "learning_rate": 3.816384180790961e-05,
      "loss": 0.9599,
      "step": 15430
    },
    {
      "epoch": 1.2411575562700965,
      "grad_norm": 2.057853937149048,
      "learning_rate": 3.812348668280872e-05,
      "loss": 1.013,
      "step": 15440
    },
    {
      "epoch": 1.2419614147909968,
      "grad_norm": 1.6940542459487915,
      "learning_rate": 3.808313155770783e-05,
      "loss": 0.9813,
      "step": 15450
    },
    {
      "epoch": 1.242765273311897,
      "grad_norm": 1.6630384922027588,
      "learning_rate": 3.804277643260694e-05,
      "loss": 1.0528,
      "step": 15460
    },
    {
      "epoch": 1.2435691318327975,
      "grad_norm": 1.4436936378479004,
      "learning_rate": 3.800242130750605e-05,
      "loss": 0.9409,
      "step": 15470
    },
    {
      "epoch": 1.2443729903536977,
      "grad_norm": 1.7026877403259277,
      "learning_rate": 3.7962066182405164e-05,
      "loss": 0.9785,
      "step": 15480
    },
    {
      "epoch": 1.245176848874598,
      "grad_norm": 1.5710142850875854,
      "learning_rate": 3.792171105730428e-05,
      "loss": 0.925,
      "step": 15490
    },
    {
      "epoch": 1.2459807073954985,
      "grad_norm": 1.6203075647354126,
      "learning_rate": 3.7881355932203394e-05,
      "loss": 1.0111,
      "step": 15500
    },
    {
      "epoch": 1.2467845659163987,
      "grad_norm": 3.0252764225006104,
      "learning_rate": 3.7841000807102505e-05,
      "loss": 0.9773,
      "step": 15510
    },
    {
      "epoch": 1.247588424437299,
      "grad_norm": 2.1563704013824463,
      "learning_rate": 3.780064568200162e-05,
      "loss": 0.9742,
      "step": 15520
    },
    {
      "epoch": 1.2483922829581993,
      "grad_norm": 2.3449900150299072,
      "learning_rate": 3.776029055690073e-05,
      "loss": 1.0818,
      "step": 15530
    },
    {
      "epoch": 1.2491961414790997,
      "grad_norm": 2.3925440311431885,
      "learning_rate": 3.771993543179984e-05,
      "loss": 1.0998,
      "step": 15540
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5898447036743164,
      "learning_rate": 3.767958030669895e-05,
      "loss": 0.936,
      "step": 15550
    },
    {
      "epoch": 1.2508038585209003,
      "grad_norm": 2.1979892253875732,
      "learning_rate": 3.763922518159806e-05,
      "loss": 1.1329,
      "step": 15560
    },
    {
      "epoch": 1.2516077170418005,
      "grad_norm": 1.892656683921814,
      "learning_rate": 3.7598870056497174e-05,
      "loss": 1.0623,
      "step": 15570
    },
    {
      "epoch": 1.252411575562701,
      "grad_norm": 1.9178494215011597,
      "learning_rate": 3.7558514931396285e-05,
      "loss": 1.038,
      "step": 15580
    },
    {
      "epoch": 1.2532154340836013,
      "grad_norm": 1.6698342561721802,
      "learning_rate": 3.75181598062954e-05,
      "loss": 0.9981,
      "step": 15590
    },
    {
      "epoch": 1.2540192926045015,
      "grad_norm": 1.7022637128829956,
      "learning_rate": 3.7477804681194514e-05,
      "loss": 0.9657,
      "step": 15600
    },
    {
      "epoch": 1.254823151125402,
      "grad_norm": 1.5800572633743286,
      "learning_rate": 3.7437449556093626e-05,
      "loss": 1.0576,
      "step": 15610
    },
    {
      "epoch": 1.2556270096463023,
      "grad_norm": 2.19415545463562,
      "learning_rate": 3.739709443099274e-05,
      "loss": 1.0083,
      "step": 15620
    },
    {
      "epoch": 1.2564308681672025,
      "grad_norm": 1.5555347204208374,
      "learning_rate": 3.7356739305891855e-05,
      "loss": 1.0514,
      "step": 15630
    },
    {
      "epoch": 1.257234726688103,
      "grad_norm": 1.9707562923431396,
      "learning_rate": 3.731638418079097e-05,
      "loss": 0.9158,
      "step": 15640
    },
    {
      "epoch": 1.2580385852090032,
      "grad_norm": 1.4888017177581787,
      "learning_rate": 3.727602905569008e-05,
      "loss": 0.9496,
      "step": 15650
    },
    {
      "epoch": 1.2588424437299035,
      "grad_norm": 1.655102014541626,
      "learning_rate": 3.723567393058919e-05,
      "loss": 1.0025,
      "step": 15660
    },
    {
      "epoch": 1.259646302250804,
      "grad_norm": 1.51974356174469,
      "learning_rate": 3.71953188054883e-05,
      "loss": 1.0251,
      "step": 15670
    },
    {
      "epoch": 1.2604501607717042,
      "grad_norm": 1.6518058776855469,
      "learning_rate": 3.715496368038741e-05,
      "loss": 0.9934,
      "step": 15680
    },
    {
      "epoch": 1.2612540192926045,
      "grad_norm": 1.6622509956359863,
      "learning_rate": 3.7114608555286523e-05,
      "loss": 1.0582,
      "step": 15690
    },
    {
      "epoch": 1.2620578778135048,
      "grad_norm": 2.096949338912964,
      "learning_rate": 3.7074253430185635e-05,
      "loss": 1.0454,
      "step": 15700
    },
    {
      "epoch": 1.262861736334405,
      "grad_norm": 1.668216586112976,
      "learning_rate": 3.7033898305084746e-05,
      "loss": 1.0954,
      "step": 15710
    },
    {
      "epoch": 1.2636655948553055,
      "grad_norm": 1.8938219547271729,
      "learning_rate": 3.699354317998386e-05,
      "loss": 0.9693,
      "step": 15720
    },
    {
      "epoch": 1.2644694533762058,
      "grad_norm": 1.4523894786834717,
      "learning_rate": 3.695318805488297e-05,
      "loss": 1.068,
      "step": 15730
    },
    {
      "epoch": 1.265273311897106,
      "grad_norm": 1.4719635248184204,
      "learning_rate": 3.691283292978208e-05,
      "loss": 1.051,
      "step": 15740
    },
    {
      "epoch": 1.2660771704180065,
      "grad_norm": 2.059270143508911,
      "learning_rate": 3.687247780468119e-05,
      "loss": 1.0024,
      "step": 15750
    },
    {
      "epoch": 1.2668810289389068,
      "grad_norm": 1.6028730869293213,
      "learning_rate": 3.683212267958031e-05,
      "loss": 0.948,
      "step": 15760
    },
    {
      "epoch": 1.267684887459807,
      "grad_norm": 2.917422294616699,
      "learning_rate": 3.679176755447942e-05,
      "loss": 0.9057,
      "step": 15770
    },
    {
      "epoch": 1.2684887459807075,
      "grad_norm": 1.3262091875076294,
      "learning_rate": 3.675141242937853e-05,
      "loss": 0.9915,
      "step": 15780
    },
    {
      "epoch": 1.2692926045016077,
      "grad_norm": 2.217057466506958,
      "learning_rate": 3.6711057304277644e-05,
      "loss": 0.958,
      "step": 15790
    },
    {
      "epoch": 1.270096463022508,
      "grad_norm": 1.6233608722686768,
      "learning_rate": 3.6670702179176755e-05,
      "loss": 0.8224,
      "step": 15800
    },
    {
      "epoch": 1.2709003215434085,
      "grad_norm": 1.6154686212539673,
      "learning_rate": 3.663034705407587e-05,
      "loss": 1.0314,
      "step": 15810
    },
    {
      "epoch": 1.2717041800643087,
      "grad_norm": 2.7980661392211914,
      "learning_rate": 3.6589991928974985e-05,
      "loss": 1.0876,
      "step": 15820
    },
    {
      "epoch": 1.272508038585209,
      "grad_norm": 1.5344147682189941,
      "learning_rate": 3.6549636803874096e-05,
      "loss": 0.9062,
      "step": 15830
    },
    {
      "epoch": 1.2733118971061093,
      "grad_norm": 3.238077163696289,
      "learning_rate": 3.650928167877321e-05,
      "loss": 1.0408,
      "step": 15840
    },
    {
      "epoch": 1.2741157556270095,
      "grad_norm": 2.143214702606201,
      "learning_rate": 3.646892655367232e-05,
      "loss": 1.0145,
      "step": 15850
    },
    {
      "epoch": 1.27491961414791,
      "grad_norm": 2.9116787910461426,
      "learning_rate": 3.642857142857143e-05,
      "loss": 0.9369,
      "step": 15860
    },
    {
      "epoch": 1.2757234726688103,
      "grad_norm": 1.6376276016235352,
      "learning_rate": 3.638821630347054e-05,
      "loss": 1.0382,
      "step": 15870
    },
    {
      "epoch": 1.2765273311897105,
      "grad_norm": 1.4802640676498413,
      "learning_rate": 3.634786117836965e-05,
      "loss": 1.0154,
      "step": 15880
    },
    {
      "epoch": 1.277331189710611,
      "grad_norm": 1.6010328531265259,
      "learning_rate": 3.630750605326877e-05,
      "loss": 0.9832,
      "step": 15890
    },
    {
      "epoch": 1.2781350482315113,
      "grad_norm": 2.688264846801758,
      "learning_rate": 3.626715092816788e-05,
      "loss": 0.9747,
      "step": 15900
    },
    {
      "epoch": 1.2789389067524115,
      "grad_norm": 1.729894995689392,
      "learning_rate": 3.6226795803066994e-05,
      "loss": 0.9999,
      "step": 15910
    },
    {
      "epoch": 1.279742765273312,
      "grad_norm": 1.8605012893676758,
      "learning_rate": 3.6186440677966105e-05,
      "loss": 0.9808,
      "step": 15920
    },
    {
      "epoch": 1.2805466237942122,
      "grad_norm": 1.628440499305725,
      "learning_rate": 3.6146085552865217e-05,
      "loss": 1.0222,
      "step": 15930
    },
    {
      "epoch": 1.2813504823151125,
      "grad_norm": 1.3880386352539062,
      "learning_rate": 3.610573042776433e-05,
      "loss": 0.9112,
      "step": 15940
    },
    {
      "epoch": 1.282154340836013,
      "grad_norm": 1.6189711093902588,
      "learning_rate": 3.606537530266344e-05,
      "loss": 1.1685,
      "step": 15950
    },
    {
      "epoch": 1.2829581993569132,
      "grad_norm": 1.9674052000045776,
      "learning_rate": 3.602502017756255e-05,
      "loss": 0.9313,
      "step": 15960
    },
    {
      "epoch": 1.2837620578778135,
      "grad_norm": 1.8231611251831055,
      "learning_rate": 3.598466505246166e-05,
      "loss": 0.8848,
      "step": 15970
    },
    {
      "epoch": 1.2845659163987138,
      "grad_norm": 1.5274815559387207,
      "learning_rate": 3.594430992736077e-05,
      "loss": 0.9129,
      "step": 15980
    },
    {
      "epoch": 1.285369774919614,
      "grad_norm": 2.585059881210327,
      "learning_rate": 3.5903954802259885e-05,
      "loss": 1.0566,
      "step": 15990
    },
    {
      "epoch": 1.2861736334405145,
      "grad_norm": 2.084228992462158,
      "learning_rate": 3.5863599677158996e-05,
      "loss": 1.0839,
      "step": 16000
    },
    {
      "epoch": 1.2861736334405145,
      "eval_yahma/alpaca-cleaned_loss": 1.2161176204681396,
      "eval_yahma/alpaca-cleaned_runtime": 115.6721,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.29,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.161,
      "step": 16000
    },
    {
      "epoch": 1.2869774919614148,
      "grad_norm": 1.84796142578125,
      "learning_rate": 3.582324455205811e-05,
      "loss": 0.9685,
      "step": 16010
    },
    {
      "epoch": 1.287781350482315,
      "grad_norm": 1.575564980506897,
      "learning_rate": 3.5782889426957226e-05,
      "loss": 1.0034,
      "step": 16020
    },
    {
      "epoch": 1.2885852090032155,
      "grad_norm": 1.797165870666504,
      "learning_rate": 3.574253430185634e-05,
      "loss": 1.0284,
      "step": 16030
    },
    {
      "epoch": 1.2893890675241158,
      "grad_norm": 1.263707160949707,
      "learning_rate": 3.570217917675545e-05,
      "loss": 1.0448,
      "step": 16040
    },
    {
      "epoch": 1.290192926045016,
      "grad_norm": 1.7160576581954956,
      "learning_rate": 3.5661824051654566e-05,
      "loss": 0.9389,
      "step": 16050
    },
    {
      "epoch": 1.2909967845659165,
      "grad_norm": 2.0201547145843506,
      "learning_rate": 3.562146892655368e-05,
      "loss": 0.9839,
      "step": 16060
    },
    {
      "epoch": 1.2918006430868167,
      "grad_norm": 1.3020087480545044,
      "learning_rate": 3.558111380145279e-05,
      "loss": 1.0517,
      "step": 16070
    },
    {
      "epoch": 1.292604501607717,
      "grad_norm": 1.6637574434280396,
      "learning_rate": 3.55407586763519e-05,
      "loss": 0.9596,
      "step": 16080
    },
    {
      "epoch": 1.2934083601286175,
      "grad_norm": 2.665010690689087,
      "learning_rate": 3.550040355125101e-05,
      "loss": 1.0649,
      "step": 16090
    },
    {
      "epoch": 1.2942122186495177,
      "grad_norm": 1.6599992513656616,
      "learning_rate": 3.546004842615012e-05,
      "loss": 0.9902,
      "step": 16100
    },
    {
      "epoch": 1.295016077170418,
      "grad_norm": 1.390799641609192,
      "learning_rate": 3.5419693301049235e-05,
      "loss": 1.0083,
      "step": 16110
    },
    {
      "epoch": 1.2958199356913183,
      "grad_norm": 4.305990695953369,
      "learning_rate": 3.5379338175948346e-05,
      "loss": 0.9338,
      "step": 16120
    },
    {
      "epoch": 1.2966237942122185,
      "grad_norm": 2.046499252319336,
      "learning_rate": 3.533898305084746e-05,
      "loss": 1.0253,
      "step": 16130
    },
    {
      "epoch": 1.297427652733119,
      "grad_norm": 2.5173068046569824,
      "learning_rate": 3.529862792574657e-05,
      "loss": 1.0404,
      "step": 16140
    },
    {
      "epoch": 1.2982315112540193,
      "grad_norm": 1.7286467552185059,
      "learning_rate": 3.525827280064569e-05,
      "loss": 1.0585,
      "step": 16150
    },
    {
      "epoch": 1.2990353697749195,
      "grad_norm": 2.4116015434265137,
      "learning_rate": 3.52179176755448e-05,
      "loss": 0.9991,
      "step": 16160
    },
    {
      "epoch": 1.29983922829582,
      "grad_norm": 2.464102029800415,
      "learning_rate": 3.517756255044391e-05,
      "loss": 0.9679,
      "step": 16170
    },
    {
      "epoch": 1.3006430868167203,
      "grad_norm": 1.9731477499008179,
      "learning_rate": 3.513720742534302e-05,
      "loss": 0.9587,
      "step": 16180
    },
    {
      "epoch": 1.3014469453376205,
      "grad_norm": 1.2134771347045898,
      "learning_rate": 3.509685230024213e-05,
      "loss": 1.0607,
      "step": 16190
    },
    {
      "epoch": 1.302250803858521,
      "grad_norm": 3.209575891494751,
      "learning_rate": 3.5056497175141244e-05,
      "loss": 0.964,
      "step": 16200
    },
    {
      "epoch": 1.3030546623794212,
      "grad_norm": 1.5849529504776,
      "learning_rate": 3.5016142050040355e-05,
      "loss": 0.8894,
      "step": 16210
    },
    {
      "epoch": 1.3038585209003215,
      "grad_norm": 2.5455405712127686,
      "learning_rate": 3.4975786924939466e-05,
      "loss": 1.0194,
      "step": 16220
    },
    {
      "epoch": 1.304662379421222,
      "grad_norm": 1.4014263153076172,
      "learning_rate": 3.493543179983858e-05,
      "loss": 0.9338,
      "step": 16230
    },
    {
      "epoch": 1.3054662379421222,
      "grad_norm": 1.5951437950134277,
      "learning_rate": 3.489507667473769e-05,
      "loss": 0.8486,
      "step": 16240
    },
    {
      "epoch": 1.3062700964630225,
      "grad_norm": 1.797086238861084,
      "learning_rate": 3.485472154963681e-05,
      "loss": 0.9986,
      "step": 16250
    },
    {
      "epoch": 1.3070739549839228,
      "grad_norm": 2.1635706424713135,
      "learning_rate": 3.481436642453592e-05,
      "loss": 0.9768,
      "step": 16260
    },
    {
      "epoch": 1.307877813504823,
      "grad_norm": 2.373884916305542,
      "learning_rate": 3.477401129943503e-05,
      "loss": 0.9809,
      "step": 16270
    },
    {
      "epoch": 1.3086816720257235,
      "grad_norm": 2.3494770526885986,
      "learning_rate": 3.473365617433415e-05,
      "loss": 1.2571,
      "step": 16280
    },
    {
      "epoch": 1.3094855305466238,
      "grad_norm": 1.4083755016326904,
      "learning_rate": 3.469330104923326e-05,
      "loss": 1.1216,
      "step": 16290
    },
    {
      "epoch": 1.310289389067524,
      "grad_norm": 1.7376712560653687,
      "learning_rate": 3.465294592413237e-05,
      "loss": 0.9893,
      "step": 16300
    },
    {
      "epoch": 1.3110932475884245,
      "grad_norm": 1.5163984298706055,
      "learning_rate": 3.461259079903148e-05,
      "loss": 1.0158,
      "step": 16310
    },
    {
      "epoch": 1.3118971061093248,
      "grad_norm": 1.9004069566726685,
      "learning_rate": 3.4572235673930594e-05,
      "loss": 1.0503,
      "step": 16320
    },
    {
      "epoch": 1.312700964630225,
      "grad_norm": 1.295363426208496,
      "learning_rate": 3.4531880548829705e-05,
      "loss": 1.1076,
      "step": 16330
    },
    {
      "epoch": 1.3135048231511255,
      "grad_norm": 1.682941198348999,
      "learning_rate": 3.4491525423728816e-05,
      "loss": 0.8961,
      "step": 16340
    },
    {
      "epoch": 1.3143086816720257,
      "grad_norm": 1.8301907777786255,
      "learning_rate": 3.445117029862793e-05,
      "loss": 1.0844,
      "step": 16350
    },
    {
      "epoch": 1.315112540192926,
      "grad_norm": 2.9198620319366455,
      "learning_rate": 3.441081517352704e-05,
      "loss": 0.9932,
      "step": 16360
    },
    {
      "epoch": 1.3159163987138265,
      "grad_norm": 2.1517717838287354,
      "learning_rate": 3.437046004842615e-05,
      "loss": 0.9756,
      "step": 16370
    },
    {
      "epoch": 1.3167202572347267,
      "grad_norm": 1.5518659353256226,
      "learning_rate": 3.433010492332526e-05,
      "loss": 0.8814,
      "step": 16380
    },
    {
      "epoch": 1.317524115755627,
      "grad_norm": 2.230884552001953,
      "learning_rate": 3.428974979822437e-05,
      "loss": 0.9049,
      "step": 16390
    },
    {
      "epoch": 1.3183279742765273,
      "grad_norm": 1.4246128797531128,
      "learning_rate": 3.4249394673123485e-05,
      "loss": 1.0478,
      "step": 16400
    },
    {
      "epoch": 1.3191318327974275,
      "grad_norm": 1.648350477218628,
      "learning_rate": 3.4209039548022596e-05,
      "loss": 1.023,
      "step": 16410
    },
    {
      "epoch": 1.319935691318328,
      "grad_norm": 1.7566518783569336,
      "learning_rate": 3.4168684422921714e-05,
      "loss": 1.1123,
      "step": 16420
    },
    {
      "epoch": 1.3207395498392283,
      "grad_norm": 1.8392665386199951,
      "learning_rate": 3.4128329297820825e-05,
      "loss": 0.9804,
      "step": 16430
    },
    {
      "epoch": 1.3215434083601285,
      "grad_norm": 1.5365403890609741,
      "learning_rate": 3.408797417271994e-05,
      "loss": 1.0917,
      "step": 16440
    },
    {
      "epoch": 1.322347266881029,
      "grad_norm": 3.783843755722046,
      "learning_rate": 3.404761904761905e-05,
      "loss": 0.9347,
      "step": 16450
    },
    {
      "epoch": 1.3231511254019293,
      "grad_norm": 2.019134044647217,
      "learning_rate": 3.400726392251816e-05,
      "loss": 0.9739,
      "step": 16460
    },
    {
      "epoch": 1.3239549839228295,
      "grad_norm": 2.0950658321380615,
      "learning_rate": 3.396690879741727e-05,
      "loss": 1.0797,
      "step": 16470
    },
    {
      "epoch": 1.32475884244373,
      "grad_norm": 1.5401766300201416,
      "learning_rate": 3.392655367231639e-05,
      "loss": 1.1162,
      "step": 16480
    },
    {
      "epoch": 1.3255627009646302,
      "grad_norm": 1.9114347696304321,
      "learning_rate": 3.38861985472155e-05,
      "loss": 1.1233,
      "step": 16490
    },
    {
      "epoch": 1.3263665594855305,
      "grad_norm": 1.4473228454589844,
      "learning_rate": 3.384584342211461e-05,
      "loss": 0.9608,
      "step": 16500
    },
    {
      "epoch": 1.327170418006431,
      "grad_norm": 1.3395804166793823,
      "learning_rate": 3.380548829701372e-05,
      "loss": 1.0014,
      "step": 16510
    },
    {
      "epoch": 1.3279742765273312,
      "grad_norm": 1.3364843130111694,
      "learning_rate": 3.3765133171912835e-05,
      "loss": 0.976,
      "step": 16520
    },
    {
      "epoch": 1.3287781350482315,
      "grad_norm": 1.3809840679168701,
      "learning_rate": 3.3724778046811946e-05,
      "loss": 1.0469,
      "step": 16530
    },
    {
      "epoch": 1.3295819935691318,
      "grad_norm": 1.4792019128799438,
      "learning_rate": 3.368442292171106e-05,
      "loss": 1.0364,
      "step": 16540
    },
    {
      "epoch": 1.330385852090032,
      "grad_norm": 2.2897255420684814,
      "learning_rate": 3.3644067796610175e-05,
      "loss": 0.9775,
      "step": 16550
    },
    {
      "epoch": 1.3311897106109325,
      "grad_norm": 2.1644399166107178,
      "learning_rate": 3.360371267150929e-05,
      "loss": 1.152,
      "step": 16560
    },
    {
      "epoch": 1.3319935691318328,
      "grad_norm": 1.5172768831253052,
      "learning_rate": 3.35633575464084e-05,
      "loss": 1.056,
      "step": 16570
    },
    {
      "epoch": 1.332797427652733,
      "grad_norm": 2.014936685562134,
      "learning_rate": 3.352300242130751e-05,
      "loss": 1.0585,
      "step": 16580
    },
    {
      "epoch": 1.3336012861736335,
      "grad_norm": 3.0378189086914062,
      "learning_rate": 3.348264729620662e-05,
      "loss": 1.045,
      "step": 16590
    },
    {
      "epoch": 1.3344051446945338,
      "grad_norm": 1.5051476955413818,
      "learning_rate": 3.344229217110573e-05,
      "loss": 1.0602,
      "step": 16600
    },
    {
      "epoch": 1.335209003215434,
      "grad_norm": 1.7232706546783447,
      "learning_rate": 3.3401937046004844e-05,
      "loss": 1.0584,
      "step": 16610
    },
    {
      "epoch": 1.3360128617363345,
      "grad_norm": 1.534652829170227,
      "learning_rate": 3.3361581920903955e-05,
      "loss": 1.0345,
      "step": 16620
    },
    {
      "epoch": 1.3368167202572347,
      "grad_norm": 2.152040719985962,
      "learning_rate": 3.3321226795803066e-05,
      "loss": 1.013,
      "step": 16630
    },
    {
      "epoch": 1.337620578778135,
      "grad_norm": 1.4761089086532593,
      "learning_rate": 3.328087167070218e-05,
      "loss": 1.0028,
      "step": 16640
    },
    {
      "epoch": 1.3384244372990355,
      "grad_norm": 1.6546519994735718,
      "learning_rate": 3.324051654560129e-05,
      "loss": 1.0939,
      "step": 16650
    },
    {
      "epoch": 1.3392282958199357,
      "grad_norm": 1.352799892425537,
      "learning_rate": 3.32001614205004e-05,
      "loss": 1.0636,
      "step": 16660
    },
    {
      "epoch": 1.340032154340836,
      "grad_norm": 1.700903058052063,
      "learning_rate": 3.315980629539951e-05,
      "loss": 0.9948,
      "step": 16670
    },
    {
      "epoch": 1.3408360128617363,
      "grad_norm": 3.408297061920166,
      "learning_rate": 3.311945117029863e-05,
      "loss": 0.9873,
      "step": 16680
    },
    {
      "epoch": 1.3416398713826365,
      "grad_norm": 1.5373080968856812,
      "learning_rate": 3.307909604519774e-05,
      "loss": 1.0707,
      "step": 16690
    },
    {
      "epoch": 1.342443729903537,
      "grad_norm": 1.3401768207550049,
      "learning_rate": 3.303874092009685e-05,
      "loss": 1.0821,
      "step": 16700
    },
    {
      "epoch": 1.3432475884244373,
      "grad_norm": 2.121828556060791,
      "learning_rate": 3.2998385794995964e-05,
      "loss": 1.0746,
      "step": 16710
    },
    {
      "epoch": 1.3440514469453375,
      "grad_norm": 1.4472429752349854,
      "learning_rate": 3.295803066989508e-05,
      "loss": 1.0205,
      "step": 16720
    },
    {
      "epoch": 1.344855305466238,
      "grad_norm": 1.9297133684158325,
      "learning_rate": 3.2917675544794194e-05,
      "loss": 1.0387,
      "step": 16730
    },
    {
      "epoch": 1.3456591639871383,
      "grad_norm": 1.7948999404907227,
      "learning_rate": 3.2877320419693305e-05,
      "loss": 0.9557,
      "step": 16740
    },
    {
      "epoch": 1.3464630225080385,
      "grad_norm": 2.6864187717437744,
      "learning_rate": 3.2836965294592416e-05,
      "loss": 0.9276,
      "step": 16750
    },
    {
      "epoch": 1.347266881028939,
      "grad_norm": 1.9042679071426392,
      "learning_rate": 3.279661016949153e-05,
      "loss": 0.9675,
      "step": 16760
    },
    {
      "epoch": 1.3480707395498392,
      "grad_norm": 1.5159459114074707,
      "learning_rate": 3.275625504439064e-05,
      "loss": 0.9691,
      "step": 16770
    },
    {
      "epoch": 1.3488745980707395,
      "grad_norm": 2.456352949142456,
      "learning_rate": 3.271589991928975e-05,
      "loss": 1.0205,
      "step": 16780
    },
    {
      "epoch": 1.34967845659164,
      "grad_norm": 1.351737380027771,
      "learning_rate": 3.267554479418886e-05,
      "loss": 0.943,
      "step": 16790
    },
    {
      "epoch": 1.3504823151125402,
      "grad_norm": 1.9734419584274292,
      "learning_rate": 3.263518966908797e-05,
      "loss": 1.0482,
      "step": 16800
    },
    {
      "epoch": 1.3512861736334405,
      "grad_norm": 1.700422763824463,
      "learning_rate": 3.259483454398709e-05,
      "loss": 1.042,
      "step": 16810
    },
    {
      "epoch": 1.3520900321543408,
      "grad_norm": 1.9596405029296875,
      "learning_rate": 3.25544794188862e-05,
      "loss": 1.0004,
      "step": 16820
    },
    {
      "epoch": 1.3528938906752412,
      "grad_norm": 1.4773586988449097,
      "learning_rate": 3.2514124293785314e-05,
      "loss": 0.9757,
      "step": 16830
    },
    {
      "epoch": 1.3536977491961415,
      "grad_norm": 1.8567955493927002,
      "learning_rate": 3.2473769168684425e-05,
      "loss": 1.0112,
      "step": 16840
    },
    {
      "epoch": 1.3545016077170418,
      "grad_norm": 1.7968416213989258,
      "learning_rate": 3.243341404358354e-05,
      "loss": 0.896,
      "step": 16850
    },
    {
      "epoch": 1.355305466237942,
      "grad_norm": 1.8025298118591309,
      "learning_rate": 3.239305891848265e-05,
      "loss": 1.0415,
      "step": 16860
    },
    {
      "epoch": 1.3561093247588425,
      "grad_norm": 1.930948257446289,
      "learning_rate": 3.235270379338176e-05,
      "loss": 0.9868,
      "step": 16870
    },
    {
      "epoch": 1.3569131832797428,
      "grad_norm": 1.6675337553024292,
      "learning_rate": 3.231234866828087e-05,
      "loss": 0.94,
      "step": 16880
    },
    {
      "epoch": 1.357717041800643,
      "grad_norm": 1.5774610042572021,
      "learning_rate": 3.227199354317998e-05,
      "loss": 0.9087,
      "step": 16890
    },
    {
      "epoch": 1.3585209003215435,
      "grad_norm": 2.006761074066162,
      "learning_rate": 3.2231638418079093e-05,
      "loss": 1.0439,
      "step": 16900
    },
    {
      "epoch": 1.3593247588424437,
      "grad_norm": 2.1158947944641113,
      "learning_rate": 3.2191283292978205e-05,
      "loss": 1.004,
      "step": 16910
    },
    {
      "epoch": 1.360128617363344,
      "grad_norm": 1.8996459245681763,
      "learning_rate": 3.215092816787732e-05,
      "loss": 0.9966,
      "step": 16920
    },
    {
      "epoch": 1.3609324758842445,
      "grad_norm": 1.500229835510254,
      "learning_rate": 3.2110573042776434e-05,
      "loss": 0.8914,
      "step": 16930
    },
    {
      "epoch": 1.3617363344051447,
      "grad_norm": 3.478053092956543,
      "learning_rate": 3.2070217917675546e-05,
      "loss": 1.037,
      "step": 16940
    },
    {
      "epoch": 1.362540192926045,
      "grad_norm": 1.6184499263763428,
      "learning_rate": 3.2029862792574664e-05,
      "loss": 0.9153,
      "step": 16950
    },
    {
      "epoch": 1.3633440514469453,
      "grad_norm": 1.5846716165542603,
      "learning_rate": 3.1989507667473775e-05,
      "loss": 1.0075,
      "step": 16960
    },
    {
      "epoch": 1.3641479099678457,
      "grad_norm": 2.4219467639923096,
      "learning_rate": 3.1949152542372887e-05,
      "loss": 1.0313,
      "step": 16970
    },
    {
      "epoch": 1.364951768488746,
      "grad_norm": 1.4937620162963867,
      "learning_rate": 3.1908797417272e-05,
      "loss": 1.0529,
      "step": 16980
    },
    {
      "epoch": 1.3657556270096463,
      "grad_norm": 1.4323031902313232,
      "learning_rate": 3.186844229217111e-05,
      "loss": 0.9315,
      "step": 16990
    },
    {
      "epoch": 1.3665594855305465,
      "grad_norm": 1.852509617805481,
      "learning_rate": 3.182808716707022e-05,
      "loss": 1.0359,
      "step": 17000
    },
    {
      "epoch": 1.367363344051447,
      "grad_norm": 1.4422820806503296,
      "learning_rate": 3.178773204196933e-05,
      "loss": 1.0239,
      "step": 17010
    },
    {
      "epoch": 1.3681672025723473,
      "grad_norm": 5.013046741485596,
      "learning_rate": 3.1747376916868443e-05,
      "loss": 0.8221,
      "step": 17020
    },
    {
      "epoch": 1.3689710610932475,
      "grad_norm": 1.6348212957382202,
      "learning_rate": 3.1707021791767555e-05,
      "loss": 0.9839,
      "step": 17030
    },
    {
      "epoch": 1.369774919614148,
      "grad_norm": 2.1785271167755127,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 1.135,
      "step": 17040
    },
    {
      "epoch": 1.3705787781350482,
      "grad_norm": 2.1272568702697754,
      "learning_rate": 3.162631154156578e-05,
      "loss": 1.1353,
      "step": 17050
    },
    {
      "epoch": 1.3713826366559485,
      "grad_norm": 3.070270299911499,
      "learning_rate": 3.158595641646489e-05,
      "loss": 1.0135,
      "step": 17060
    },
    {
      "epoch": 1.372186495176849,
      "grad_norm": 1.5802148580551147,
      "learning_rate": 3.1545601291364e-05,
      "loss": 1.0158,
      "step": 17070
    },
    {
      "epoch": 1.3729903536977492,
      "grad_norm": 1.7264318466186523,
      "learning_rate": 3.150524616626312e-05,
      "loss": 0.9694,
      "step": 17080
    },
    {
      "epoch": 1.3737942122186495,
      "grad_norm": 2.184271812438965,
      "learning_rate": 3.146489104116223e-05,
      "loss": 1.0871,
      "step": 17090
    },
    {
      "epoch": 1.3745980707395498,
      "grad_norm": 1.7589523792266846,
      "learning_rate": 3.142453591606134e-05,
      "loss": 1.0919,
      "step": 17100
    },
    {
      "epoch": 1.3754019292604502,
      "grad_norm": 1.6874173879623413,
      "learning_rate": 3.138418079096045e-05,
      "loss": 1.1151,
      "step": 17110
    },
    {
      "epoch": 1.3762057877813505,
      "grad_norm": 1.6541131734848022,
      "learning_rate": 3.1343825665859564e-05,
      "loss": 1.118,
      "step": 17120
    },
    {
      "epoch": 1.3770096463022508,
      "grad_norm": 1.4665961265563965,
      "learning_rate": 3.1303470540758675e-05,
      "loss": 1.0108,
      "step": 17130
    },
    {
      "epoch": 1.377813504823151,
      "grad_norm": 2.157575845718384,
      "learning_rate": 3.1263115415657787e-05,
      "loss": 1.0388,
      "step": 17140
    },
    {
      "epoch": 1.3786173633440515,
      "grad_norm": 1.7408186197280884,
      "learning_rate": 3.1222760290556905e-05,
      "loss": 0.9634,
      "step": 17150
    },
    {
      "epoch": 1.3794212218649518,
      "grad_norm": 1.8646273612976074,
      "learning_rate": 3.1182405165456016e-05,
      "loss": 1.1186,
      "step": 17160
    },
    {
      "epoch": 1.380225080385852,
      "grad_norm": 1.4674321413040161,
      "learning_rate": 3.114205004035513e-05,
      "loss": 0.9969,
      "step": 17170
    },
    {
      "epoch": 1.3810289389067525,
      "grad_norm": 1.7009894847869873,
      "learning_rate": 3.110169491525424e-05,
      "loss": 1.0342,
      "step": 17180
    },
    {
      "epoch": 1.3818327974276527,
      "grad_norm": 1.473688006401062,
      "learning_rate": 3.106133979015335e-05,
      "loss": 0.9908,
      "step": 17190
    },
    {
      "epoch": 1.382636655948553,
      "grad_norm": 1.7846876382827759,
      "learning_rate": 3.102098466505246e-05,
      "loss": 0.9985,
      "step": 17200
    },
    {
      "epoch": 1.3834405144694535,
      "grad_norm": 1.3931621313095093,
      "learning_rate": 3.098062953995158e-05,
      "loss": 1.059,
      "step": 17210
    },
    {
      "epoch": 1.3842443729903537,
      "grad_norm": 1.5702584981918335,
      "learning_rate": 3.094027441485069e-05,
      "loss": 1.0033,
      "step": 17220
    },
    {
      "epoch": 1.385048231511254,
      "grad_norm": 1.5941625833511353,
      "learning_rate": 3.08999192897498e-05,
      "loss": 1.056,
      "step": 17230
    },
    {
      "epoch": 1.3858520900321543,
      "grad_norm": 1.9544049501419067,
      "learning_rate": 3.0859564164648914e-05,
      "loss": 1.0103,
      "step": 17240
    },
    {
      "epoch": 1.3866559485530547,
      "grad_norm": 1.4904630184173584,
      "learning_rate": 3.0819209039548025e-05,
      "loss": 1.074,
      "step": 17250
    },
    {
      "epoch": 1.387459807073955,
      "grad_norm": 2.1309163570404053,
      "learning_rate": 3.0778853914447137e-05,
      "loss": 1.0918,
      "step": 17260
    },
    {
      "epoch": 1.3882636655948553,
      "grad_norm": 1.3276787996292114,
      "learning_rate": 3.073849878934625e-05,
      "loss": 1.0247,
      "step": 17270
    },
    {
      "epoch": 1.3890675241157555,
      "grad_norm": 2.309291124343872,
      "learning_rate": 3.069814366424536e-05,
      "loss": 0.9255,
      "step": 17280
    },
    {
      "epoch": 1.389871382636656,
      "grad_norm": 1.6173782348632812,
      "learning_rate": 3.065778853914447e-05,
      "loss": 1.0273,
      "step": 17290
    },
    {
      "epoch": 1.3906752411575563,
      "grad_norm": 2.0882816314697266,
      "learning_rate": 3.061743341404358e-05,
      "loss": 1.0799,
      "step": 17300
    },
    {
      "epoch": 1.3914790996784565,
      "grad_norm": 1.4501557350158691,
      "learning_rate": 3.057707828894269e-05,
      "loss": 0.9923,
      "step": 17310
    },
    {
      "epoch": 1.392282958199357,
      "grad_norm": 1.792034387588501,
      "learning_rate": 3.0536723163841805e-05,
      "loss": 1.075,
      "step": 17320
    },
    {
      "epoch": 1.3930868167202572,
      "grad_norm": 1.9524577856063843,
      "learning_rate": 3.049636803874092e-05,
      "loss": 1.1379,
      "step": 17330
    },
    {
      "epoch": 1.3938906752411575,
      "grad_norm": 1.2398579120635986,
      "learning_rate": 3.0456012913640038e-05,
      "loss": 0.9899,
      "step": 17340
    },
    {
      "epoch": 1.394694533762058,
      "grad_norm": 1.7992384433746338,
      "learning_rate": 3.041565778853915e-05,
      "loss": 1.0277,
      "step": 17350
    },
    {
      "epoch": 1.3954983922829582,
      "grad_norm": 2.0077879428863525,
      "learning_rate": 3.037530266343826e-05,
      "loss": 1.0122,
      "step": 17360
    },
    {
      "epoch": 1.3963022508038585,
      "grad_norm": 1.895276665687561,
      "learning_rate": 3.033494753833737e-05,
      "loss": 1.0742,
      "step": 17370
    },
    {
      "epoch": 1.3971061093247588,
      "grad_norm": 1.7623761892318726,
      "learning_rate": 3.0294592413236483e-05,
      "loss": 1.0696,
      "step": 17380
    },
    {
      "epoch": 1.3979099678456592,
      "grad_norm": 2.2184484004974365,
      "learning_rate": 3.0254237288135594e-05,
      "loss": 0.9861,
      "step": 17390
    },
    {
      "epoch": 1.3987138263665595,
      "grad_norm": 1.6133798360824585,
      "learning_rate": 3.0213882163034706e-05,
      "loss": 1.043,
      "step": 17400
    },
    {
      "epoch": 1.3995176848874598,
      "grad_norm": 1.8008503913879395,
      "learning_rate": 3.0173527037933817e-05,
      "loss": 1.035,
      "step": 17410
    },
    {
      "epoch": 1.40032154340836,
      "grad_norm": 2.023869037628174,
      "learning_rate": 3.0133171912832932e-05,
      "loss": 1.0238,
      "step": 17420
    },
    {
      "epoch": 1.4011254019292605,
      "grad_norm": 1.7300667762756348,
      "learning_rate": 3.0092816787732043e-05,
      "loss": 1.0263,
      "step": 17430
    },
    {
      "epoch": 1.4019292604501608,
      "grad_norm": 1.5736720561981201,
      "learning_rate": 3.0052461662631155e-05,
      "loss": 1.0354,
      "step": 17440
    },
    {
      "epoch": 1.402733118971061,
      "grad_norm": 1.8412175178527832,
      "learning_rate": 3.0012106537530266e-05,
      "loss": 1.0392,
      "step": 17450
    },
    {
      "epoch": 1.4035369774919615,
      "grad_norm": 1.3952903747558594,
      "learning_rate": 2.9971751412429377e-05,
      "loss": 0.9645,
      "step": 17460
    },
    {
      "epoch": 1.4043408360128617,
      "grad_norm": 1.6282439231872559,
      "learning_rate": 2.9931396287328496e-05,
      "loss": 1.1068,
      "step": 17470
    },
    {
      "epoch": 1.405144694533762,
      "grad_norm": 2.133103132247925,
      "learning_rate": 2.9891041162227607e-05,
      "loss": 0.9765,
      "step": 17480
    },
    {
      "epoch": 1.4059485530546625,
      "grad_norm": 2.174391508102417,
      "learning_rate": 2.9850686037126718e-05,
      "loss": 1.093,
      "step": 17490
    },
    {
      "epoch": 1.4067524115755627,
      "grad_norm": 2.0915133953094482,
      "learning_rate": 2.981033091202583e-05,
      "loss": 0.983,
      "step": 17500
    },
    {
      "epoch": 1.407556270096463,
      "grad_norm": 1.40506112575531,
      "learning_rate": 2.976997578692494e-05,
      "loss": 0.9038,
      "step": 17510
    },
    {
      "epoch": 1.4083601286173635,
      "grad_norm": 1.5525906085968018,
      "learning_rate": 2.9729620661824052e-05,
      "loss": 1.0389,
      "step": 17520
    },
    {
      "epoch": 1.4091639871382637,
      "grad_norm": 1.559783697128296,
      "learning_rate": 2.9689265536723164e-05,
      "loss": 0.996,
      "step": 17530
    },
    {
      "epoch": 1.409967845659164,
      "grad_norm": 1.8317124843597412,
      "learning_rate": 2.964891041162228e-05,
      "loss": 1.1198,
      "step": 17540
    },
    {
      "epoch": 1.4107717041800643,
      "grad_norm": 1.9380903244018555,
      "learning_rate": 2.960855528652139e-05,
      "loss": 1.0118,
      "step": 17550
    },
    {
      "epoch": 1.4115755627009645,
      "grad_norm": 2.3650496006011963,
      "learning_rate": 2.95682001614205e-05,
      "loss": 0.9,
      "step": 17560
    },
    {
      "epoch": 1.412379421221865,
      "grad_norm": 1.747936487197876,
      "learning_rate": 2.9527845036319613e-05,
      "loss": 1.1245,
      "step": 17570
    },
    {
      "epoch": 1.4131832797427653,
      "grad_norm": 3.0566165447235107,
      "learning_rate": 2.9487489911218724e-05,
      "loss": 1.0397,
      "step": 17580
    },
    {
      "epoch": 1.4139871382636655,
      "grad_norm": 1.790842890739441,
      "learning_rate": 2.9447134786117835e-05,
      "loss": 0.893,
      "step": 17590
    },
    {
      "epoch": 1.414790996784566,
      "grad_norm": 2.7602193355560303,
      "learning_rate": 2.9406779661016953e-05,
      "loss": 1.0208,
      "step": 17600
    },
    {
      "epoch": 1.4155948553054662,
      "grad_norm": 2.4181101322174072,
      "learning_rate": 2.9366424535916065e-05,
      "loss": 1.0231,
      "step": 17610
    },
    {
      "epoch": 1.4163987138263665,
      "grad_norm": 1.5748388767242432,
      "learning_rate": 2.9326069410815176e-05,
      "loss": 0.9656,
      "step": 17620
    },
    {
      "epoch": 1.417202572347267,
      "grad_norm": 1.6972712278366089,
      "learning_rate": 2.9285714285714288e-05,
      "loss": 0.8361,
      "step": 17630
    },
    {
      "epoch": 1.4180064308681672,
      "grad_norm": 1.8888636827468872,
      "learning_rate": 2.92453591606134e-05,
      "loss": 0.9267,
      "step": 17640
    },
    {
      "epoch": 1.4188102893890675,
      "grad_norm": 1.4726215600967407,
      "learning_rate": 2.9205004035512514e-05,
      "loss": 0.9852,
      "step": 17650
    },
    {
      "epoch": 1.419614147909968,
      "grad_norm": 1.650896430015564,
      "learning_rate": 2.9164648910411625e-05,
      "loss": 0.9401,
      "step": 17660
    },
    {
      "epoch": 1.4204180064308682,
      "grad_norm": 1.576968789100647,
      "learning_rate": 2.9124293785310736e-05,
      "loss": 1.0903,
      "step": 17670
    },
    {
      "epoch": 1.4212218649517685,
      "grad_norm": 2.186856985092163,
      "learning_rate": 2.9083938660209848e-05,
      "loss": 0.9751,
      "step": 17680
    },
    {
      "epoch": 1.4220257234726688,
      "grad_norm": 1.621895432472229,
      "learning_rate": 2.904358353510896e-05,
      "loss": 0.9995,
      "step": 17690
    },
    {
      "epoch": 1.422829581993569,
      "grad_norm": 1.6178061962127686,
      "learning_rate": 2.900322841000807e-05,
      "loss": 0.9885,
      "step": 17700
    },
    {
      "epoch": 1.4236334405144695,
      "grad_norm": 3.5141470432281494,
      "learning_rate": 2.8962873284907182e-05,
      "loss": 0.8756,
      "step": 17710
    },
    {
      "epoch": 1.4244372990353698,
      "grad_norm": 1.3309582471847534,
      "learning_rate": 2.8922518159806293e-05,
      "loss": 0.9736,
      "step": 17720
    },
    {
      "epoch": 1.42524115755627,
      "grad_norm": 1.4174798727035522,
      "learning_rate": 2.888216303470541e-05,
      "loss": 1.1252,
      "step": 17730
    },
    {
      "epoch": 1.4260450160771705,
      "grad_norm": 1.849108338356018,
      "learning_rate": 2.8841807909604523e-05,
      "loss": 0.9253,
      "step": 17740
    },
    {
      "epoch": 1.4268488745980707,
      "grad_norm": 1.832971215248108,
      "learning_rate": 2.8801452784503634e-05,
      "loss": 0.9816,
      "step": 17750
    },
    {
      "epoch": 1.427652733118971,
      "grad_norm": 1.9136488437652588,
      "learning_rate": 2.8761097659402745e-05,
      "loss": 1.0795,
      "step": 17760
    },
    {
      "epoch": 1.4284565916398715,
      "grad_norm": 1.530927300453186,
      "learning_rate": 2.872074253430186e-05,
      "loss": 1.0021,
      "step": 17770
    },
    {
      "epoch": 1.4292604501607717,
      "grad_norm": 2.640871047973633,
      "learning_rate": 2.868038740920097e-05,
      "loss": 0.9635,
      "step": 17780
    },
    {
      "epoch": 1.430064308681672,
      "grad_norm": 1.6071547269821167,
      "learning_rate": 2.8640032284100083e-05,
      "loss": 1.0267,
      "step": 17790
    },
    {
      "epoch": 1.4308681672025725,
      "grad_norm": 1.395981788635254,
      "learning_rate": 2.8599677158999194e-05,
      "loss": 0.939,
      "step": 17800
    },
    {
      "epoch": 1.4316720257234727,
      "grad_norm": 1.8101909160614014,
      "learning_rate": 2.8559322033898306e-05,
      "loss": 1.0454,
      "step": 17810
    },
    {
      "epoch": 1.432475884244373,
      "grad_norm": 1.7596745491027832,
      "learning_rate": 2.8518966908797417e-05,
      "loss": 1.0245,
      "step": 17820
    },
    {
      "epoch": 1.4332797427652733,
      "grad_norm": 1.522783875465393,
      "learning_rate": 2.847861178369653e-05,
      "loss": 1.007,
      "step": 17830
    },
    {
      "epoch": 1.4340836012861735,
      "grad_norm": 1.8518741130828857,
      "learning_rate": 2.843825665859564e-05,
      "loss": 1.0237,
      "step": 17840
    },
    {
      "epoch": 1.434887459807074,
      "grad_norm": 1.7670881748199463,
      "learning_rate": 2.8397901533494754e-05,
      "loss": 0.9962,
      "step": 17850
    },
    {
      "epoch": 1.4356913183279743,
      "grad_norm": 1.9033243656158447,
      "learning_rate": 2.8357546408393866e-05,
      "loss": 0.9878,
      "step": 17860
    },
    {
      "epoch": 1.4364951768488745,
      "grad_norm": 2.2080554962158203,
      "learning_rate": 2.831719128329298e-05,
      "loss": 1.0062,
      "step": 17870
    },
    {
      "epoch": 1.437299035369775,
      "grad_norm": 1.6603493690490723,
      "learning_rate": 2.8276836158192095e-05,
      "loss": 0.9851,
      "step": 17880
    },
    {
      "epoch": 1.4381028938906752,
      "grad_norm": 1.9345533847808838,
      "learning_rate": 2.8236481033091207e-05,
      "loss": 1.0904,
      "step": 17890
    },
    {
      "epoch": 1.4389067524115755,
      "grad_norm": 1.8384795188903809,
      "learning_rate": 2.8196125907990318e-05,
      "loss": 0.9186,
      "step": 17900
    },
    {
      "epoch": 1.439710610932476,
      "grad_norm": 1.4033855199813843,
      "learning_rate": 2.815577078288943e-05,
      "loss": 1.002,
      "step": 17910
    },
    {
      "epoch": 1.4405144694533762,
      "grad_norm": 1.9541116952896118,
      "learning_rate": 2.811541565778854e-05,
      "loss": 1.0194,
      "step": 17920
    },
    {
      "epoch": 1.4413183279742765,
      "grad_norm": 1.4612561464309692,
      "learning_rate": 2.8075060532687652e-05,
      "loss": 1.0518,
      "step": 17930
    },
    {
      "epoch": 1.442122186495177,
      "grad_norm": 1.920513391494751,
      "learning_rate": 2.8034705407586764e-05,
      "loss": 1.0144,
      "step": 17940
    },
    {
      "epoch": 1.4429260450160772,
      "grad_norm": 1.4925817251205444,
      "learning_rate": 2.7994350282485875e-05,
      "loss": 1.0138,
      "step": 17950
    },
    {
      "epoch": 1.4437299035369775,
      "grad_norm": 1.4273995161056519,
      "learning_rate": 2.7953995157384986e-05,
      "loss": 0.9795,
      "step": 17960
    },
    {
      "epoch": 1.4445337620578778,
      "grad_norm": 1.4286296367645264,
      "learning_rate": 2.79136400322841e-05,
      "loss": 1.0259,
      "step": 17970
    },
    {
      "epoch": 1.445337620578778,
      "grad_norm": 1.9147309064865112,
      "learning_rate": 2.7873284907183212e-05,
      "loss": 1.1585,
      "step": 17980
    },
    {
      "epoch": 1.4461414790996785,
      "grad_norm": 2.371657133102417,
      "learning_rate": 2.7832929782082324e-05,
      "loss": 1.0628,
      "step": 17990
    },
    {
      "epoch": 1.4469453376205788,
      "grad_norm": 2.483755588531494,
      "learning_rate": 2.7792574656981442e-05,
      "loss": 1.0018,
      "step": 18000
    },
    {
      "epoch": 1.4469453376205788,
      "eval_yahma/alpaca-cleaned_loss": 1.2110522985458374,
      "eval_yahma/alpaca-cleaned_runtime": 115.7566,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.278,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.16,
      "step": 18000
    },
    {
      "epoch": 1.447749196141479,
      "grad_norm": 1.9039057493209839,
      "learning_rate": 2.7752219531880553e-05,
      "loss": 1.0801,
      "step": 18010
    },
    {
      "epoch": 1.4485530546623795,
      "grad_norm": 3.932539224624634,
      "learning_rate": 2.7711864406779665e-05,
      "loss": 0.9159,
      "step": 18020
    },
    {
      "epoch": 1.4493569131832797,
      "grad_norm": 1.7187069654464722,
      "learning_rate": 2.7675544794188862e-05,
      "loss": 1.1043,
      "step": 18030
    },
    {
      "epoch": 1.45016077170418,
      "grad_norm": 2.409668445587158,
      "learning_rate": 2.7635189669087974e-05,
      "loss": 0.9993,
      "step": 18040
    },
    {
      "epoch": 1.4509646302250805,
      "grad_norm": 1.4845926761627197,
      "learning_rate": 2.7594834543987085e-05,
      "loss": 0.9123,
      "step": 18050
    },
    {
      "epoch": 1.4517684887459807,
      "grad_norm": 1.4308907985687256,
      "learning_rate": 2.7554479418886196e-05,
      "loss": 1.0153,
      "step": 18060
    },
    {
      "epoch": 1.452572347266881,
      "grad_norm": 1.526442050933838,
      "learning_rate": 2.751412429378531e-05,
      "loss": 0.8933,
      "step": 18070
    },
    {
      "epoch": 1.4533762057877815,
      "grad_norm": 1.9068630933761597,
      "learning_rate": 2.7473769168684426e-05,
      "loss": 0.9328,
      "step": 18080
    },
    {
      "epoch": 1.4541800643086817,
      "grad_norm": 1.8453564643859863,
      "learning_rate": 2.7433414043583537e-05,
      "loss": 0.9831,
      "step": 18090
    },
    {
      "epoch": 1.454983922829582,
      "grad_norm": 1.743841290473938,
      "learning_rate": 2.7393058918482652e-05,
      "loss": 1.0306,
      "step": 18100
    },
    {
      "epoch": 1.4557877813504823,
      "grad_norm": 1.9150433540344238,
      "learning_rate": 2.7352703793381763e-05,
      "loss": 1.1571,
      "step": 18110
    },
    {
      "epoch": 1.4565916398713825,
      "grad_norm": 2.1634671688079834,
      "learning_rate": 2.7312348668280875e-05,
      "loss": 0.9588,
      "step": 18120
    },
    {
      "epoch": 1.457395498392283,
      "grad_norm": 1.8215436935424805,
      "learning_rate": 2.7271993543179986e-05,
      "loss": 1.1203,
      "step": 18130
    },
    {
      "epoch": 1.4581993569131833,
      "grad_norm": 1.6399837732315063,
      "learning_rate": 2.7231638418079097e-05,
      "loss": 1.0164,
      "step": 18140
    },
    {
      "epoch": 1.4590032154340835,
      "grad_norm": 2.877127170562744,
      "learning_rate": 2.719128329297821e-05,
      "loss": 1.0546,
      "step": 18150
    },
    {
      "epoch": 1.459807073954984,
      "grad_norm": 2.673527717590332,
      "learning_rate": 2.715092816787732e-05,
      "loss": 0.9971,
      "step": 18160
    },
    {
      "epoch": 1.4606109324758842,
      "grad_norm": 1.6876904964447021,
      "learning_rate": 2.711057304277643e-05,
      "loss": 0.991,
      "step": 18170
    },
    {
      "epoch": 1.4614147909967845,
      "grad_norm": 2.3266868591308594,
      "learning_rate": 2.7070217917675543e-05,
      "loss": 0.9961,
      "step": 18180
    },
    {
      "epoch": 1.462218649517685,
      "grad_norm": 4.623996734619141,
      "learning_rate": 2.7029862792574658e-05,
      "loss": 0.9838,
      "step": 18190
    },
    {
      "epoch": 1.4630225080385852,
      "grad_norm": 1.4012879133224487,
      "learning_rate": 2.698950766747377e-05,
      "loss": 1.0015,
      "step": 18200
    },
    {
      "epoch": 1.4638263665594855,
      "grad_norm": 1.4820497035980225,
      "learning_rate": 2.6949152542372884e-05,
      "loss": 1.0434,
      "step": 18210
    },
    {
      "epoch": 1.464630225080386,
      "grad_norm": 1.5504990816116333,
      "learning_rate": 2.6908797417272e-05,
      "loss": 0.9886,
      "step": 18220
    },
    {
      "epoch": 1.4654340836012862,
      "grad_norm": 1.4550745487213135,
      "learning_rate": 2.686844229217111e-05,
      "loss": 1.0432,
      "step": 18230
    },
    {
      "epoch": 1.4662379421221865,
      "grad_norm": 1.790533423423767,
      "learning_rate": 2.682808716707022e-05,
      "loss": 0.9935,
      "step": 18240
    },
    {
      "epoch": 1.4670418006430868,
      "grad_norm": 2.5110254287719727,
      "learning_rate": 2.6787732041969332e-05,
      "loss": 0.9956,
      "step": 18250
    },
    {
      "epoch": 1.467845659163987,
      "grad_norm": 1.7681093215942383,
      "learning_rate": 2.6747376916868444e-05,
      "loss": 1.0096,
      "step": 18260
    },
    {
      "epoch": 1.4686495176848875,
      "grad_norm": 2.1158554553985596,
      "learning_rate": 2.6707021791767555e-05,
      "loss": 1.0045,
      "step": 18270
    },
    {
      "epoch": 1.4694533762057878,
      "grad_norm": 1.3785464763641357,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.9776,
      "step": 18280
    },
    {
      "epoch": 1.470257234726688,
      "grad_norm": 3.132347583770752,
      "learning_rate": 2.6626311541565778e-05,
      "loss": 0.9183,
      "step": 18290
    },
    {
      "epoch": 1.4710610932475885,
      "grad_norm": 1.5687378644943237,
      "learning_rate": 2.6585956416464893e-05,
      "loss": 1.0263,
      "step": 18300
    },
    {
      "epoch": 1.4718649517684887,
      "grad_norm": 1.2770296335220337,
      "learning_rate": 2.6545601291364004e-05,
      "loss": 0.8838,
      "step": 18310
    },
    {
      "epoch": 1.472668810289389,
      "grad_norm": 1.7857643365859985,
      "learning_rate": 2.6505246166263115e-05,
      "loss": 0.9491,
      "step": 18320
    },
    {
      "epoch": 1.4734726688102895,
      "grad_norm": 2.11606502532959,
      "learning_rate": 2.6464891041162227e-05,
      "loss": 1.1035,
      "step": 18330
    },
    {
      "epoch": 1.4742765273311897,
      "grad_norm": 1.88895583152771,
      "learning_rate": 2.6424535916061345e-05,
      "loss": 1.0237,
      "step": 18340
    },
    {
      "epoch": 1.47508038585209,
      "grad_norm": 2.3139593601226807,
      "learning_rate": 2.6384180790960456e-05,
      "loss": 1.0029,
      "step": 18350
    },
    {
      "epoch": 1.4758842443729905,
      "grad_norm": 1.4084957838058472,
      "learning_rate": 2.6343825665859568e-05,
      "loss": 0.9766,
      "step": 18360
    },
    {
      "epoch": 1.4766881028938907,
      "grad_norm": 1.5821465253829956,
      "learning_rate": 2.630347054075868e-05,
      "loss": 1.0581,
      "step": 18370
    },
    {
      "epoch": 1.477491961414791,
      "grad_norm": 1.7947590351104736,
      "learning_rate": 2.626311541565779e-05,
      "loss": 0.8901,
      "step": 18380
    },
    {
      "epoch": 1.4782958199356913,
      "grad_norm": 2.6382060050964355,
      "learning_rate": 2.6222760290556902e-05,
      "loss": 1.0321,
      "step": 18390
    },
    {
      "epoch": 1.4790996784565915,
      "grad_norm": 1.636795997619629,
      "learning_rate": 2.6182405165456013e-05,
      "loss": 1.073,
      "step": 18400
    },
    {
      "epoch": 1.479903536977492,
      "grad_norm": 1.451741337776184,
      "learning_rate": 2.6142050040355124e-05,
      "loss": 0.9504,
      "step": 18410
    },
    {
      "epoch": 1.4807073954983923,
      "grad_norm": 1.6076511144638062,
      "learning_rate": 2.610169491525424e-05,
      "loss": 1.0157,
      "step": 18420
    },
    {
      "epoch": 1.4815112540192925,
      "grad_norm": 1.6737004518508911,
      "learning_rate": 2.606133979015335e-05,
      "loss": 0.9674,
      "step": 18430
    },
    {
      "epoch": 1.482315112540193,
      "grad_norm": 1.4484913349151611,
      "learning_rate": 2.6020984665052462e-05,
      "loss": 0.9444,
      "step": 18440
    },
    {
      "epoch": 1.4831189710610932,
      "grad_norm": 1.8022609949111938,
      "learning_rate": 2.5980629539951573e-05,
      "loss": 1.045,
      "step": 18450
    },
    {
      "epoch": 1.4839228295819935,
      "grad_norm": 1.4874672889709473,
      "learning_rate": 2.5940274414850685e-05,
      "loss": 0.9895,
      "step": 18460
    },
    {
      "epoch": 1.484726688102894,
      "grad_norm": 1.8093199729919434,
      "learning_rate": 2.5899919289749803e-05,
      "loss": 1.0687,
      "step": 18470
    },
    {
      "epoch": 1.4855305466237942,
      "grad_norm": 2.6661243438720703,
      "learning_rate": 2.5859564164648914e-05,
      "loss": 1.1001,
      "step": 18480
    },
    {
      "epoch": 1.4863344051446945,
      "grad_norm": 2.4542813301086426,
      "learning_rate": 2.5819209039548026e-05,
      "loss": 1.0176,
      "step": 18490
    },
    {
      "epoch": 1.487138263665595,
      "grad_norm": 1.6964333057403564,
      "learning_rate": 2.5778853914447137e-05,
      "loss": 0.8609,
      "step": 18500
    },
    {
      "epoch": 1.4879421221864952,
      "grad_norm": 1.7633438110351562,
      "learning_rate": 2.573849878934625e-05,
      "loss": 1.0015,
      "step": 18510
    },
    {
      "epoch": 1.4887459807073955,
      "grad_norm": 1.8267382383346558,
      "learning_rate": 2.569814366424536e-05,
      "loss": 1.1222,
      "step": 18520
    },
    {
      "epoch": 1.4895498392282958,
      "grad_norm": 2.0830318927764893,
      "learning_rate": 2.5657788539144474e-05,
      "loss": 1.1318,
      "step": 18530
    },
    {
      "epoch": 1.490353697749196,
      "grad_norm": 1.4734503030776978,
      "learning_rate": 2.5617433414043586e-05,
      "loss": 0.9939,
      "step": 18540
    },
    {
      "epoch": 1.4911575562700965,
      "grad_norm": 2.606203079223633,
      "learning_rate": 2.5577078288942697e-05,
      "loss": 1.0286,
      "step": 18550
    },
    {
      "epoch": 1.4919614147909968,
      "grad_norm": 3.1389873027801514,
      "learning_rate": 2.553672316384181e-05,
      "loss": 1.0314,
      "step": 18560
    },
    {
      "epoch": 1.492765273311897,
      "grad_norm": 1.8032082319259644,
      "learning_rate": 2.549636803874092e-05,
      "loss": 1.0433,
      "step": 18570
    },
    {
      "epoch": 1.4935691318327975,
      "grad_norm": 2.33467173576355,
      "learning_rate": 2.545601291364003e-05,
      "loss": 1.0446,
      "step": 18580
    },
    {
      "epoch": 1.4943729903536977,
      "grad_norm": 2.114933490753174,
      "learning_rate": 2.5415657788539143e-05,
      "loss": 1.0179,
      "step": 18590
    },
    {
      "epoch": 1.495176848874598,
      "grad_norm": 1.1957430839538574,
      "learning_rate": 2.5375302663438254e-05,
      "loss": 1.0259,
      "step": 18600
    },
    {
      "epoch": 1.4959807073954985,
      "grad_norm": 1.8650671243667603,
      "learning_rate": 2.5334947538337372e-05,
      "loss": 0.934,
      "step": 18610
    },
    {
      "epoch": 1.4967845659163987,
      "grad_norm": 1.485527753829956,
      "learning_rate": 2.5294592413236483e-05,
      "loss": 1.0833,
      "step": 18620
    },
    {
      "epoch": 1.497588424437299,
      "grad_norm": 1.4764578342437744,
      "learning_rate": 2.5254237288135595e-05,
      "loss": 0.9306,
      "step": 18630
    },
    {
      "epoch": 1.4983922829581995,
      "grad_norm": 2.014840602874756,
      "learning_rate": 2.5213882163034706e-05,
      "loss": 0.9984,
      "step": 18640
    },
    {
      "epoch": 1.4991961414790997,
      "grad_norm": 1.6000443696975708,
      "learning_rate": 2.517352703793382e-05,
      "loss": 1.0134,
      "step": 18650
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7050846815109253,
      "learning_rate": 2.5133171912832932e-05,
      "loss": 0.9482,
      "step": 18660
    },
    {
      "epoch": 1.5008038585209005,
      "grad_norm": 1.9964686632156372,
      "learning_rate": 2.5092816787732044e-05,
      "loss": 1.0569,
      "step": 18670
    },
    {
      "epoch": 1.5016077170418005,
      "grad_norm": 1.524829387664795,
      "learning_rate": 2.5052461662631155e-05,
      "loss": 1.126,
      "step": 18680
    },
    {
      "epoch": 1.502411575562701,
      "grad_norm": 1.630293607711792,
      "learning_rate": 2.5012106537530266e-05,
      "loss": 1.0917,
      "step": 18690
    },
    {
      "epoch": 1.5032154340836013,
      "grad_norm": 1.819430947303772,
      "learning_rate": 2.497175141242938e-05,
      "loss": 0.9468,
      "step": 18700
    },
    {
      "epoch": 1.5040192926045015,
      "grad_norm": 1.743780493736267,
      "learning_rate": 2.4931396287328493e-05,
      "loss": 0.9956,
      "step": 18710
    },
    {
      "epoch": 1.504823151125402,
      "grad_norm": 1.9177618026733398,
      "learning_rate": 2.4891041162227604e-05,
      "loss": 0.9998,
      "step": 18720
    },
    {
      "epoch": 1.5056270096463023,
      "grad_norm": 1.604160189628601,
      "learning_rate": 2.4850686037126715e-05,
      "loss": 0.9501,
      "step": 18730
    },
    {
      "epoch": 1.5064308681672025,
      "grad_norm": 2.4709489345550537,
      "learning_rate": 2.4810330912025827e-05,
      "loss": 1.013,
      "step": 18740
    },
    {
      "epoch": 1.507234726688103,
      "grad_norm": 2.3379924297332764,
      "learning_rate": 2.476997578692494e-05,
      "loss": 1.0388,
      "step": 18750
    },
    {
      "epoch": 1.5080385852090032,
      "grad_norm": 1.7188684940338135,
      "learning_rate": 2.4729620661824053e-05,
      "loss": 1.03,
      "step": 18760
    },
    {
      "epoch": 1.5088424437299035,
      "grad_norm": 3.812397003173828,
      "learning_rate": 2.4689265536723168e-05,
      "loss": 1.0547,
      "step": 18770
    },
    {
      "epoch": 1.509646302250804,
      "grad_norm": 1.712337851524353,
      "learning_rate": 2.464891041162228e-05,
      "loss": 1.0071,
      "step": 18780
    },
    {
      "epoch": 1.510450160771704,
      "grad_norm": 2.7004611492156982,
      "learning_rate": 2.460855528652139e-05,
      "loss": 1.1183,
      "step": 18790
    },
    {
      "epoch": 1.5112540192926045,
      "grad_norm": 1.4686659574508667,
      "learning_rate": 2.45682001614205e-05,
      "loss": 0.9288,
      "step": 18800
    },
    {
      "epoch": 1.512057877813505,
      "grad_norm": 1.755505919456482,
      "learning_rate": 2.4527845036319613e-05,
      "loss": 1.0728,
      "step": 18810
    },
    {
      "epoch": 1.512861736334405,
      "grad_norm": 4.083408832550049,
      "learning_rate": 2.4487489911218724e-05,
      "loss": 0.9492,
      "step": 18820
    },
    {
      "epoch": 1.5136655948553055,
      "grad_norm": 1.7185946702957153,
      "learning_rate": 2.4447134786117836e-05,
      "loss": 0.8437,
      "step": 18830
    },
    {
      "epoch": 1.5144694533762058,
      "grad_norm": 1.4967041015625,
      "learning_rate": 2.440677966101695e-05,
      "loss": 0.9956,
      "step": 18840
    },
    {
      "epoch": 1.515273311897106,
      "grad_norm": 1.3148831129074097,
      "learning_rate": 2.4366424535916062e-05,
      "loss": 0.9735,
      "step": 18850
    },
    {
      "epoch": 1.5160771704180065,
      "grad_norm": 1.886210322380066,
      "learning_rate": 2.4326069410815173e-05,
      "loss": 1.0393,
      "step": 18860
    },
    {
      "epoch": 1.5168810289389068,
      "grad_norm": 1.3368439674377441,
      "learning_rate": 2.4285714285714288e-05,
      "loss": 0.9677,
      "step": 18870
    },
    {
      "epoch": 1.517684887459807,
      "grad_norm": 1.8139729499816895,
      "learning_rate": 2.42453591606134e-05,
      "loss": 1.0196,
      "step": 18880
    },
    {
      "epoch": 1.5184887459807075,
      "grad_norm": 2.2600748538970947,
      "learning_rate": 2.420500403551251e-05,
      "loss": 1.0883,
      "step": 18890
    },
    {
      "epoch": 1.5192926045016077,
      "grad_norm": 2.071951389312744,
      "learning_rate": 2.4164648910411625e-05,
      "loss": 0.9809,
      "step": 18900
    },
    {
      "epoch": 1.520096463022508,
      "grad_norm": 3.309037685394287,
      "learning_rate": 2.4124293785310737e-05,
      "loss": 0.9566,
      "step": 18910
    },
    {
      "epoch": 1.5209003215434085,
      "grad_norm": 1.5997929573059082,
      "learning_rate": 2.4083938660209848e-05,
      "loss": 1.0077,
      "step": 18920
    },
    {
      "epoch": 1.5217041800643085,
      "grad_norm": 2.123086929321289,
      "learning_rate": 2.404358353510896e-05,
      "loss": 1.2028,
      "step": 18930
    },
    {
      "epoch": 1.522508038585209,
      "grad_norm": 2.219896078109741,
      "learning_rate": 2.400322841000807e-05,
      "loss": 1.0008,
      "step": 18940
    },
    {
      "epoch": 1.5233118971061095,
      "grad_norm": 1.7317404747009277,
      "learning_rate": 2.3962873284907182e-05,
      "loss": 1.1455,
      "step": 18950
    },
    {
      "epoch": 1.5241157556270095,
      "grad_norm": 1.5645885467529297,
      "learning_rate": 2.3922518159806294e-05,
      "loss": 0.9871,
      "step": 18960
    },
    {
      "epoch": 1.52491961414791,
      "grad_norm": 1.4457658529281616,
      "learning_rate": 2.388216303470541e-05,
      "loss": 1.0322,
      "step": 18970
    },
    {
      "epoch": 1.5257234726688103,
      "grad_norm": 1.5804071426391602,
      "learning_rate": 2.3841807909604523e-05,
      "loss": 1.0548,
      "step": 18980
    },
    {
      "epoch": 1.5265273311897105,
      "grad_norm": 1.545935869216919,
      "learning_rate": 2.3801452784503634e-05,
      "loss": 1.0185,
      "step": 18990
    },
    {
      "epoch": 1.527331189710611,
      "grad_norm": 2.3571431636810303,
      "learning_rate": 2.3761097659402746e-05,
      "loss": 1.0888,
      "step": 19000
    },
    {
      "epoch": 1.5281350482315113,
      "grad_norm": 1.867585301399231,
      "learning_rate": 2.3720742534301857e-05,
      "loss": 0.9695,
      "step": 19010
    },
    {
      "epoch": 1.5289389067524115,
      "grad_norm": 2.631490468978882,
      "learning_rate": 2.368038740920097e-05,
      "loss": 0.9966,
      "step": 19020
    },
    {
      "epoch": 1.529742765273312,
      "grad_norm": 1.501379370689392,
      "learning_rate": 2.3640032284100083e-05,
      "loss": 0.9514,
      "step": 19030
    },
    {
      "epoch": 1.5305466237942122,
      "grad_norm": 1.7019522190093994,
      "learning_rate": 2.3599677158999195e-05,
      "loss": 1.0341,
      "step": 19040
    },
    {
      "epoch": 1.5313504823151125,
      "grad_norm": 2.4788031578063965,
      "learning_rate": 2.3559322033898306e-05,
      "loss": 0.9932,
      "step": 19050
    },
    {
      "epoch": 1.532154340836013,
      "grad_norm": 1.5615988969802856,
      "learning_rate": 2.3518966908797417e-05,
      "loss": 1.016,
      "step": 19060
    },
    {
      "epoch": 1.532958199356913,
      "grad_norm": 1.50218665599823,
      "learning_rate": 2.347861178369653e-05,
      "loss": 1.0888,
      "step": 19070
    },
    {
      "epoch": 1.5337620578778135,
      "grad_norm": 1.6570556163787842,
      "learning_rate": 2.3438256658595644e-05,
      "loss": 0.8625,
      "step": 19080
    },
    {
      "epoch": 1.534565916398714,
      "grad_norm": 2.3664464950561523,
      "learning_rate": 2.3397901533494755e-05,
      "loss": 1.0421,
      "step": 19090
    },
    {
      "epoch": 1.535369774919614,
      "grad_norm": 2.178680181503296,
      "learning_rate": 2.335754640839387e-05,
      "loss": 1.017,
      "step": 19100
    },
    {
      "epoch": 1.5361736334405145,
      "grad_norm": 1.6579536199569702,
      "learning_rate": 2.331719128329298e-05,
      "loss": 1.046,
      "step": 19110
    },
    {
      "epoch": 1.5369774919614148,
      "grad_norm": 3.513983726501465,
      "learning_rate": 2.3276836158192092e-05,
      "loss": 1.0616,
      "step": 19120
    },
    {
      "epoch": 1.537781350482315,
      "grad_norm": 2.0785059928894043,
      "learning_rate": 2.3236481033091204e-05,
      "loss": 0.9541,
      "step": 19130
    },
    {
      "epoch": 1.5385852090032155,
      "grad_norm": 1.9155325889587402,
      "learning_rate": 2.3196125907990315e-05,
      "loss": 1.0438,
      "step": 19140
    },
    {
      "epoch": 1.5393890675241158,
      "grad_norm": 2.0071442127227783,
      "learning_rate": 2.3155770782889426e-05,
      "loss": 0.9538,
      "step": 19150
    },
    {
      "epoch": 1.540192926045016,
      "grad_norm": 2.359255313873291,
      "learning_rate": 2.3115415657788538e-05,
      "loss": 0.8942,
      "step": 19160
    },
    {
      "epoch": 1.5409967845659165,
      "grad_norm": 1.5436384677886963,
      "learning_rate": 2.3075060532687653e-05,
      "loss": 0.9818,
      "step": 19170
    },
    {
      "epoch": 1.5418006430868167,
      "grad_norm": 1.5936756134033203,
      "learning_rate": 2.3034705407586764e-05,
      "loss": 1.0091,
      "step": 19180
    },
    {
      "epoch": 1.542604501607717,
      "grad_norm": 2.787342071533203,
      "learning_rate": 2.2994350282485875e-05,
      "loss": 1.0598,
      "step": 19190
    },
    {
      "epoch": 1.5434083601286175,
      "grad_norm": 2.2612195014953613,
      "learning_rate": 2.295399515738499e-05,
      "loss": 1.1404,
      "step": 19200
    },
    {
      "epoch": 1.5442122186495175,
      "grad_norm": 3.06294584274292,
      "learning_rate": 2.29136400322841e-05,
      "loss": 1.1383,
      "step": 19210
    },
    {
      "epoch": 1.545016077170418,
      "grad_norm": 2.707146644592285,
      "learning_rate": 2.2873284907183213e-05,
      "loss": 0.9306,
      "step": 19220
    },
    {
      "epoch": 1.5458199356913185,
      "grad_norm": 1.7285178899765015,
      "learning_rate": 2.2832929782082328e-05,
      "loss": 1.0923,
      "step": 19230
    },
    {
      "epoch": 1.5466237942122185,
      "grad_norm": 1.8518304824829102,
      "learning_rate": 2.279257465698144e-05,
      "loss": 1.0411,
      "step": 19240
    },
    {
      "epoch": 1.547427652733119,
      "grad_norm": 2.2175393104553223,
      "learning_rate": 2.275221953188055e-05,
      "loss": 1.0426,
      "step": 19250
    },
    {
      "epoch": 1.5482315112540193,
      "grad_norm": 1.9369691610336304,
      "learning_rate": 2.271186440677966e-05,
      "loss": 0.952,
      "step": 19260
    },
    {
      "epoch": 1.5490353697749195,
      "grad_norm": 2.876096725463867,
      "learning_rate": 2.2671509281678773e-05,
      "loss": 1.0326,
      "step": 19270
    },
    {
      "epoch": 1.54983922829582,
      "grad_norm": 1.7443819046020508,
      "learning_rate": 2.2631154156577884e-05,
      "loss": 1.1068,
      "step": 19280
    },
    {
      "epoch": 1.5506430868167203,
      "grad_norm": 1.8379837274551392,
      "learning_rate": 2.2590799031476996e-05,
      "loss": 1.0209,
      "step": 19290
    },
    {
      "epoch": 1.5514469453376205,
      "grad_norm": 1.8857988119125366,
      "learning_rate": 2.255044390637611e-05,
      "loss": 1.1399,
      "step": 19300
    },
    {
      "epoch": 1.552250803858521,
      "grad_norm": 1.553567886352539,
      "learning_rate": 2.2510088781275225e-05,
      "loss": 0.9507,
      "step": 19310
    },
    {
      "epoch": 1.5530546623794212,
      "grad_norm": 1.7635307312011719,
      "learning_rate": 2.2469733656174337e-05,
      "loss": 0.9853,
      "step": 19320
    },
    {
      "epoch": 1.5538585209003215,
      "grad_norm": 2.045858860015869,
      "learning_rate": 2.2429378531073448e-05,
      "loss": 0.9357,
      "step": 19330
    },
    {
      "epoch": 1.554662379421222,
      "grad_norm": 1.603248953819275,
      "learning_rate": 2.238902340597256e-05,
      "loss": 0.9685,
      "step": 19340
    },
    {
      "epoch": 1.555466237942122,
      "grad_norm": 2.1176679134368896,
      "learning_rate": 2.234866828087167e-05,
      "loss": 1.1459,
      "step": 19350
    },
    {
      "epoch": 1.5562700964630225,
      "grad_norm": 1.4296454191207886,
      "learning_rate": 2.2308313155770785e-05,
      "loss": 1.0233,
      "step": 19360
    },
    {
      "epoch": 1.557073954983923,
      "grad_norm": 1.4487553834915161,
      "learning_rate": 2.2267958030669897e-05,
      "loss": 1.0689,
      "step": 19370
    },
    {
      "epoch": 1.557877813504823,
      "grad_norm": 1.4253681898117065,
      "learning_rate": 2.2227602905569008e-05,
      "loss": 1.0238,
      "step": 19380
    },
    {
      "epoch": 1.5586816720257235,
      "grad_norm": 2.9197847843170166,
      "learning_rate": 2.218724778046812e-05,
      "loss": 0.9624,
      "step": 19390
    },
    {
      "epoch": 1.5594855305466238,
      "grad_norm": 1.4034744501113892,
      "learning_rate": 2.214689265536723e-05,
      "loss": 1.1532,
      "step": 19400
    },
    {
      "epoch": 1.560289389067524,
      "grad_norm": 1.631948709487915,
      "learning_rate": 2.2106537530266346e-05,
      "loss": 1.0096,
      "step": 19410
    },
    {
      "epoch": 1.5610932475884245,
      "grad_norm": 1.686239242553711,
      "learning_rate": 2.2066182405165457e-05,
      "loss": 0.9291,
      "step": 19420
    },
    {
      "epoch": 1.5618971061093248,
      "grad_norm": 1.7068378925323486,
      "learning_rate": 2.2025827280064572e-05,
      "loss": 1.0057,
      "step": 19430
    },
    {
      "epoch": 1.562700964630225,
      "grad_norm": 2.1420505046844482,
      "learning_rate": 2.1985472154963683e-05,
      "loss": 1.0527,
      "step": 19440
    },
    {
      "epoch": 1.5635048231511255,
      "grad_norm": 1.7549521923065186,
      "learning_rate": 2.1945117029862795e-05,
      "loss": 1.0262,
      "step": 19450
    },
    {
      "epoch": 1.5643086816720257,
      "grad_norm": 2.7982192039489746,
      "learning_rate": 2.1904761904761906e-05,
      "loss": 0.9992,
      "step": 19460
    },
    {
      "epoch": 1.565112540192926,
      "grad_norm": 1.7734447717666626,
      "learning_rate": 2.1864406779661017e-05,
      "loss": 0.9371,
      "step": 19470
    },
    {
      "epoch": 1.5659163987138265,
      "grad_norm": 2.7094318866729736,
      "learning_rate": 2.182405165456013e-05,
      "loss": 1.0623,
      "step": 19480
    },
    {
      "epoch": 1.5667202572347267,
      "grad_norm": 2.0985891819000244,
      "learning_rate": 2.178369652945924e-05,
      "loss": 1.1057,
      "step": 19490
    },
    {
      "epoch": 1.567524115755627,
      "grad_norm": 2.8843741416931152,
      "learning_rate": 2.1743341404358355e-05,
      "loss": 1.089,
      "step": 19500
    },
    {
      "epoch": 1.5683279742765275,
      "grad_norm": 2.498589038848877,
      "learning_rate": 2.1702986279257466e-05,
      "loss": 0.9217,
      "step": 19510
    },
    {
      "epoch": 1.5691318327974275,
      "grad_norm": 3.109459400177002,
      "learning_rate": 2.1662631154156577e-05,
      "loss": 0.9486,
      "step": 19520
    },
    {
      "epoch": 1.569935691318328,
      "grad_norm": 1.6474716663360596,
      "learning_rate": 2.1622276029055692e-05,
      "loss": 0.9679,
      "step": 19530
    },
    {
      "epoch": 1.5707395498392283,
      "grad_norm": 2.079991102218628,
      "learning_rate": 2.1581920903954804e-05,
      "loss": 0.9412,
      "step": 19540
    },
    {
      "epoch": 1.5715434083601285,
      "grad_norm": 2.6996138095855713,
      "learning_rate": 2.1541565778853915e-05,
      "loss": 0.881,
      "step": 19550
    },
    {
      "epoch": 1.572347266881029,
      "grad_norm": 1.8078988790512085,
      "learning_rate": 2.150121065375303e-05,
      "loss": 1.062,
      "step": 19560
    },
    {
      "epoch": 1.5731511254019293,
      "grad_norm": 1.8167033195495605,
      "learning_rate": 2.146085552865214e-05,
      "loss": 1.0287,
      "step": 19570
    },
    {
      "epoch": 1.5739549839228295,
      "grad_norm": 1.9593679904937744,
      "learning_rate": 2.1420500403551252e-05,
      "loss": 1.0442,
      "step": 19580
    },
    {
      "epoch": 1.57475884244373,
      "grad_norm": 2.62164568901062,
      "learning_rate": 2.1380145278450364e-05,
      "loss": 1.0162,
      "step": 19590
    },
    {
      "epoch": 1.5755627009646302,
      "grad_norm": 1.6859263181686401,
      "learning_rate": 2.1339790153349475e-05,
      "loss": 0.8896,
      "step": 19600
    },
    {
      "epoch": 1.5763665594855305,
      "grad_norm": 2.3182373046875,
      "learning_rate": 2.1299435028248587e-05,
      "loss": 0.9433,
      "step": 19610
    },
    {
      "epoch": 1.577170418006431,
      "grad_norm": 2.2859318256378174,
      "learning_rate": 2.1259079903147698e-05,
      "loss": 1.0641,
      "step": 19620
    },
    {
      "epoch": 1.5779742765273312,
      "grad_norm": 4.590075492858887,
      "learning_rate": 2.1218724778046813e-05,
      "loss": 1.0006,
      "step": 19630
    },
    {
      "epoch": 1.5787781350482315,
      "grad_norm": 1.6839805841445923,
      "learning_rate": 2.1178369652945927e-05,
      "loss": 1.1331,
      "step": 19640
    },
    {
      "epoch": 1.579581993569132,
      "grad_norm": 1.8310493230819702,
      "learning_rate": 2.113801452784504e-05,
      "loss": 1.1686,
      "step": 19650
    },
    {
      "epoch": 1.580385852090032,
      "grad_norm": 1.4298887252807617,
      "learning_rate": 2.109765940274415e-05,
      "loss": 0.9636,
      "step": 19660
    },
    {
      "epoch": 1.5811897106109325,
      "grad_norm": 2.003631353378296,
      "learning_rate": 2.105730427764326e-05,
      "loss": 0.8991,
      "step": 19670
    },
    {
      "epoch": 1.5819935691318328,
      "grad_norm": 1.770169973373413,
      "learning_rate": 2.1016949152542373e-05,
      "loss": 1.0196,
      "step": 19680
    },
    {
      "epoch": 1.582797427652733,
      "grad_norm": 3.1248345375061035,
      "learning_rate": 2.0976594027441488e-05,
      "loss": 1.057,
      "step": 19690
    },
    {
      "epoch": 1.5836012861736335,
      "grad_norm": 1.6270357370376587,
      "learning_rate": 2.09362389023406e-05,
      "loss": 1.0216,
      "step": 19700
    },
    {
      "epoch": 1.5844051446945338,
      "grad_norm": 2.0626096725463867,
      "learning_rate": 2.089588377723971e-05,
      "loss": 1.0524,
      "step": 19710
    },
    {
      "epoch": 1.585209003215434,
      "grad_norm": 2.2830660343170166,
      "learning_rate": 2.0855528652138822e-05,
      "loss": 1.0987,
      "step": 19720
    },
    {
      "epoch": 1.5860128617363345,
      "grad_norm": 1.951399803161621,
      "learning_rate": 2.0815173527037933e-05,
      "loss": 0.907,
      "step": 19730
    },
    {
      "epoch": 1.5868167202572347,
      "grad_norm": 3.4775824546813965,
      "learning_rate": 2.0774818401937048e-05,
      "loss": 1.0355,
      "step": 19740
    },
    {
      "epoch": 1.587620578778135,
      "grad_norm": 1.6632407903671265,
      "learning_rate": 2.073446327683616e-05,
      "loss": 1.0947,
      "step": 19750
    },
    {
      "epoch": 1.5884244372990355,
      "grad_norm": 1.8771129846572876,
      "learning_rate": 2.0694108151735274e-05,
      "loss": 0.993,
      "step": 19760
    },
    {
      "epoch": 1.5892282958199357,
      "grad_norm": 1.9345898628234863,
      "learning_rate": 2.0653753026634385e-05,
      "loss": 0.9974,
      "step": 19770
    },
    {
      "epoch": 1.590032154340836,
      "grad_norm": 1.9626299142837524,
      "learning_rate": 2.0613397901533497e-05,
      "loss": 1.053,
      "step": 19780
    },
    {
      "epoch": 1.5908360128617365,
      "grad_norm": 1.9407083988189697,
      "learning_rate": 2.0573042776432608e-05,
      "loss": 0.9735,
      "step": 19790
    },
    {
      "epoch": 1.5916398713826365,
      "grad_norm": 1.5738252401351929,
      "learning_rate": 2.053268765133172e-05,
      "loss": 0.9544,
      "step": 19800
    },
    {
      "epoch": 1.592443729903537,
      "grad_norm": 1.4336590766906738,
      "learning_rate": 2.049233252623083e-05,
      "loss": 0.9506,
      "step": 19810
    },
    {
      "epoch": 1.5932475884244373,
      "grad_norm": 2.0407192707061768,
      "learning_rate": 2.0451977401129946e-05,
      "loss": 1.0241,
      "step": 19820
    },
    {
      "epoch": 1.5940514469453375,
      "grad_norm": 1.9290578365325928,
      "learning_rate": 2.0411622276029057e-05,
      "loss": 0.9519,
      "step": 19830
    },
    {
      "epoch": 1.594855305466238,
      "grad_norm": 1.5437543392181396,
      "learning_rate": 2.0371267150928168e-05,
      "loss": 0.9266,
      "step": 19840
    },
    {
      "epoch": 1.5956591639871383,
      "grad_norm": 1.6253694295883179,
      "learning_rate": 2.033091202582728e-05,
      "loss": 1.0576,
      "step": 19850
    },
    {
      "epoch": 1.5964630225080385,
      "grad_norm": 2.7759857177734375,
      "learning_rate": 2.0290556900726394e-05,
      "loss": 1.0464,
      "step": 19860
    },
    {
      "epoch": 1.597266881028939,
      "grad_norm": 1.6483947038650513,
      "learning_rate": 2.0250201775625506e-05,
      "loss": 0.9771,
      "step": 19870
    },
    {
      "epoch": 1.5980707395498392,
      "grad_norm": 1.52219820022583,
      "learning_rate": 2.0209846650524617e-05,
      "loss": 0.9558,
      "step": 19880
    },
    {
      "epoch": 1.5988745980707395,
      "grad_norm": 1.429937481880188,
      "learning_rate": 2.0169491525423732e-05,
      "loss": 0.95,
      "step": 19890
    },
    {
      "epoch": 1.59967845659164,
      "grad_norm": 1.8432539701461792,
      "learning_rate": 2.0129136400322843e-05,
      "loss": 1.0402,
      "step": 19900
    },
    {
      "epoch": 1.6004823151125402,
      "grad_norm": 1.6091192960739136,
      "learning_rate": 2.0088781275221955e-05,
      "loss": 1.0194,
      "step": 19910
    },
    {
      "epoch": 1.6012861736334405,
      "grad_norm": 2.1835174560546875,
      "learning_rate": 2.0048426150121066e-05,
      "loss": 0.9036,
      "step": 19920
    },
    {
      "epoch": 1.602090032154341,
      "grad_norm": 1.876979947090149,
      "learning_rate": 2.0008071025020177e-05,
      "loss": 1.0052,
      "step": 19930
    },
    {
      "epoch": 1.602893890675241,
      "grad_norm": 1.3231323957443237,
      "learning_rate": 1.996771589991929e-05,
      "loss": 0.9938,
      "step": 19940
    },
    {
      "epoch": 1.6036977491961415,
      "grad_norm": 2.0215790271759033,
      "learning_rate": 1.99273607748184e-05,
      "loss": 1.0777,
      "step": 19950
    },
    {
      "epoch": 1.6045016077170418,
      "grad_norm": 1.4858729839324951,
      "learning_rate": 1.9887005649717515e-05,
      "loss": 0.9382,
      "step": 19960
    },
    {
      "epoch": 1.605305466237942,
      "grad_norm": 1.8996670246124268,
      "learning_rate": 1.984665052461663e-05,
      "loss": 1.0613,
      "step": 19970
    },
    {
      "epoch": 1.6061093247588425,
      "grad_norm": 1.6118154525756836,
      "learning_rate": 1.980629539951574e-05,
      "loss": 1.0106,
      "step": 19980
    },
    {
      "epoch": 1.6069131832797428,
      "grad_norm": 2.23201847076416,
      "learning_rate": 1.9765940274414852e-05,
      "loss": 0.982,
      "step": 19990
    },
    {
      "epoch": 1.607717041800643,
      "grad_norm": 2.0391416549682617,
      "learning_rate": 1.9725585149313964e-05,
      "loss": 1.0126,
      "step": 20000
    },
    {
      "epoch": 1.607717041800643,
      "eval_yahma/alpaca-cleaned_loss": 1.2087749242782593,
      "eval_yahma/alpaca-cleaned_runtime": 115.7363,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.281,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.16,
      "step": 20000
    },
    {
      "epoch": 1.6085209003215435,
      "grad_norm": 1.4210705757141113,
      "learning_rate": 1.9685230024213075e-05,
      "loss": 0.9249,
      "step": 20010
    },
    {
      "epoch": 1.6093247588424437,
      "grad_norm": 1.8334485292434692,
      "learning_rate": 1.964487489911219e-05,
      "loss": 1.1195,
      "step": 20020
    },
    {
      "epoch": 1.610128617363344,
      "grad_norm": 1.5518391132354736,
      "learning_rate": 1.96045197740113e-05,
      "loss": 0.9409,
      "step": 20030
    },
    {
      "epoch": 1.6109324758842445,
      "grad_norm": 1.3544491529464722,
      "learning_rate": 1.9564164648910413e-05,
      "loss": 0.9035,
      "step": 20040
    },
    {
      "epoch": 1.6117363344051447,
      "grad_norm": 1.4562795162200928,
      "learning_rate": 1.9523809523809524e-05,
      "loss": 1.0332,
      "step": 20050
    },
    {
      "epoch": 1.612540192926045,
      "grad_norm": 1.9066096544265747,
      "learning_rate": 1.9483454398708635e-05,
      "loss": 0.9654,
      "step": 20060
    },
    {
      "epoch": 1.6133440514469455,
      "grad_norm": 2.1256048679351807,
      "learning_rate": 1.944309927360775e-05,
      "loss": 1.1411,
      "step": 20070
    },
    {
      "epoch": 1.6141479099678455,
      "grad_norm": 2.309495449066162,
      "learning_rate": 1.940274414850686e-05,
      "loss": 0.9443,
      "step": 20080
    },
    {
      "epoch": 1.614951768488746,
      "grad_norm": 1.5459156036376953,
      "learning_rate": 1.9362389023405976e-05,
      "loss": 1.0025,
      "step": 20090
    },
    {
      "epoch": 1.6157556270096463,
      "grad_norm": 1.8760969638824463,
      "learning_rate": 1.9322033898305087e-05,
      "loss": 1.1442,
      "step": 20100
    },
    {
      "epoch": 1.6165594855305465,
      "grad_norm": 2.290050506591797,
      "learning_rate": 1.92816787732042e-05,
      "loss": 1.0269,
      "step": 20110
    },
    {
      "epoch": 1.617363344051447,
      "grad_norm": 1.5273579359054565,
      "learning_rate": 1.924132364810331e-05,
      "loss": 0.9352,
      "step": 20120
    },
    {
      "epoch": 1.6181672025723473,
      "grad_norm": 1.4741283655166626,
      "learning_rate": 1.920096852300242e-05,
      "loss": 0.962,
      "step": 20130
    },
    {
      "epoch": 1.6189710610932475,
      "grad_norm": 2.3171274662017822,
      "learning_rate": 1.9160613397901533e-05,
      "loss": 1.0589,
      "step": 20140
    },
    {
      "epoch": 1.619774919614148,
      "grad_norm": 2.144007444381714,
      "learning_rate": 1.9120258272800648e-05,
      "loss": 1.0362,
      "step": 20150
    },
    {
      "epoch": 1.6205787781350482,
      "grad_norm": 2.317404270172119,
      "learning_rate": 1.907990314769976e-05,
      "loss": 0.972,
      "step": 20160
    },
    {
      "epoch": 1.6213826366559485,
      "grad_norm": 1.6954485177993774,
      "learning_rate": 1.903954802259887e-05,
      "loss": 1.0767,
      "step": 20170
    },
    {
      "epoch": 1.622186495176849,
      "grad_norm": 1.7256394624710083,
      "learning_rate": 1.8999192897497982e-05,
      "loss": 1.0305,
      "step": 20180
    },
    {
      "epoch": 1.6229903536977492,
      "grad_norm": 1.4580116271972656,
      "learning_rate": 1.8958837772397097e-05,
      "loss": 0.9812,
      "step": 20190
    },
    {
      "epoch": 1.6237942122186495,
      "grad_norm": 2.0422894954681396,
      "learning_rate": 1.8918482647296208e-05,
      "loss": 0.9744,
      "step": 20200
    },
    {
      "epoch": 1.62459807073955,
      "grad_norm": 4.765537738800049,
      "learning_rate": 1.887812752219532e-05,
      "loss": 1.0139,
      "step": 20210
    },
    {
      "epoch": 1.62540192926045,
      "grad_norm": 2.389866828918457,
      "learning_rate": 1.8837772397094434e-05,
      "loss": 1.1453,
      "step": 20220
    },
    {
      "epoch": 1.6262057877813505,
      "grad_norm": 1.4161128997802734,
      "learning_rate": 1.8797417271993545e-05,
      "loss": 0.9437,
      "step": 20230
    },
    {
      "epoch": 1.6270096463022508,
      "grad_norm": 1.9139914512634277,
      "learning_rate": 1.8757062146892657e-05,
      "loss": 1.0214,
      "step": 20240
    },
    {
      "epoch": 1.627813504823151,
      "grad_norm": 1.738409161567688,
      "learning_rate": 1.8716707021791768e-05,
      "loss": 1.026,
      "step": 20250
    },
    {
      "epoch": 1.6286173633440515,
      "grad_norm": 2.470611333847046,
      "learning_rate": 1.867635189669088e-05,
      "loss": 1.0338,
      "step": 20260
    },
    {
      "epoch": 1.6294212218649518,
      "grad_norm": 1.3114768266677856,
      "learning_rate": 1.863599677158999e-05,
      "loss": 1.0561,
      "step": 20270
    },
    {
      "epoch": 1.630225080385852,
      "grad_norm": 1.7882541418075562,
      "learning_rate": 1.8595641646489102e-05,
      "loss": 1.0617,
      "step": 20280
    },
    {
      "epoch": 1.6310289389067525,
      "grad_norm": 2.458045244216919,
      "learning_rate": 1.8555286521388217e-05,
      "loss": 0.9396,
      "step": 20290
    },
    {
      "epoch": 1.6318327974276527,
      "grad_norm": 2.7625696659088135,
      "learning_rate": 1.851493139628733e-05,
      "loss": 0.9944,
      "step": 20300
    },
    {
      "epoch": 1.632636655948553,
      "grad_norm": 1.760848879814148,
      "learning_rate": 1.8474576271186443e-05,
      "loss": 0.9271,
      "step": 20310
    },
    {
      "epoch": 1.6334405144694535,
      "grad_norm": 1.9309755563735962,
      "learning_rate": 1.8434221146085554e-05,
      "loss": 1.0556,
      "step": 20320
    },
    {
      "epoch": 1.6342443729903537,
      "grad_norm": 2.2995901107788086,
      "learning_rate": 1.8393866020984666e-05,
      "loss": 1.0896,
      "step": 20330
    },
    {
      "epoch": 1.635048231511254,
      "grad_norm": 2.4556405544281006,
      "learning_rate": 1.8353510895883777e-05,
      "loss": 0.929,
      "step": 20340
    },
    {
      "epoch": 1.6358520900321545,
      "grad_norm": 2.014082431793213,
      "learning_rate": 1.8313155770782892e-05,
      "loss": 0.9454,
      "step": 20350
    },
    {
      "epoch": 1.6366559485530545,
      "grad_norm": 1.7907990217208862,
      "learning_rate": 1.8272800645682003e-05,
      "loss": 0.956,
      "step": 20360
    },
    {
      "epoch": 1.637459807073955,
      "grad_norm": 1.6757053136825562,
      "learning_rate": 1.8232445520581115e-05,
      "loss": 1.0474,
      "step": 20370
    },
    {
      "epoch": 1.6382636655948553,
      "grad_norm": 1.374757170677185,
      "learning_rate": 1.8192090395480226e-05,
      "loss": 0.8232,
      "step": 20380
    },
    {
      "epoch": 1.6390675241157555,
      "grad_norm": 1.8884340524673462,
      "learning_rate": 1.8151735270379337e-05,
      "loss": 0.9609,
      "step": 20390
    },
    {
      "epoch": 1.639871382636656,
      "grad_norm": 1.5904606580734253,
      "learning_rate": 1.811138014527845e-05,
      "loss": 0.9653,
      "step": 20400
    },
    {
      "epoch": 1.6406752411575563,
      "grad_norm": 1.5124174356460571,
      "learning_rate": 1.8071025020177564e-05,
      "loss": 1.0028,
      "step": 20410
    },
    {
      "epoch": 1.6414790996784565,
      "grad_norm": 2.074727773666382,
      "learning_rate": 1.8030669895076678e-05,
      "loss": 0.9784,
      "step": 20420
    },
    {
      "epoch": 1.642282958199357,
      "grad_norm": 3.159682035446167,
      "learning_rate": 1.799031476997579e-05,
      "loss": 1.0564,
      "step": 20430
    },
    {
      "epoch": 1.6430868167202572,
      "grad_norm": 2.477046012878418,
      "learning_rate": 1.79499596448749e-05,
      "loss": 0.9725,
      "step": 20440
    },
    {
      "epoch": 1.6438906752411575,
      "grad_norm": 1.718367576599121,
      "learning_rate": 1.7909604519774012e-05,
      "loss": 0.9811,
      "step": 20450
    },
    {
      "epoch": 1.644694533762058,
      "grad_norm": 2.1779701709747314,
      "learning_rate": 1.7869249394673124e-05,
      "loss": 0.9761,
      "step": 20460
    },
    {
      "epoch": 1.6454983922829582,
      "grad_norm": 2.076404094696045,
      "learning_rate": 1.7828894269572235e-05,
      "loss": 0.9694,
      "step": 20470
    },
    {
      "epoch": 1.6463022508038585,
      "grad_norm": 1.9192440509796143,
      "learning_rate": 1.778853914447135e-05,
      "loss": 1.0546,
      "step": 20480
    },
    {
      "epoch": 1.647106109324759,
      "grad_norm": 3.688655138015747,
      "learning_rate": 1.774818401937046e-05,
      "loss": 1.0202,
      "step": 20490
    },
    {
      "epoch": 1.647909967845659,
      "grad_norm": 1.932462215423584,
      "learning_rate": 1.7707828894269573e-05,
      "loss": 0.9977,
      "step": 20500
    },
    {
      "epoch": 1.6487138263665595,
      "grad_norm": 2.313615322113037,
      "learning_rate": 1.7667473769168684e-05,
      "loss": 1.0029,
      "step": 20510
    },
    {
      "epoch": 1.6495176848874598,
      "grad_norm": 1.4296419620513916,
      "learning_rate": 1.76271186440678e-05,
      "loss": 1.0456,
      "step": 20520
    },
    {
      "epoch": 1.65032154340836,
      "grad_norm": 2.292299747467041,
      "learning_rate": 1.758676351896691e-05,
      "loss": 1.0806,
      "step": 20530
    },
    {
      "epoch": 1.6511254019292605,
      "grad_norm": 2.147498369216919,
      "learning_rate": 1.754640839386602e-05,
      "loss": 1.0389,
      "step": 20540
    },
    {
      "epoch": 1.6519292604501608,
      "grad_norm": 1.8983690738677979,
      "learning_rate": 1.7506053268765136e-05,
      "loss": 0.9936,
      "step": 20550
    },
    {
      "epoch": 1.652733118971061,
      "grad_norm": 1.4978177547454834,
      "learning_rate": 1.7465698143664248e-05,
      "loss": 1.0733,
      "step": 20560
    },
    {
      "epoch": 1.6535369774919615,
      "grad_norm": 3.058656692504883,
      "learning_rate": 1.742534301856336e-05,
      "loss": 0.8716,
      "step": 20570
    },
    {
      "epoch": 1.6543408360128617,
      "grad_norm": 1.651357889175415,
      "learning_rate": 1.738498789346247e-05,
      "loss": 1.0162,
      "step": 20580
    },
    {
      "epoch": 1.655144694533762,
      "grad_norm": 1.9416139125823975,
      "learning_rate": 1.734463276836158e-05,
      "loss": 0.999,
      "step": 20590
    },
    {
      "epoch": 1.6559485530546625,
      "grad_norm": 2.939375400543213,
      "learning_rate": 1.7304277643260693e-05,
      "loss": 1.0727,
      "step": 20600
    },
    {
      "epoch": 1.6567524115755627,
      "grad_norm": 2.1564691066741943,
      "learning_rate": 1.7263922518159804e-05,
      "loss": 1.0074,
      "step": 20610
    },
    {
      "epoch": 1.657556270096463,
      "grad_norm": 2.003150701522827,
      "learning_rate": 1.722356739305892e-05,
      "loss": 1.1418,
      "step": 20620
    },
    {
      "epoch": 1.6583601286173635,
      "grad_norm": 1.420788049697876,
      "learning_rate": 1.718321226795803e-05,
      "loss": 0.8737,
      "step": 20630
    },
    {
      "epoch": 1.6591639871382635,
      "grad_norm": 1.5822371244430542,
      "learning_rate": 1.7142857142857145e-05,
      "loss": 0.9014,
      "step": 20640
    },
    {
      "epoch": 1.659967845659164,
      "grad_norm": 1.6281102895736694,
      "learning_rate": 1.7102502017756257e-05,
      "loss": 0.8962,
      "step": 20650
    },
    {
      "epoch": 1.6607717041800643,
      "grad_norm": 2.509737491607666,
      "learning_rate": 1.7062146892655368e-05,
      "loss": 0.9401,
      "step": 20660
    },
    {
      "epoch": 1.6615755627009645,
      "grad_norm": 1.9507536888122559,
      "learning_rate": 1.702179176755448e-05,
      "loss": 0.9771,
      "step": 20670
    },
    {
      "epoch": 1.662379421221865,
      "grad_norm": 2.431785821914673,
      "learning_rate": 1.6981436642453594e-05,
      "loss": 1.0511,
      "step": 20680
    },
    {
      "epoch": 1.6631832797427653,
      "grad_norm": 2.554717540740967,
      "learning_rate": 1.6941081517352705e-05,
      "loss": 0.9411,
      "step": 20690
    },
    {
      "epoch": 1.6639871382636655,
      "grad_norm": 2.0644285678863525,
      "learning_rate": 1.6900726392251817e-05,
      "loss": 0.9656,
      "step": 20700
    },
    {
      "epoch": 1.664790996784566,
      "grad_norm": 1.919416069984436,
      "learning_rate": 1.6860371267150928e-05,
      "loss": 0.8719,
      "step": 20710
    },
    {
      "epoch": 1.6655948553054662,
      "grad_norm": 1.4564859867095947,
      "learning_rate": 1.682001614205004e-05,
      "loss": 0.8866,
      "step": 20720
    },
    {
      "epoch": 1.6663987138263665,
      "grad_norm": 2.2104880809783936,
      "learning_rate": 1.677966101694915e-05,
      "loss": 1.0019,
      "step": 20730
    },
    {
      "epoch": 1.667202572347267,
      "grad_norm": 1.7495410442352295,
      "learning_rate": 1.6739305891848266e-05,
      "loss": 1.1196,
      "step": 20740
    },
    {
      "epoch": 1.6680064308681672,
      "grad_norm": 1.9178359508514404,
      "learning_rate": 1.669895076674738e-05,
      "loss": 1.0092,
      "step": 20750
    },
    {
      "epoch": 1.6688102893890675,
      "grad_norm": 2.948913812637329,
      "learning_rate": 1.6658595641646492e-05,
      "loss": 0.9096,
      "step": 20760
    },
    {
      "epoch": 1.669614147909968,
      "grad_norm": 1.9237521886825562,
      "learning_rate": 1.6618240516545603e-05,
      "loss": 1.0704,
      "step": 20770
    },
    {
      "epoch": 1.670418006430868,
      "grad_norm": 2.0535099506378174,
      "learning_rate": 1.6577885391444715e-05,
      "loss": 0.9703,
      "step": 20780
    },
    {
      "epoch": 1.6712218649517685,
      "grad_norm": 2.0139174461364746,
      "learning_rate": 1.6537530266343826e-05,
      "loss": 0.9179,
      "step": 20790
    },
    {
      "epoch": 1.6720257234726688,
      "grad_norm": 1.4897915124893188,
      "learning_rate": 1.6497175141242937e-05,
      "loss": 0.9509,
      "step": 20800
    },
    {
      "epoch": 1.672829581993569,
      "grad_norm": 2.042825937271118,
      "learning_rate": 1.6456820016142052e-05,
      "loss": 0.8686,
      "step": 20810
    },
    {
      "epoch": 1.6736334405144695,
      "grad_norm": 2.0794460773468018,
      "learning_rate": 1.6416464891041163e-05,
      "loss": 1.0376,
      "step": 20820
    },
    {
      "epoch": 1.6744372990353698,
      "grad_norm": 1.6484997272491455,
      "learning_rate": 1.6380145278450364e-05,
      "loss": 0.7965,
      "step": 20830
    },
    {
      "epoch": 1.67524115755627,
      "grad_norm": 1.6736949682235718,
      "learning_rate": 1.6339790153349476e-05,
      "loss": 0.9812,
      "step": 20840
    },
    {
      "epoch": 1.6760450160771705,
      "grad_norm": 1.5247830152511597,
      "learning_rate": 1.6299435028248587e-05,
      "loss": 0.9694,
      "step": 20850
    },
    {
      "epoch": 1.6768488745980707,
      "grad_norm": 2.0274441242218018,
      "learning_rate": 1.6259079903147702e-05,
      "loss": 1.0599,
      "step": 20860
    },
    {
      "epoch": 1.677652733118971,
      "grad_norm": 1.847269058227539,
      "learning_rate": 1.6218724778046813e-05,
      "loss": 0.9876,
      "step": 20870
    },
    {
      "epoch": 1.6784565916398715,
      "grad_norm": 2.002469301223755,
      "learning_rate": 1.6178369652945924e-05,
      "loss": 1.0208,
      "step": 20880
    },
    {
      "epoch": 1.6792604501607717,
      "grad_norm": 1.3857256174087524,
      "learning_rate": 1.613801452784504e-05,
      "loss": 1.0983,
      "step": 20890
    },
    {
      "epoch": 1.680064308681672,
      "grad_norm": 1.6506829261779785,
      "learning_rate": 1.609765940274415e-05,
      "loss": 0.9916,
      "step": 20900
    },
    {
      "epoch": 1.6808681672025725,
      "grad_norm": 1.6178910732269287,
      "learning_rate": 1.6057304277643262e-05,
      "loss": 0.8901,
      "step": 20910
    },
    {
      "epoch": 1.6816720257234725,
      "grad_norm": 2.434352397918701,
      "learning_rate": 1.6016949152542373e-05,
      "loss": 0.9237,
      "step": 20920
    },
    {
      "epoch": 1.682475884244373,
      "grad_norm": 1.6109217405319214,
      "learning_rate": 1.5976594027441485e-05,
      "loss": 1.1368,
      "step": 20930
    },
    {
      "epoch": 1.6832797427652733,
      "grad_norm": 2.161876678466797,
      "learning_rate": 1.5936238902340596e-05,
      "loss": 1.0398,
      "step": 20940
    },
    {
      "epoch": 1.6840836012861735,
      "grad_norm": 1.8207350969314575,
      "learning_rate": 1.5895883777239707e-05,
      "loss": 0.95,
      "step": 20950
    },
    {
      "epoch": 1.684887459807074,
      "grad_norm": 1.5055922269821167,
      "learning_rate": 1.5855528652138822e-05,
      "loss": 1.0249,
      "step": 20960
    },
    {
      "epoch": 1.6856913183279743,
      "grad_norm": 1.4368202686309814,
      "learning_rate": 1.5815173527037937e-05,
      "loss": 1.0758,
      "step": 20970
    },
    {
      "epoch": 1.6864951768488745,
      "grad_norm": 3.190652847290039,
      "learning_rate": 1.5774818401937048e-05,
      "loss": 1.1185,
      "step": 20980
    },
    {
      "epoch": 1.687299035369775,
      "grad_norm": 3.332230567932129,
      "learning_rate": 1.573446327683616e-05,
      "loss": 1.0188,
      "step": 20990
    },
    {
      "epoch": 1.6881028938906752,
      "grad_norm": 2.1496293544769287,
      "learning_rate": 1.569410815173527e-05,
      "loss": 1.0209,
      "step": 21000
    },
    {
      "epoch": 1.6889067524115755,
      "grad_norm": 1.6579251289367676,
      "learning_rate": 1.5653753026634382e-05,
      "loss": 0.9776,
      "step": 21010
    },
    {
      "epoch": 1.689710610932476,
      "grad_norm": 2.849315881729126,
      "learning_rate": 1.5613397901533494e-05,
      "loss": 0.9393,
      "step": 21020
    },
    {
      "epoch": 1.6905144694533762,
      "grad_norm": 1.442819595336914,
      "learning_rate": 1.557304277643261e-05,
      "loss": 1.192,
      "step": 21030
    },
    {
      "epoch": 1.6913183279742765,
      "grad_norm": 1.701747179031372,
      "learning_rate": 1.553268765133172e-05,
      "loss": 1.0754,
      "step": 21040
    },
    {
      "epoch": 1.692122186495177,
      "grad_norm": 1.8358919620513916,
      "learning_rate": 1.549233252623083e-05,
      "loss": 0.9335,
      "step": 21050
    },
    {
      "epoch": 1.692926045016077,
      "grad_norm": 2.1475460529327393,
      "learning_rate": 1.5451977401129943e-05,
      "loss": 0.897,
      "step": 21060
    },
    {
      "epoch": 1.6937299035369775,
      "grad_norm": 2.243940830230713,
      "learning_rate": 1.5411622276029057e-05,
      "loss": 1.0031,
      "step": 21070
    },
    {
      "epoch": 1.694533762057878,
      "grad_norm": 2.350111246109009,
      "learning_rate": 1.537126715092817e-05,
      "loss": 1.1756,
      "step": 21080
    },
    {
      "epoch": 1.695337620578778,
      "grad_norm": 1.7451907396316528,
      "learning_rate": 1.5330912025827283e-05,
      "loss": 1.0719,
      "step": 21090
    },
    {
      "epoch": 1.6961414790996785,
      "grad_norm": 1.607172966003418,
      "learning_rate": 1.5290556900726395e-05,
      "loss": 1.0103,
      "step": 21100
    },
    {
      "epoch": 1.6969453376205788,
      "grad_norm": 1.8843369483947754,
      "learning_rate": 1.5250201775625506e-05,
      "loss": 1.0322,
      "step": 21110
    },
    {
      "epoch": 1.697749196141479,
      "grad_norm": 1.918947458267212,
      "learning_rate": 1.5209846650524618e-05,
      "loss": 0.9708,
      "step": 21120
    },
    {
      "epoch": 1.6985530546623795,
      "grad_norm": 1.444896936416626,
      "learning_rate": 1.5169491525423729e-05,
      "loss": 0.9646,
      "step": 21130
    },
    {
      "epoch": 1.6993569131832797,
      "grad_norm": 1.6434109210968018,
      "learning_rate": 1.512913640032284e-05,
      "loss": 0.9987,
      "step": 21140
    },
    {
      "epoch": 1.70016077170418,
      "grad_norm": 2.072153091430664,
      "learning_rate": 1.5088781275221953e-05,
      "loss": 1.1089,
      "step": 21150
    },
    {
      "epoch": 1.7009646302250805,
      "grad_norm": 2.1333658695220947,
      "learning_rate": 1.5048426150121066e-05,
      "loss": 0.9657,
      "step": 21160
    },
    {
      "epoch": 1.7017684887459807,
      "grad_norm": 1.6481126546859741,
      "learning_rate": 1.500807102502018e-05,
      "loss": 0.9615,
      "step": 21170
    },
    {
      "epoch": 1.702572347266881,
      "grad_norm": 1.8588632345199585,
      "learning_rate": 1.496771589991929e-05,
      "loss": 1.0058,
      "step": 21180
    },
    {
      "epoch": 1.7033762057877815,
      "grad_norm": 1.585315465927124,
      "learning_rate": 1.4927360774818402e-05,
      "loss": 0.9077,
      "step": 21190
    },
    {
      "epoch": 1.7041800643086815,
      "grad_norm": 1.5307081937789917,
      "learning_rate": 1.4887005649717514e-05,
      "loss": 0.996,
      "step": 21200
    },
    {
      "epoch": 1.704983922829582,
      "grad_norm": 1.5457983016967773,
      "learning_rate": 1.4846650524616627e-05,
      "loss": 0.9997,
      "step": 21210
    },
    {
      "epoch": 1.7057877813504825,
      "grad_norm": 1.6371421813964844,
      "learning_rate": 1.480629539951574e-05,
      "loss": 1.1061,
      "step": 21220
    },
    {
      "epoch": 1.7065916398713825,
      "grad_norm": 2.6698966026306152,
      "learning_rate": 1.4765940274414853e-05,
      "loss": 1.0207,
      "step": 21230
    },
    {
      "epoch": 1.707395498392283,
      "grad_norm": 2.1621170043945312,
      "learning_rate": 1.4725585149313964e-05,
      "loss": 1.042,
      "step": 21240
    },
    {
      "epoch": 1.7081993569131833,
      "grad_norm": 1.244496464729309,
      "learning_rate": 1.4685230024213075e-05,
      "loss": 0.9997,
      "step": 21250
    },
    {
      "epoch": 1.7090032154340835,
      "grad_norm": 1.8812851905822754,
      "learning_rate": 1.4644874899112187e-05,
      "loss": 1.0055,
      "step": 21260
    },
    {
      "epoch": 1.709807073954984,
      "grad_norm": 1.6612039804458618,
      "learning_rate": 1.46045197740113e-05,
      "loss": 1.0835,
      "step": 21270
    },
    {
      "epoch": 1.7106109324758842,
      "grad_norm": 1.7879507541656494,
      "learning_rate": 1.4564164648910411e-05,
      "loss": 1.0079,
      "step": 21280
    },
    {
      "epoch": 1.7114147909967845,
      "grad_norm": 1.9121391773223877,
      "learning_rate": 1.4523809523809526e-05,
      "loss": 0.9546,
      "step": 21290
    },
    {
      "epoch": 1.712218649517685,
      "grad_norm": 1.898646593093872,
      "learning_rate": 1.4483454398708637e-05,
      "loss": 0.9777,
      "step": 21300
    },
    {
      "epoch": 1.7130225080385852,
      "grad_norm": 2.991395950317383,
      "learning_rate": 1.4443099273607749e-05,
      "loss": 1.0865,
      "step": 21310
    },
    {
      "epoch": 1.7138263665594855,
      "grad_norm": 1.890356183052063,
      "learning_rate": 1.440274414850686e-05,
      "loss": 0.9562,
      "step": 21320
    },
    {
      "epoch": 1.714630225080386,
      "grad_norm": 3.2333405017852783,
      "learning_rate": 1.4362389023405973e-05,
      "loss": 0.9115,
      "step": 21330
    },
    {
      "epoch": 1.715434083601286,
      "grad_norm": 1.6147865056991577,
      "learning_rate": 1.4322033898305085e-05,
      "loss": 0.9789,
      "step": 21340
    },
    {
      "epoch": 1.7162379421221865,
      "grad_norm": 1.7564023733139038,
      "learning_rate": 1.42816787732042e-05,
      "loss": 0.8648,
      "step": 21350
    },
    {
      "epoch": 1.717041800643087,
      "grad_norm": 1.7845501899719238,
      "learning_rate": 1.424132364810331e-05,
      "loss": 1.1064,
      "step": 21360
    },
    {
      "epoch": 1.717845659163987,
      "grad_norm": 1.3877874612808228,
      "learning_rate": 1.4200968523002422e-05,
      "loss": 1.0212,
      "step": 21370
    },
    {
      "epoch": 1.7186495176848875,
      "grad_norm": 2.5669033527374268,
      "learning_rate": 1.4160613397901535e-05,
      "loss": 0.98,
      "step": 21380
    },
    {
      "epoch": 1.7194533762057878,
      "grad_norm": 2.023080825805664,
      "learning_rate": 1.4120258272800646e-05,
      "loss": 1.0809,
      "step": 21390
    },
    {
      "epoch": 1.720257234726688,
      "grad_norm": 3.2553956508636475,
      "learning_rate": 1.4079903147699758e-05,
      "loss": 0.8719,
      "step": 21400
    },
    {
      "epoch": 1.7210610932475885,
      "grad_norm": 1.4506080150604248,
      "learning_rate": 1.403954802259887e-05,
      "loss": 0.9757,
      "step": 21410
    },
    {
      "epoch": 1.7218649517684887,
      "grad_norm": 1.6034953594207764,
      "learning_rate": 1.3999192897497984e-05,
      "loss": 1.0364,
      "step": 21420
    },
    {
      "epoch": 1.722668810289389,
      "grad_norm": 1.687049388885498,
      "learning_rate": 1.3958837772397095e-05,
      "loss": 1.0933,
      "step": 21430
    },
    {
      "epoch": 1.7234726688102895,
      "grad_norm": 1.4316902160644531,
      "learning_rate": 1.3918482647296208e-05,
      "loss": 1.0475,
      "step": 21440
    },
    {
      "epoch": 1.7242765273311897,
      "grad_norm": 2.5947353839874268,
      "learning_rate": 1.387812752219532e-05,
      "loss": 0.9705,
      "step": 21450
    },
    {
      "epoch": 1.72508038585209,
      "grad_norm": 1.9034242630004883,
      "learning_rate": 1.3837772397094431e-05,
      "loss": 1.0019,
      "step": 21460
    },
    {
      "epoch": 1.7258842443729905,
      "grad_norm": 1.8161375522613525,
      "learning_rate": 1.3797417271993542e-05,
      "loss": 0.9571,
      "step": 21470
    },
    {
      "epoch": 1.7266881028938905,
      "grad_norm": 3.107001304626465,
      "learning_rate": 1.3757062146892655e-05,
      "loss": 1.1365,
      "step": 21480
    },
    {
      "epoch": 1.727491961414791,
      "grad_norm": 1.6083005666732788,
      "learning_rate": 1.3716707021791769e-05,
      "loss": 1.0522,
      "step": 21490
    },
    {
      "epoch": 1.7282958199356915,
      "grad_norm": 2.342423915863037,
      "learning_rate": 1.3676351896690882e-05,
      "loss": 0.9688,
      "step": 21500
    },
    {
      "epoch": 1.7290996784565915,
      "grad_norm": 1.732395887374878,
      "learning_rate": 1.3635996771589993e-05,
      "loss": 1.0353,
      "step": 21510
    },
    {
      "epoch": 1.729903536977492,
      "grad_norm": 2.039433240890503,
      "learning_rate": 1.3595641646489104e-05,
      "loss": 1.0223,
      "step": 21520
    },
    {
      "epoch": 1.7307073954983923,
      "grad_norm": 1.7352782487869263,
      "learning_rate": 1.3555286521388216e-05,
      "loss": 0.9912,
      "step": 21530
    },
    {
      "epoch": 1.7315112540192925,
      "grad_norm": 1.8595890998840332,
      "learning_rate": 1.3514931396287329e-05,
      "loss": 0.9412,
      "step": 21540
    },
    {
      "epoch": 1.732315112540193,
      "grad_norm": 1.7216383218765259,
      "learning_rate": 1.3474576271186442e-05,
      "loss": 0.9254,
      "step": 21550
    },
    {
      "epoch": 1.7331189710610932,
      "grad_norm": 2.412393808364868,
      "learning_rate": 1.3434221146085555e-05,
      "loss": 1.0077,
      "step": 21560
    },
    {
      "epoch": 1.7339228295819935,
      "grad_norm": 1.396226167678833,
      "learning_rate": 1.3393866020984666e-05,
      "loss": 1.0498,
      "step": 21570
    },
    {
      "epoch": 1.734726688102894,
      "grad_norm": 2.1518938541412354,
      "learning_rate": 1.3353510895883778e-05,
      "loss": 1.0775,
      "step": 21580
    },
    {
      "epoch": 1.7355305466237942,
      "grad_norm": 1.9329767227172852,
      "learning_rate": 1.3313155770782889e-05,
      "loss": 0.9458,
      "step": 21590
    },
    {
      "epoch": 1.7363344051446945,
      "grad_norm": 2.125119209289551,
      "learning_rate": 1.3272800645682002e-05,
      "loss": 1.0538,
      "step": 21600
    },
    {
      "epoch": 1.737138263665595,
      "grad_norm": 1.8424369096755981,
      "learning_rate": 1.3232445520581113e-05,
      "loss": 0.98,
      "step": 21610
    },
    {
      "epoch": 1.737942122186495,
      "grad_norm": 2.402310848236084,
      "learning_rate": 1.3192090395480228e-05,
      "loss": 1.0943,
      "step": 21620
    },
    {
      "epoch": 1.7387459807073955,
      "grad_norm": 2.104525089263916,
      "learning_rate": 1.315173527037934e-05,
      "loss": 0.9797,
      "step": 21630
    },
    {
      "epoch": 1.739549839228296,
      "grad_norm": 2.5234835147857666,
      "learning_rate": 1.3111380145278451e-05,
      "loss": 1.0867,
      "step": 21640
    },
    {
      "epoch": 1.740353697749196,
      "grad_norm": 2.5901427268981934,
      "learning_rate": 1.3071025020177562e-05,
      "loss": 0.9924,
      "step": 21650
    },
    {
      "epoch": 1.7411575562700965,
      "grad_norm": 1.5725637674331665,
      "learning_rate": 1.3030669895076675e-05,
      "loss": 1.0287,
      "step": 21660
    },
    {
      "epoch": 1.7419614147909968,
      "grad_norm": 1.4871200323104858,
      "learning_rate": 1.2990314769975787e-05,
      "loss": 1.004,
      "step": 21670
    },
    {
      "epoch": 1.742765273311897,
      "grad_norm": 2.0282976627349854,
      "learning_rate": 1.2949959644874901e-05,
      "loss": 0.9733,
      "step": 21680
    },
    {
      "epoch": 1.7435691318327975,
      "grad_norm": 1.8308261632919312,
      "learning_rate": 1.2909604519774013e-05,
      "loss": 1.0513,
      "step": 21690
    },
    {
      "epoch": 1.7443729903536977,
      "grad_norm": 1.5598034858703613,
      "learning_rate": 1.2869249394673124e-05,
      "loss": 0.9191,
      "step": 21700
    },
    {
      "epoch": 1.745176848874598,
      "grad_norm": 1.718119502067566,
      "learning_rate": 1.2828894269572237e-05,
      "loss": 0.9828,
      "step": 21710
    },
    {
      "epoch": 1.7459807073954985,
      "grad_norm": 2.4848999977111816,
      "learning_rate": 1.2788539144471349e-05,
      "loss": 0.9463,
      "step": 21720
    },
    {
      "epoch": 1.7467845659163987,
      "grad_norm": 2.8907392024993896,
      "learning_rate": 1.274818401937046e-05,
      "loss": 1.0016,
      "step": 21730
    },
    {
      "epoch": 1.747588424437299,
      "grad_norm": 1.360023856163025,
      "learning_rate": 1.2707828894269571e-05,
      "loss": 0.891,
      "step": 21740
    },
    {
      "epoch": 1.7483922829581995,
      "grad_norm": 2.0196022987365723,
      "learning_rate": 1.2667473769168686e-05,
      "loss": 1.0393,
      "step": 21750
    },
    {
      "epoch": 1.7491961414790995,
      "grad_norm": 1.7127809524536133,
      "learning_rate": 1.2627118644067797e-05,
      "loss": 1.08,
      "step": 21760
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.263516902923584,
      "learning_rate": 1.258676351896691e-05,
      "loss": 1.0002,
      "step": 21770
    },
    {
      "epoch": 1.7508038585209005,
      "grad_norm": 1.9506158828735352,
      "learning_rate": 1.2546408393866022e-05,
      "loss": 1.0205,
      "step": 21780
    },
    {
      "epoch": 1.7516077170418005,
      "grad_norm": 1.926055669784546,
      "learning_rate": 1.2506053268765133e-05,
      "loss": 0.8754,
      "step": 21790
    },
    {
      "epoch": 1.752411575562701,
      "grad_norm": 1.4109805822372437,
      "learning_rate": 1.2465698143664246e-05,
      "loss": 0.9293,
      "step": 21800
    },
    {
      "epoch": 1.7532154340836013,
      "grad_norm": 1.531019687652588,
      "learning_rate": 1.2425343018563358e-05,
      "loss": 1.0311,
      "step": 21810
    },
    {
      "epoch": 1.7540192926045015,
      "grad_norm": 2.2952654361724854,
      "learning_rate": 1.238498789346247e-05,
      "loss": 0.9757,
      "step": 21820
    },
    {
      "epoch": 1.754823151125402,
      "grad_norm": 2.8620126247406006,
      "learning_rate": 1.2344632768361584e-05,
      "loss": 0.9115,
      "step": 21830
    },
    {
      "epoch": 1.7556270096463023,
      "grad_norm": 1.715909481048584,
      "learning_rate": 1.2304277643260695e-05,
      "loss": 1.0013,
      "step": 21840
    },
    {
      "epoch": 1.7564308681672025,
      "grad_norm": 2.3659169673919678,
      "learning_rate": 1.2263922518159806e-05,
      "loss": 0.9478,
      "step": 21850
    },
    {
      "epoch": 1.757234726688103,
      "grad_norm": 2.386011838912964,
      "learning_rate": 1.2223567393058918e-05,
      "loss": 1.0195,
      "step": 21860
    },
    {
      "epoch": 1.7580385852090032,
      "grad_norm": 2.0422582626342773,
      "learning_rate": 1.2183212267958031e-05,
      "loss": 0.9469,
      "step": 21870
    },
    {
      "epoch": 1.7588424437299035,
      "grad_norm": 1.73795747756958,
      "learning_rate": 1.2142857142857144e-05,
      "loss": 0.9687,
      "step": 21880
    },
    {
      "epoch": 1.759646302250804,
      "grad_norm": 2.1826019287109375,
      "learning_rate": 1.2102502017756255e-05,
      "loss": 0.9925,
      "step": 21890
    },
    {
      "epoch": 1.760450160771704,
      "grad_norm": 1.6059566736221313,
      "learning_rate": 1.2062146892655368e-05,
      "loss": 0.9586,
      "step": 21900
    },
    {
      "epoch": 1.7612540192926045,
      "grad_norm": 1.6619853973388672,
      "learning_rate": 1.202179176755448e-05,
      "loss": 0.9419,
      "step": 21910
    },
    {
      "epoch": 1.762057877813505,
      "grad_norm": 1.4436253309249878,
      "learning_rate": 1.1981436642453591e-05,
      "loss": 1.037,
      "step": 21920
    },
    {
      "epoch": 1.762861736334405,
      "grad_norm": 1.9889531135559082,
      "learning_rate": 1.1941081517352704e-05,
      "loss": 1.0703,
      "step": 21930
    },
    {
      "epoch": 1.7636655948553055,
      "grad_norm": 1.9301440715789795,
      "learning_rate": 1.1900726392251817e-05,
      "loss": 0.964,
      "step": 21940
    },
    {
      "epoch": 1.7644694533762058,
      "grad_norm": 2.3707351684570312,
      "learning_rate": 1.1860371267150929e-05,
      "loss": 0.968,
      "step": 21950
    },
    {
      "epoch": 1.765273311897106,
      "grad_norm": 1.9376273155212402,
      "learning_rate": 1.1820016142050042e-05,
      "loss": 0.8785,
      "step": 21960
    },
    {
      "epoch": 1.7660771704180065,
      "grad_norm": 2.0454390048980713,
      "learning_rate": 1.1779661016949153e-05,
      "loss": 0.8985,
      "step": 21970
    },
    {
      "epoch": 1.7668810289389068,
      "grad_norm": 2.0367326736450195,
      "learning_rate": 1.1739305891848264e-05,
      "loss": 0.8673,
      "step": 21980
    },
    {
      "epoch": 1.767684887459807,
      "grad_norm": 1.7736015319824219,
      "learning_rate": 1.1698950766747377e-05,
      "loss": 0.9391,
      "step": 21990
    },
    {
      "epoch": 1.7684887459807075,
      "grad_norm": 1.967177391052246,
      "learning_rate": 1.165859564164649e-05,
      "loss": 1.0408,
      "step": 22000
    },
    {
      "epoch": 1.7684887459807075,
      "eval_yahma/alpaca-cleaned_loss": 1.2060405015945435,
      "eval_yahma/alpaca-cleaned_runtime": 115.751,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.278,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.16,
      "step": 22000
    },
    {
      "epoch": 1.7692926045016077,
      "grad_norm": 2.349555492401123,
      "learning_rate": 1.1618240516545602e-05,
      "loss": 1.0432,
      "step": 22010
    },
    {
      "epoch": 1.770096463022508,
      "grad_norm": 2.213883638381958,
      "learning_rate": 1.1577885391444713e-05,
      "loss": 0.8698,
      "step": 22020
    },
    {
      "epoch": 1.7709003215434085,
      "grad_norm": 1.762036919593811,
      "learning_rate": 1.1537530266343826e-05,
      "loss": 0.8771,
      "step": 22030
    },
    {
      "epoch": 1.7717041800643085,
      "grad_norm": 0.9419066905975342,
      "learning_rate": 1.1497175141242938e-05,
      "loss": 0.813,
      "step": 22040
    },
    {
      "epoch": 1.772508038585209,
      "grad_norm": 1.4384018182754517,
      "learning_rate": 1.145682001614205e-05,
      "loss": 0.9929,
      "step": 22050
    },
    {
      "epoch": 1.7733118971061095,
      "grad_norm": 2.7204196453094482,
      "learning_rate": 1.1416464891041164e-05,
      "loss": 0.9315,
      "step": 22060
    },
    {
      "epoch": 1.7741157556270095,
      "grad_norm": 1.8378400802612305,
      "learning_rate": 1.1376109765940275e-05,
      "loss": 1.0011,
      "step": 22070
    },
    {
      "epoch": 1.77491961414791,
      "grad_norm": 1.4868669509887695,
      "learning_rate": 1.1335754640839387e-05,
      "loss": 1.0055,
      "step": 22080
    },
    {
      "epoch": 1.7757234726688103,
      "grad_norm": 1.6384830474853516,
      "learning_rate": 1.1295399515738498e-05,
      "loss": 1.095,
      "step": 22090
    },
    {
      "epoch": 1.7765273311897105,
      "grad_norm": 1.8896377086639404,
      "learning_rate": 1.1255044390637613e-05,
      "loss": 1.0446,
      "step": 22100
    },
    {
      "epoch": 1.777331189710611,
      "grad_norm": 1.6007559299468994,
      "learning_rate": 1.1214689265536724e-05,
      "loss": 1.0298,
      "step": 22110
    },
    {
      "epoch": 1.7781350482315113,
      "grad_norm": 1.9358576536178589,
      "learning_rate": 1.1174334140435835e-05,
      "loss": 0.9502,
      "step": 22120
    },
    {
      "epoch": 1.7789389067524115,
      "grad_norm": 1.6456011533737183,
      "learning_rate": 1.1133979015334948e-05,
      "loss": 0.9911,
      "step": 22130
    },
    {
      "epoch": 1.779742765273312,
      "grad_norm": 2.3452653884887695,
      "learning_rate": 1.109362389023406e-05,
      "loss": 0.9298,
      "step": 22140
    },
    {
      "epoch": 1.7805466237942122,
      "grad_norm": 1.9005801677703857,
      "learning_rate": 1.1053268765133173e-05,
      "loss": 0.9438,
      "step": 22150
    },
    {
      "epoch": 1.7813504823151125,
      "grad_norm": 2.6864964962005615,
      "learning_rate": 1.1012913640032286e-05,
      "loss": 0.9279,
      "step": 22160
    },
    {
      "epoch": 1.782154340836013,
      "grad_norm": 1.7829868793487549,
      "learning_rate": 1.0972558514931397e-05,
      "loss": 0.9321,
      "step": 22170
    },
    {
      "epoch": 1.782958199356913,
      "grad_norm": 2.1801998615264893,
      "learning_rate": 1.0932203389830509e-05,
      "loss": 0.9647,
      "step": 22180
    },
    {
      "epoch": 1.7837620578778135,
      "grad_norm": 1.683989405632019,
      "learning_rate": 1.089184826472962e-05,
      "loss": 0.971,
      "step": 22190
    },
    {
      "epoch": 1.784565916398714,
      "grad_norm": 1.9461780786514282,
      "learning_rate": 1.0851493139628733e-05,
      "loss": 1.0014,
      "step": 22200
    },
    {
      "epoch": 1.785369774919614,
      "grad_norm": 1.9797353744506836,
      "learning_rate": 1.0811138014527846e-05,
      "loss": 1.0008,
      "step": 22210
    },
    {
      "epoch": 1.7861736334405145,
      "grad_norm": 1.445884108543396,
      "learning_rate": 1.0770782889426957e-05,
      "loss": 0.9971,
      "step": 22220
    },
    {
      "epoch": 1.7869774919614148,
      "grad_norm": 2.3050432205200195,
      "learning_rate": 1.073042776432607e-05,
      "loss": 0.9472,
      "step": 22230
    },
    {
      "epoch": 1.787781350482315,
      "grad_norm": 1.5249042510986328,
      "learning_rate": 1.0690072639225182e-05,
      "loss": 1.1952,
      "step": 22240
    },
    {
      "epoch": 1.7885852090032155,
      "grad_norm": 1.5306737422943115,
      "learning_rate": 1.0649717514124293e-05,
      "loss": 1.1246,
      "step": 22250
    },
    {
      "epoch": 1.7893890675241158,
      "grad_norm": 1.9988269805908203,
      "learning_rate": 1.0609362389023406e-05,
      "loss": 1.0048,
      "step": 22260
    },
    {
      "epoch": 1.790192926045016,
      "grad_norm": 1.5958483219146729,
      "learning_rate": 1.056900726392252e-05,
      "loss": 0.9573,
      "step": 22270
    },
    {
      "epoch": 1.7909967845659165,
      "grad_norm": 1.8712174892425537,
      "learning_rate": 1.052865213882163e-05,
      "loss": 0.8928,
      "step": 22280
    },
    {
      "epoch": 1.7918006430868167,
      "grad_norm": 1.641789197921753,
      "learning_rate": 1.0488297013720744e-05,
      "loss": 0.9694,
      "step": 22290
    },
    {
      "epoch": 1.792604501607717,
      "grad_norm": 3.96578311920166,
      "learning_rate": 1.0447941888619855e-05,
      "loss": 1.0946,
      "step": 22300
    },
    {
      "epoch": 1.7934083601286175,
      "grad_norm": 1.7721596956253052,
      "learning_rate": 1.0407586763518967e-05,
      "loss": 1.0712,
      "step": 22310
    },
    {
      "epoch": 1.7942122186495175,
      "grad_norm": 1.861772894859314,
      "learning_rate": 1.036723163841808e-05,
      "loss": 0.9103,
      "step": 22320
    },
    {
      "epoch": 1.795016077170418,
      "grad_norm": 1.7235984802246094,
      "learning_rate": 1.0326876513317193e-05,
      "loss": 1.0103,
      "step": 22330
    },
    {
      "epoch": 1.7958199356913185,
      "grad_norm": 1.3846309185028076,
      "learning_rate": 1.0286521388216304e-05,
      "loss": 0.9419,
      "step": 22340
    },
    {
      "epoch": 1.7966237942122185,
      "grad_norm": 3.1098134517669678,
      "learning_rate": 1.0246166263115415e-05,
      "loss": 0.9782,
      "step": 22350
    },
    {
      "epoch": 1.797427652733119,
      "grad_norm": 2.1430604457855225,
      "learning_rate": 1.0205811138014528e-05,
      "loss": 0.9843,
      "step": 22360
    },
    {
      "epoch": 1.7982315112540193,
      "grad_norm": 1.5648924112319946,
      "learning_rate": 1.016545601291364e-05,
      "loss": 0.9367,
      "step": 22370
    },
    {
      "epoch": 1.7990353697749195,
      "grad_norm": 1.7868338823318481,
      "learning_rate": 1.0125100887812753e-05,
      "loss": 0.985,
      "step": 22380
    },
    {
      "epoch": 1.79983922829582,
      "grad_norm": 2.122217893600464,
      "learning_rate": 1.0084745762711866e-05,
      "loss": 1.0524,
      "step": 22390
    },
    {
      "epoch": 1.8006430868167203,
      "grad_norm": 3.732097625732422,
      "learning_rate": 1.0044390637610977e-05,
      "loss": 0.9872,
      "step": 22400
    },
    {
      "epoch": 1.8014469453376205,
      "grad_norm": 1.2948299646377563,
      "learning_rate": 1.0004035512510089e-05,
      "loss": 0.9207,
      "step": 22410
    },
    {
      "epoch": 1.802250803858521,
      "grad_norm": 1.9187204837799072,
      "learning_rate": 9.9636803874092e-06,
      "loss": 1.1478,
      "step": 22420
    },
    {
      "epoch": 1.8030546623794212,
      "grad_norm": 1.4700504541397095,
      "learning_rate": 9.923325262308315e-06,
      "loss": 1.0152,
      "step": 22430
    },
    {
      "epoch": 1.8038585209003215,
      "grad_norm": 1.5490188598632812,
      "learning_rate": 9.882970137207426e-06,
      "loss": 0.9741,
      "step": 22440
    },
    {
      "epoch": 1.804662379421222,
      "grad_norm": 1.7711931467056274,
      "learning_rate": 9.842615012106538e-06,
      "loss": 0.9791,
      "step": 22450
    },
    {
      "epoch": 1.805466237942122,
      "grad_norm": 2.2049827575683594,
      "learning_rate": 9.80225988700565e-06,
      "loss": 0.9413,
      "step": 22460
    },
    {
      "epoch": 1.8062700964630225,
      "grad_norm": 2.086760997772217,
      "learning_rate": 9.761904761904762e-06,
      "loss": 1.029,
      "step": 22470
    },
    {
      "epoch": 1.807073954983923,
      "grad_norm": 4.496650218963623,
      "learning_rate": 9.721549636803875e-06,
      "loss": 1.0884,
      "step": 22480
    },
    {
      "epoch": 1.807877813504823,
      "grad_norm": 1.6099193096160889,
      "learning_rate": 9.681194511702988e-06,
      "loss": 0.9469,
      "step": 22490
    },
    {
      "epoch": 1.8086816720257235,
      "grad_norm": 2.1339876651763916,
      "learning_rate": 9.6408393866021e-06,
      "loss": 1.0124,
      "step": 22500
    },
    {
      "epoch": 1.8094855305466238,
      "grad_norm": 1.8483517169952393,
      "learning_rate": 9.60048426150121e-06,
      "loss": 0.9827,
      "step": 22510
    },
    {
      "epoch": 1.810289389067524,
      "grad_norm": 2.988699197769165,
      "learning_rate": 9.560129136400324e-06,
      "loss": 1.1299,
      "step": 22520
    },
    {
      "epoch": 1.8110932475884245,
      "grad_norm": 1.7932300567626953,
      "learning_rate": 9.519774011299435e-06,
      "loss": 0.9944,
      "step": 22530
    },
    {
      "epoch": 1.8118971061093248,
      "grad_norm": 2.6900038719177246,
      "learning_rate": 9.479418886198548e-06,
      "loss": 1.0668,
      "step": 22540
    },
    {
      "epoch": 1.812700964630225,
      "grad_norm": 1.5085099935531616,
      "learning_rate": 9.43906376109766e-06,
      "loss": 1.0023,
      "step": 22550
    },
    {
      "epoch": 1.8135048231511255,
      "grad_norm": 2.3199098110198975,
      "learning_rate": 9.398708635996773e-06,
      "loss": 1.1052,
      "step": 22560
    },
    {
      "epoch": 1.8143086816720257,
      "grad_norm": 1.4957773685455322,
      "learning_rate": 9.358353510895884e-06,
      "loss": 1.0233,
      "step": 22570
    },
    {
      "epoch": 1.815112540192926,
      "grad_norm": 2.718686819076538,
      "learning_rate": 9.317998385794995e-06,
      "loss": 1.0884,
      "step": 22580
    },
    {
      "epoch": 1.8159163987138265,
      "grad_norm": 2.2220447063446045,
      "learning_rate": 9.277643260694108e-06,
      "loss": 0.9866,
      "step": 22590
    },
    {
      "epoch": 1.8167202572347267,
      "grad_norm": 1.6045563220977783,
      "learning_rate": 9.237288135593222e-06,
      "loss": 1.0357,
      "step": 22600
    },
    {
      "epoch": 1.817524115755627,
      "grad_norm": 1.5356566905975342,
      "learning_rate": 9.196933010492333e-06,
      "loss": 0.9883,
      "step": 22610
    },
    {
      "epoch": 1.8183279742765275,
      "grad_norm": 1.7813926935195923,
      "learning_rate": 9.156577885391446e-06,
      "loss": 0.9772,
      "step": 22620
    },
    {
      "epoch": 1.8191318327974275,
      "grad_norm": 2.866159439086914,
      "learning_rate": 9.116222760290557e-06,
      "loss": 0.9915,
      "step": 22630
    },
    {
      "epoch": 1.819935691318328,
      "grad_norm": 2.222846508026123,
      "learning_rate": 9.075867635189669e-06,
      "loss": 1.0007,
      "step": 22640
    },
    {
      "epoch": 1.8207395498392283,
      "grad_norm": 2.0583643913269043,
      "learning_rate": 9.035512510088782e-06,
      "loss": 1.0692,
      "step": 22650
    },
    {
      "epoch": 1.8215434083601285,
      "grad_norm": 1.5456589460372925,
      "learning_rate": 8.995157384987895e-06,
      "loss": 0.9902,
      "step": 22660
    },
    {
      "epoch": 1.822347266881029,
      "grad_norm": 1.7286393642425537,
      "learning_rate": 8.954802259887006e-06,
      "loss": 1.0302,
      "step": 22670
    },
    {
      "epoch": 1.8231511254019293,
      "grad_norm": 1.6188042163848877,
      "learning_rate": 8.914447134786118e-06,
      "loss": 1.1087,
      "step": 22680
    },
    {
      "epoch": 1.8239549839228295,
      "grad_norm": 1.6614141464233398,
      "learning_rate": 8.87409200968523e-06,
      "loss": 1.0115,
      "step": 22690
    },
    {
      "epoch": 1.82475884244373,
      "grad_norm": 1.9828251600265503,
      "learning_rate": 8.833736884584342e-06,
      "loss": 0.8943,
      "step": 22700
    },
    {
      "epoch": 1.8255627009646302,
      "grad_norm": 1.587049126625061,
      "learning_rate": 8.793381759483455e-06,
      "loss": 0.9652,
      "step": 22710
    },
    {
      "epoch": 1.8263665594855305,
      "grad_norm": 2.128166675567627,
      "learning_rate": 8.753026634382568e-06,
      "loss": 1.0803,
      "step": 22720
    },
    {
      "epoch": 1.827170418006431,
      "grad_norm": 2.181746006011963,
      "learning_rate": 8.71267150928168e-06,
      "loss": 1.0535,
      "step": 22730
    },
    {
      "epoch": 1.8279742765273312,
      "grad_norm": 1.4443950653076172,
      "learning_rate": 8.67231638418079e-06,
      "loss": 0.9807,
      "step": 22740
    },
    {
      "epoch": 1.8287781350482315,
      "grad_norm": 4.426495552062988,
      "learning_rate": 8.631961259079902e-06,
      "loss": 0.9908,
      "step": 22750
    },
    {
      "epoch": 1.829581993569132,
      "grad_norm": 1.8822588920593262,
      "learning_rate": 8.591606133979015e-06,
      "loss": 1.083,
      "step": 22760
    },
    {
      "epoch": 1.830385852090032,
      "grad_norm": 1.6437195539474487,
      "learning_rate": 8.551251008878128e-06,
      "loss": 1.0005,
      "step": 22770
    },
    {
      "epoch": 1.8311897106109325,
      "grad_norm": 1.8927476406097412,
      "learning_rate": 8.51089588377724e-06,
      "loss": 1.0809,
      "step": 22780
    },
    {
      "epoch": 1.8319935691318328,
      "grad_norm": 2.30253267288208,
      "learning_rate": 8.470540758676353e-06,
      "loss": 0.9831,
      "step": 22790
    },
    {
      "epoch": 1.832797427652733,
      "grad_norm": 2.003490447998047,
      "learning_rate": 8.430185633575464e-06,
      "loss": 1.022,
      "step": 22800
    },
    {
      "epoch": 1.8336012861736335,
      "grad_norm": 3.780564308166504,
      "learning_rate": 8.389830508474575e-06,
      "loss": 0.9888,
      "step": 22810
    },
    {
      "epoch": 1.8344051446945338,
      "grad_norm": 1.9336515665054321,
      "learning_rate": 8.34947538337369e-06,
      "loss": 1.0131,
      "step": 22820
    },
    {
      "epoch": 1.835209003215434,
      "grad_norm": 1.8983111381530762,
      "learning_rate": 8.309120258272802e-06,
      "loss": 1.161,
      "step": 22830
    },
    {
      "epoch": 1.8360128617363345,
      "grad_norm": 2.518303871154785,
      "learning_rate": 8.268765133171913e-06,
      "loss": 1.1041,
      "step": 22840
    },
    {
      "epoch": 1.8368167202572347,
      "grad_norm": 1.8314872980117798,
      "learning_rate": 8.228410008071026e-06,
      "loss": 1.0529,
      "step": 22850
    },
    {
      "epoch": 1.837620578778135,
      "grad_norm": 1.640785813331604,
      "learning_rate": 8.188054882970137e-06,
      "loss": 0.9747,
      "step": 22860
    },
    {
      "epoch": 1.8384244372990355,
      "grad_norm": 1.3812037706375122,
      "learning_rate": 8.14769975786925e-06,
      "loss": 1.0021,
      "step": 22870
    },
    {
      "epoch": 1.8392282958199357,
      "grad_norm": 1.4298020601272583,
      "learning_rate": 8.107344632768362e-06,
      "loss": 0.9967,
      "step": 22880
    },
    {
      "epoch": 1.840032154340836,
      "grad_norm": 1.9113152027130127,
      "learning_rate": 8.066989507667475e-06,
      "loss": 0.8342,
      "step": 22890
    },
    {
      "epoch": 1.8408360128617365,
      "grad_norm": 2.049980878829956,
      "learning_rate": 8.026634382566586e-06,
      "loss": 1.002,
      "step": 22900
    },
    {
      "epoch": 1.8416398713826365,
      "grad_norm": 1.9057393074035645,
      "learning_rate": 7.986279257465698e-06,
      "loss": 0.9655,
      "step": 22910
    },
    {
      "epoch": 1.842443729903537,
      "grad_norm": 1.5741013288497925,
      "learning_rate": 7.94592413236481e-06,
      "loss": 1.0502,
      "step": 22920
    },
    {
      "epoch": 1.8432475884244373,
      "grad_norm": 2.031848192214966,
      "learning_rate": 7.905569007263924e-06,
      "loss": 1.0311,
      "step": 22930
    },
    {
      "epoch": 1.8440514469453375,
      "grad_norm": 3.889958381652832,
      "learning_rate": 7.865213882163035e-06,
      "loss": 0.9455,
      "step": 22940
    },
    {
      "epoch": 1.844855305466238,
      "grad_norm": 1.9395737648010254,
      "learning_rate": 7.824858757062148e-06,
      "loss": 0.9401,
      "step": 22950
    },
    {
      "epoch": 1.8456591639871383,
      "grad_norm": 4.533421516418457,
      "learning_rate": 7.78450363196126e-06,
      "loss": 0.9171,
      "step": 22960
    },
    {
      "epoch": 1.8464630225080385,
      "grad_norm": 1.5359047651290894,
      "learning_rate": 7.74414850686037e-06,
      "loss": 0.9464,
      "step": 22970
    },
    {
      "epoch": 1.847266881028939,
      "grad_norm": 1.548888921737671,
      "learning_rate": 7.703793381759484e-06,
      "loss": 1.0143,
      "step": 22980
    },
    {
      "epoch": 1.8480707395498392,
      "grad_norm": 1.649289608001709,
      "learning_rate": 7.663438256658597e-06,
      "loss": 1.1122,
      "step": 22990
    },
    {
      "epoch": 1.8488745980707395,
      "grad_norm": 2.1079747676849365,
      "learning_rate": 7.623083131557708e-06,
      "loss": 0.9877,
      "step": 23000
    },
    {
      "epoch": 1.84967845659164,
      "grad_norm": 1.611446738243103,
      "learning_rate": 7.58272800645682e-06,
      "loss": 0.9571,
      "step": 23010
    },
    {
      "epoch": 1.8504823151125402,
      "grad_norm": 1.5056588649749756,
      "learning_rate": 7.542372881355933e-06,
      "loss": 0.9905,
      "step": 23020
    },
    {
      "epoch": 1.8512861736334405,
      "grad_norm": 1.9945638179779053,
      "learning_rate": 7.502017756255045e-06,
      "loss": 1.0252,
      "step": 23030
    },
    {
      "epoch": 1.852090032154341,
      "grad_norm": 1.8758174180984497,
      "learning_rate": 7.461662631154156e-06,
      "loss": 1.038,
      "step": 23040
    },
    {
      "epoch": 1.852893890675241,
      "grad_norm": 2.0755865573883057,
      "learning_rate": 7.425343018563358e-06,
      "loss": 1.0526,
      "step": 23050
    },
    {
      "epoch": 1.8536977491961415,
      "grad_norm": 1.5491857528686523,
      "learning_rate": 7.3849878934624694e-06,
      "loss": 1.0086,
      "step": 23060
    },
    {
      "epoch": 1.8545016077170418,
      "grad_norm": 3.110924482345581,
      "learning_rate": 7.3446327683615825e-06,
      "loss": 0.9758,
      "step": 23070
    },
    {
      "epoch": 1.855305466237942,
      "grad_norm": 1.3394163846969604,
      "learning_rate": 7.304277643260695e-06,
      "loss": 1.0135,
      "step": 23080
    },
    {
      "epoch": 1.8561093247588425,
      "grad_norm": 1.9073662757873535,
      "learning_rate": 7.263922518159806e-06,
      "loss": 1.0456,
      "step": 23090
    },
    {
      "epoch": 1.8569131832797428,
      "grad_norm": 1.4355270862579346,
      "learning_rate": 7.223567393058919e-06,
      "loss": 1.045,
      "step": 23100
    },
    {
      "epoch": 1.857717041800643,
      "grad_norm": 2.270494222640991,
      "learning_rate": 7.183212267958031e-06,
      "loss": 0.9851,
      "step": 23110
    },
    {
      "epoch": 1.8585209003215435,
      "grad_norm": 2.288757085800171,
      "learning_rate": 7.142857142857143e-06,
      "loss": 1.0242,
      "step": 23120
    },
    {
      "epoch": 1.8593247588424437,
      "grad_norm": 1.5285948514938354,
      "learning_rate": 7.102502017756256e-06,
      "loss": 1.0258,
      "step": 23130
    },
    {
      "epoch": 1.860128617363344,
      "grad_norm": 1.900128722190857,
      "learning_rate": 7.062146892655368e-06,
      "loss": 0.8494,
      "step": 23140
    },
    {
      "epoch": 1.8609324758842445,
      "grad_norm": 1.519309639930725,
      "learning_rate": 7.021791767554479e-06,
      "loss": 1.0318,
      "step": 23150
    },
    {
      "epoch": 1.8617363344051447,
      "grad_norm": 1.7744423151016235,
      "learning_rate": 6.9814366424535916e-06,
      "loss": 0.9396,
      "step": 23160
    },
    {
      "epoch": 1.862540192926045,
      "grad_norm": 1.5099612474441528,
      "learning_rate": 6.941081517352705e-06,
      "loss": 0.9436,
      "step": 23170
    },
    {
      "epoch": 1.8633440514469455,
      "grad_norm": 1.8056023120880127,
      "learning_rate": 6.900726392251816e-06,
      "loss": 0.9437,
      "step": 23180
    },
    {
      "epoch": 1.8641479099678455,
      "grad_norm": 2.491077423095703,
      "learning_rate": 6.860371267150928e-06,
      "loss": 0.9675,
      "step": 23190
    },
    {
      "epoch": 1.864951768488746,
      "grad_norm": 1.5670075416564941,
      "learning_rate": 6.820016142050041e-06,
      "loss": 0.9382,
      "step": 23200
    },
    {
      "epoch": 1.8657556270096463,
      "grad_norm": 1.6534690856933594,
      "learning_rate": 6.779661016949153e-06,
      "loss": 1.0577,
      "step": 23210
    },
    {
      "epoch": 1.8665594855305465,
      "grad_norm": 1.7739144563674927,
      "learning_rate": 6.739305891848265e-06,
      "loss": 1.0413,
      "step": 23220
    },
    {
      "epoch": 1.867363344051447,
      "grad_norm": 2.2446448802948,
      "learning_rate": 6.698950766747378e-06,
      "loss": 0.9624,
      "step": 23230
    },
    {
      "epoch": 1.8681672025723473,
      "grad_norm": 1.8075047731399536,
      "learning_rate": 6.658595641646489e-06,
      "loss": 1.0003,
      "step": 23240
    },
    {
      "epoch": 1.8689710610932475,
      "grad_norm": 1.8375296592712402,
      "learning_rate": 6.6182405165456015e-06,
      "loss": 0.9927,
      "step": 23250
    },
    {
      "epoch": 1.869774919614148,
      "grad_norm": 1.985586166381836,
      "learning_rate": 6.5778853914447145e-06,
      "loss": 1.0433,
      "step": 23260
    },
    {
      "epoch": 1.8705787781350482,
      "grad_norm": 1.910912275314331,
      "learning_rate": 6.537530266343826e-06,
      "loss": 0.9427,
      "step": 23270
    },
    {
      "epoch": 1.8713826366559485,
      "grad_norm": 2.1846415996551514,
      "learning_rate": 6.497175141242938e-06,
      "loss": 0.9872,
      "step": 23280
    },
    {
      "epoch": 1.872186495176849,
      "grad_norm": 1.6068943738937378,
      "learning_rate": 6.4568200161420495e-06,
      "loss": 0.9514,
      "step": 23290
    },
    {
      "epoch": 1.8729903536977492,
      "grad_norm": 1.9065032005310059,
      "learning_rate": 6.4164648910411625e-06,
      "loss": 1.0191,
      "step": 23300
    },
    {
      "epoch": 1.8737942122186495,
      "grad_norm": 1.668900489807129,
      "learning_rate": 6.376109765940275e-06,
      "loss": 0.964,
      "step": 23310
    },
    {
      "epoch": 1.87459807073955,
      "grad_norm": 1.5289818048477173,
      "learning_rate": 6.335754640839386e-06,
      "loss": 0.9577,
      "step": 23320
    },
    {
      "epoch": 1.87540192926045,
      "grad_norm": 1.8854244947433472,
      "learning_rate": 6.295399515738499e-06,
      "loss": 1.1001,
      "step": 23330
    },
    {
      "epoch": 1.8762057877813505,
      "grad_norm": 2.1930129528045654,
      "learning_rate": 6.255044390637611e-06,
      "loss": 1.036,
      "step": 23340
    },
    {
      "epoch": 1.8770096463022508,
      "grad_norm": 1.6574493646621704,
      "learning_rate": 6.214689265536724e-06,
      "loss": 1.0617,
      "step": 23350
    },
    {
      "epoch": 1.877813504823151,
      "grad_norm": 1.4845741987228394,
      "learning_rate": 6.174334140435836e-06,
      "loss": 1.0416,
      "step": 23360
    },
    {
      "epoch": 1.8786173633440515,
      "grad_norm": 1.7336281538009644,
      "learning_rate": 6.133979015334948e-06,
      "loss": 0.9253,
      "step": 23370
    },
    {
      "epoch": 1.8794212218649518,
      "grad_norm": 1.5522409677505493,
      "learning_rate": 6.093623890234059e-06,
      "loss": 1.0626,
      "step": 23380
    },
    {
      "epoch": 1.880225080385852,
      "grad_norm": 2.1845033168792725,
      "learning_rate": 6.0532687651331724e-06,
      "loss": 1.1008,
      "step": 23390
    },
    {
      "epoch": 1.8810289389067525,
      "grad_norm": 1.3744220733642578,
      "learning_rate": 6.012913640032285e-06,
      "loss": 1.0222,
      "step": 23400
    },
    {
      "epoch": 1.8818327974276527,
      "grad_norm": 1.8450615406036377,
      "learning_rate": 5.972558514931397e-06,
      "loss": 1.1045,
      "step": 23410
    },
    {
      "epoch": 1.882636655948553,
      "grad_norm": 3.2560296058654785,
      "learning_rate": 5.932203389830509e-06,
      "loss": 1.0165,
      "step": 23420
    },
    {
      "epoch": 1.8834405144694535,
      "grad_norm": 3.120931386947632,
      "learning_rate": 5.8918482647296204e-06,
      "loss": 0.9879,
      "step": 23430
    },
    {
      "epoch": 1.8842443729903537,
      "grad_norm": 2.820199728012085,
      "learning_rate": 5.8514931396287335e-06,
      "loss": 0.9497,
      "step": 23440
    },
    {
      "epoch": 1.885048231511254,
      "grad_norm": 2.5963895320892334,
      "learning_rate": 5.811138014527846e-06,
      "loss": 1.067,
      "step": 23450
    },
    {
      "epoch": 1.8858520900321545,
      "grad_norm": 1.6238845586776733,
      "learning_rate": 5.770782889426957e-06,
      "loss": 1.0693,
      "step": 23460
    },
    {
      "epoch": 1.8866559485530545,
      "grad_norm": 1.703322172164917,
      "learning_rate": 5.73042776432607e-06,
      "loss": 0.8865,
      "step": 23470
    },
    {
      "epoch": 1.887459807073955,
      "grad_norm": 2.031500816345215,
      "learning_rate": 5.6900726392251815e-06,
      "loss": 1.0144,
      "step": 23480
    },
    {
      "epoch": 1.8882636655948553,
      "grad_norm": 1.9934537410736084,
      "learning_rate": 5.649717514124294e-06,
      "loss": 0.9858,
      "step": 23490
    },
    {
      "epoch": 1.8890675241157555,
      "grad_norm": 1.85928475856781,
      "learning_rate": 5.609362389023407e-06,
      "loss": 0.9204,
      "step": 23500
    },
    {
      "epoch": 1.889871382636656,
      "grad_norm": 1.7416903972625732,
      "learning_rate": 5.569007263922518e-06,
      "loss": 0.9034,
      "step": 23510
    },
    {
      "epoch": 1.8906752411575563,
      "grad_norm": 2.1307573318481445,
      "learning_rate": 5.52865213882163e-06,
      "loss": 0.8743,
      "step": 23520
    },
    {
      "epoch": 1.8914790996784565,
      "grad_norm": 1.882643461227417,
      "learning_rate": 5.4882970137207426e-06,
      "loss": 1.0894,
      "step": 23530
    },
    {
      "epoch": 1.892282958199357,
      "grad_norm": 1.6504333019256592,
      "learning_rate": 5.447941888619855e-06,
      "loss": 1.132,
      "step": 23540
    },
    {
      "epoch": 1.8930868167202572,
      "grad_norm": 1.440651297569275,
      "learning_rate": 5.407586763518967e-06,
      "loss": 1.0758,
      "step": 23550
    },
    {
      "epoch": 1.8938906752411575,
      "grad_norm": 2.3337113857269287,
      "learning_rate": 5.367231638418079e-06,
      "loss": 0.9673,
      "step": 23560
    },
    {
      "epoch": 1.894694533762058,
      "grad_norm": 2.217491865158081,
      "learning_rate": 5.326876513317191e-06,
      "loss": 1.0566,
      "step": 23570
    },
    {
      "epoch": 1.8954983922829582,
      "grad_norm": 2.5645487308502197,
      "learning_rate": 5.286521388216304e-06,
      "loss": 0.999,
      "step": 23580
    },
    {
      "epoch": 1.8963022508038585,
      "grad_norm": 1.5525003671646118,
      "learning_rate": 5.246166263115416e-06,
      "loss": 1.1088,
      "step": 23590
    },
    {
      "epoch": 1.897106109324759,
      "grad_norm": 2.1636111736297607,
      "learning_rate": 5.205811138014528e-06,
      "loss": 0.9676,
      "step": 23600
    },
    {
      "epoch": 1.897909967845659,
      "grad_norm": 1.9286320209503174,
      "learning_rate": 5.16545601291364e-06,
      "loss": 1.0241,
      "step": 23610
    },
    {
      "epoch": 1.8987138263665595,
      "grad_norm": 2.1557018756866455,
      "learning_rate": 5.1251008878127525e-06,
      "loss": 0.9821,
      "step": 23620
    },
    {
      "epoch": 1.8995176848874598,
      "grad_norm": 1.5869606733322144,
      "learning_rate": 5.084745762711865e-06,
      "loss": 0.9518,
      "step": 23630
    },
    {
      "epoch": 1.90032154340836,
      "grad_norm": 2.810943365097046,
      "learning_rate": 5.044390637610977e-06,
      "loss": 1.0004,
      "step": 23640
    },
    {
      "epoch": 1.9011254019292605,
      "grad_norm": 1.8571282625198364,
      "learning_rate": 5.004035512510089e-06,
      "loss": 1.0284,
      "step": 23650
    },
    {
      "epoch": 1.9019292604501608,
      "grad_norm": 2.991077423095703,
      "learning_rate": 4.963680387409201e-06,
      "loss": 0.9655,
      "step": 23660
    },
    {
      "epoch": 1.902733118971061,
      "grad_norm": 2.18481183052063,
      "learning_rate": 4.9233252623083135e-06,
      "loss": 0.917,
      "step": 23670
    },
    {
      "epoch": 1.9035369774919615,
      "grad_norm": 1.5696027278900146,
      "learning_rate": 4.882970137207426e-06,
      "loss": 0.9889,
      "step": 23680
    },
    {
      "epoch": 1.9043408360128617,
      "grad_norm": 1.5635316371917725,
      "learning_rate": 4.842615012106538e-06,
      "loss": 1.0119,
      "step": 23690
    },
    {
      "epoch": 1.905144694533762,
      "grad_norm": 2.4708731174468994,
      "learning_rate": 4.80225988700565e-06,
      "loss": 0.9785,
      "step": 23700
    },
    {
      "epoch": 1.9059485530546625,
      "grad_norm": 1.5477855205535889,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 0.9873,
      "step": 23710
    },
    {
      "epoch": 1.9067524115755627,
      "grad_norm": 1.5106024742126465,
      "learning_rate": 4.721549636803875e-06,
      "loss": 1.0617,
      "step": 23720
    },
    {
      "epoch": 1.907556270096463,
      "grad_norm": 2.2407634258270264,
      "learning_rate": 4.681194511702987e-06,
      "loss": 1.0268,
      "step": 23730
    },
    {
      "epoch": 1.9083601286173635,
      "grad_norm": 1.9189608097076416,
      "learning_rate": 4.640839386602098e-06,
      "loss": 1.0089,
      "step": 23740
    },
    {
      "epoch": 1.9091639871382635,
      "grad_norm": 2.416452407836914,
      "learning_rate": 4.600484261501211e-06,
      "loss": 0.9246,
      "step": 23750
    },
    {
      "epoch": 1.909967845659164,
      "grad_norm": 3.1344220638275146,
      "learning_rate": 4.560129136400323e-06,
      "loss": 1.1536,
      "step": 23760
    },
    {
      "epoch": 1.9107717041800643,
      "grad_norm": 4.287921905517578,
      "learning_rate": 4.519774011299436e-06,
      "loss": 1.0714,
      "step": 23770
    },
    {
      "epoch": 1.9115755627009645,
      "grad_norm": 2.1450698375701904,
      "learning_rate": 4.479418886198548e-06,
      "loss": 1.1106,
      "step": 23780
    },
    {
      "epoch": 1.912379421221865,
      "grad_norm": 1.7077056169509888,
      "learning_rate": 4.439063761097659e-06,
      "loss": 1.0095,
      "step": 23790
    },
    {
      "epoch": 1.9131832797427653,
      "grad_norm": 1.9565300941467285,
      "learning_rate": 4.398708635996772e-06,
      "loss": 0.8488,
      "step": 23800
    },
    {
      "epoch": 1.9139871382636655,
      "grad_norm": 2.304868221282959,
      "learning_rate": 4.358353510895884e-06,
      "loss": 1.1288,
      "step": 23810
    },
    {
      "epoch": 1.914790996784566,
      "grad_norm": 1.3317753076553345,
      "learning_rate": 4.317998385794996e-06,
      "loss": 0.9373,
      "step": 23820
    },
    {
      "epoch": 1.9155948553054662,
      "grad_norm": 2.6593315601348877,
      "learning_rate": 4.277643260694109e-06,
      "loss": 1.0363,
      "step": 23830
    },
    {
      "epoch": 1.9163987138263665,
      "grad_norm": 2.0847318172454834,
      "learning_rate": 4.23728813559322e-06,
      "loss": 1.0535,
      "step": 23840
    },
    {
      "epoch": 1.917202572347267,
      "grad_norm": 1.3046883344650269,
      "learning_rate": 4.1969330104923325e-06,
      "loss": 0.8945,
      "step": 23850
    },
    {
      "epoch": 1.9180064308681672,
      "grad_norm": 1.510403037071228,
      "learning_rate": 4.156577885391445e-06,
      "loss": 1.064,
      "step": 23860
    },
    {
      "epoch": 1.9188102893890675,
      "grad_norm": 1.7287099361419678,
      "learning_rate": 4.116222760290557e-06,
      "loss": 1.0929,
      "step": 23870
    },
    {
      "epoch": 1.919614147909968,
      "grad_norm": 2.49249267578125,
      "learning_rate": 4.075867635189669e-06,
      "loss": 0.9446,
      "step": 23880
    },
    {
      "epoch": 1.920418006430868,
      "grad_norm": 1.937441349029541,
      "learning_rate": 4.035512510088781e-06,
      "loss": 1.106,
      "step": 23890
    },
    {
      "epoch": 1.9212218649517685,
      "grad_norm": 1.676645278930664,
      "learning_rate": 3.9951573849878936e-06,
      "loss": 1.0443,
      "step": 23900
    },
    {
      "epoch": 1.9220257234726688,
      "grad_norm": 1.4668079614639282,
      "learning_rate": 3.954802259887006e-06,
      "loss": 1.0366,
      "step": 23910
    },
    {
      "epoch": 1.922829581993569,
      "grad_norm": 1.5843942165374756,
      "learning_rate": 3.914447134786118e-06,
      "loss": 0.9147,
      "step": 23920
    },
    {
      "epoch": 1.9236334405144695,
      "grad_norm": 1.6224534511566162,
      "learning_rate": 3.87409200968523e-06,
      "loss": 0.9329,
      "step": 23930
    },
    {
      "epoch": 1.9244372990353698,
      "grad_norm": 1.3980032205581665,
      "learning_rate": 3.833736884584342e-06,
      "loss": 0.9348,
      "step": 23940
    },
    {
      "epoch": 1.92524115755627,
      "grad_norm": 1.8087605237960815,
      "learning_rate": 3.7933817594834546e-06,
      "loss": 1.0116,
      "step": 23950
    },
    {
      "epoch": 1.9260450160771705,
      "grad_norm": 2.1880173683166504,
      "learning_rate": 3.7530266343825673e-06,
      "loss": 1.0284,
      "step": 23960
    },
    {
      "epoch": 1.9268488745980707,
      "grad_norm": 2.184781789779663,
      "learning_rate": 3.712671509281679e-06,
      "loss": 0.889,
      "step": 23970
    },
    {
      "epoch": 1.927652733118971,
      "grad_norm": 2.029803991317749,
      "learning_rate": 3.6723163841807913e-06,
      "loss": 1.0188,
      "step": 23980
    },
    {
      "epoch": 1.9284565916398715,
      "grad_norm": 1.5179870128631592,
      "learning_rate": 3.631961259079903e-06,
      "loss": 0.8826,
      "step": 23990
    },
    {
      "epoch": 1.9292604501607717,
      "grad_norm": 2.0556914806365967,
      "learning_rate": 3.5916061339790157e-06,
      "loss": 0.9491,
      "step": 24000
    },
    {
      "epoch": 1.9292604501607717,
      "eval_yahma/alpaca-cleaned_loss": 1.203926920890808,
      "eval_yahma/alpaca-cleaned_runtime": 115.7109,
      "eval_yahma/alpaca-cleaned_samples_per_second": 17.284,
      "eval_yahma/alpaca-cleaned_steps_per_second": 2.161,
      "step": 24000
    },
    {
      "epoch": 1.930064308681672,
      "grad_norm": 1.6474958658218384,
      "learning_rate": 3.551251008878128e-06,
      "loss": 0.9901,
      "step": 24010
    },
    {
      "epoch": 1.9308681672025725,
      "grad_norm": 1.4724137783050537,
      "learning_rate": 3.5108958837772397e-06,
      "loss": 1.0064,
      "step": 24020
    },
    {
      "epoch": 1.9316720257234725,
      "grad_norm": 1.539610505104065,
      "learning_rate": 3.4705407586763523e-06,
      "loss": 0.9541,
      "step": 24030
    },
    {
      "epoch": 1.932475884244373,
      "grad_norm": 1.5626659393310547,
      "learning_rate": 3.430185633575464e-06,
      "loss": 0.9884,
      "step": 24040
    },
    {
      "epoch": 1.9332797427652733,
      "grad_norm": 2.267127275466919,
      "learning_rate": 3.3898305084745763e-06,
      "loss": 0.9561,
      "step": 24050
    },
    {
      "epoch": 1.9340836012861735,
      "grad_norm": 1.9397815465927124,
      "learning_rate": 3.349475383373689e-06,
      "loss": 1.0845,
      "step": 24060
    },
    {
      "epoch": 1.934887459807074,
      "grad_norm": 1.6136093139648438,
      "learning_rate": 3.3091202582728007e-06,
      "loss": 0.9394,
      "step": 24070
    },
    {
      "epoch": 1.9356913183279743,
      "grad_norm": 1.8694920539855957,
      "learning_rate": 3.268765133171913e-06,
      "loss": 0.9832,
      "step": 24080
    },
    {
      "epoch": 1.9364951768488745,
      "grad_norm": 1.9546782970428467,
      "learning_rate": 3.2284100080710247e-06,
      "loss": 1.0085,
      "step": 24090
    },
    {
      "epoch": 1.937299035369775,
      "grad_norm": 1.237918496131897,
      "learning_rate": 3.1880548829701374e-06,
      "loss": 1.0026,
      "step": 24100
    },
    {
      "epoch": 1.9381028938906752,
      "grad_norm": 1.7807506322860718,
      "learning_rate": 3.1476997578692496e-06,
      "loss": 0.9878,
      "step": 24110
    },
    {
      "epoch": 1.9389067524115755,
      "grad_norm": 1.604943037033081,
      "learning_rate": 3.107344632768362e-06,
      "loss": 1.0486,
      "step": 24120
    },
    {
      "epoch": 1.939710610932476,
      "grad_norm": 1.4765275716781616,
      "learning_rate": 3.066989507667474e-06,
      "loss": 0.9899,
      "step": 24130
    },
    {
      "epoch": 1.9405144694533762,
      "grad_norm": 2.7262089252471924,
      "learning_rate": 3.0266343825665862e-06,
      "loss": 0.9322,
      "step": 24140
    },
    {
      "epoch": 1.9413183279742765,
      "grad_norm": 2.0875954627990723,
      "learning_rate": 2.9862792574656984e-06,
      "loss": 1.0717,
      "step": 24150
    },
    {
      "epoch": 1.942122186495177,
      "grad_norm": 1.2878005504608154,
      "learning_rate": 2.9459241323648102e-06,
      "loss": 0.9872,
      "step": 24160
    },
    {
      "epoch": 1.942926045016077,
      "grad_norm": 2.3931329250335693,
      "learning_rate": 2.905569007263923e-06,
      "loss": 1.009,
      "step": 24170
    },
    {
      "epoch": 1.9437299035369775,
      "grad_norm": 2.563086748123169,
      "learning_rate": 2.865213882163035e-06,
      "loss": 1.098,
      "step": 24180
    },
    {
      "epoch": 1.944533762057878,
      "grad_norm": 1.4600781202316284,
      "learning_rate": 2.824858757062147e-06,
      "loss": 1.0331,
      "step": 24190
    },
    {
      "epoch": 1.945337620578778,
      "grad_norm": 2.167815685272217,
      "learning_rate": 2.784503631961259e-06,
      "loss": 1.0031,
      "step": 24200
    },
    {
      "epoch": 1.9461414790996785,
      "grad_norm": 1.4191280603408813,
      "learning_rate": 2.7441485068603713e-06,
      "loss": 0.962,
      "step": 24210
    },
    {
      "epoch": 1.9469453376205788,
      "grad_norm": 2.354363441467285,
      "learning_rate": 2.7037933817594835e-06,
      "loss": 1.054,
      "step": 24220
    },
    {
      "epoch": 1.947749196141479,
      "grad_norm": 1.9436718225479126,
      "learning_rate": 2.6634382566585957e-06,
      "loss": 1.0494,
      "step": 24230
    },
    {
      "epoch": 1.9485530546623795,
      "grad_norm": 1.4945588111877441,
      "learning_rate": 2.623083131557708e-06,
      "loss": 0.9375,
      "step": 24240
    },
    {
      "epoch": 1.9493569131832797,
      "grad_norm": 2.525233268737793,
      "learning_rate": 2.58272800645682e-06,
      "loss": 1.0606,
      "step": 24250
    },
    {
      "epoch": 1.95016077170418,
      "grad_norm": 2.243128776550293,
      "learning_rate": 2.5423728813559323e-06,
      "loss": 1.0543,
      "step": 24260
    },
    {
      "epoch": 1.9509646302250805,
      "grad_norm": 1.4152499437332153,
      "learning_rate": 2.5020177562550446e-06,
      "loss": 1.0039,
      "step": 24270
    },
    {
      "epoch": 1.9517684887459807,
      "grad_norm": 1.6305155754089355,
      "learning_rate": 2.4616626311541568e-06,
      "loss": 0.97,
      "step": 24280
    },
    {
      "epoch": 1.952572347266881,
      "grad_norm": 1.5360208749771118,
      "learning_rate": 2.421307506053269e-06,
      "loss": 0.9765,
      "step": 24290
    },
    {
      "epoch": 1.9533762057877815,
      "grad_norm": 2.4453868865966797,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 1.0101,
      "step": 24300
    },
    {
      "epoch": 1.9541800643086815,
      "grad_norm": 1.925690770149231,
      "learning_rate": 2.3405972558514934e-06,
      "loss": 1.1117,
      "step": 24310
    },
    {
      "epoch": 1.954983922829582,
      "grad_norm": 2.877366542816162,
      "learning_rate": 2.3002421307506056e-06,
      "loss": 1.0363,
      "step": 24320
    },
    {
      "epoch": 1.9557877813504825,
      "grad_norm": 1.9311004877090454,
      "learning_rate": 2.259887005649718e-06,
      "loss": 0.9606,
      "step": 24330
    },
    {
      "epoch": 1.9565916398713825,
      "grad_norm": 2.90724515914917,
      "learning_rate": 2.2195318805488296e-06,
      "loss": 0.8997,
      "step": 24340
    },
    {
      "epoch": 1.957395498392283,
      "grad_norm": 2.0841379165649414,
      "learning_rate": 2.179176755447942e-06,
      "loss": 1.0371,
      "step": 24350
    },
    {
      "epoch": 1.9581993569131833,
      "grad_norm": 2.0677711963653564,
      "learning_rate": 2.1388216303470545e-06,
      "loss": 1.0479,
      "step": 24360
    },
    {
      "epoch": 1.9590032154340835,
      "grad_norm": 1.8002878427505493,
      "learning_rate": 2.0984665052461662e-06,
      "loss": 0.9957,
      "step": 24370
    },
    {
      "epoch": 1.959807073954984,
      "grad_norm": 1.5182082653045654,
      "learning_rate": 2.0581113801452785e-06,
      "loss": 0.9324,
      "step": 24380
    },
    {
      "epoch": 1.9606109324758842,
      "grad_norm": 1.88400399684906,
      "learning_rate": 2.0177562550443907e-06,
      "loss": 1.0915,
      "step": 24390
    },
    {
      "epoch": 1.9614147909967845,
      "grad_norm": 2.4193053245544434,
      "learning_rate": 1.977401129943503e-06,
      "loss": 1.0839,
      "step": 24400
    },
    {
      "epoch": 1.962218649517685,
      "grad_norm": 1.627624273300171,
      "learning_rate": 1.937046004842615e-06,
      "loss": 1.0113,
      "step": 24410
    },
    {
      "epoch": 1.9630225080385852,
      "grad_norm": 1.4581716060638428,
      "learning_rate": 1.8966908797417273e-06,
      "loss": 1.0207,
      "step": 24420
    },
    {
      "epoch": 1.9638263665594855,
      "grad_norm": 2.1971490383148193,
      "learning_rate": 1.8563357546408395e-06,
      "loss": 0.9363,
      "step": 24430
    },
    {
      "epoch": 1.964630225080386,
      "grad_norm": 1.8994516134262085,
      "learning_rate": 1.8159806295399515e-06,
      "loss": 0.9914,
      "step": 24440
    },
    {
      "epoch": 1.965434083601286,
      "grad_norm": 1.8573304414749146,
      "learning_rate": 1.775625504439064e-06,
      "loss": 0.9137,
      "step": 24450
    },
    {
      "epoch": 1.9662379421221865,
      "grad_norm": 1.2724213600158691,
      "learning_rate": 1.7352703793381762e-06,
      "loss": 0.9722,
      "step": 24460
    },
    {
      "epoch": 1.967041800643087,
      "grad_norm": 1.9029781818389893,
      "learning_rate": 1.6949152542372882e-06,
      "loss": 0.9432,
      "step": 24470
    },
    {
      "epoch": 1.967845659163987,
      "grad_norm": 1.6016614437103271,
      "learning_rate": 1.6545601291364004e-06,
      "loss": 1.0239,
      "step": 24480
    },
    {
      "epoch": 1.9686495176848875,
      "grad_norm": 1.7589869499206543,
      "learning_rate": 1.6142050040355124e-06,
      "loss": 1.0125,
      "step": 24490
    },
    {
      "epoch": 1.9694533762057878,
      "grad_norm": 1.6908783912658691,
      "learning_rate": 1.5738498789346248e-06,
      "loss": 0.8737,
      "step": 24500
    },
    {
      "epoch": 1.970257234726688,
      "grad_norm": 1.8626710176467896,
      "learning_rate": 1.533494753833737e-06,
      "loss": 0.9974,
      "step": 24510
    },
    {
      "epoch": 1.9710610932475885,
      "grad_norm": 1.257219672203064,
      "learning_rate": 1.4931396287328492e-06,
      "loss": 1.039,
      "step": 24520
    },
    {
      "epoch": 1.9718649517684887,
      "grad_norm": 1.5754961967468262,
      "learning_rate": 1.4527845036319614e-06,
      "loss": 0.967,
      "step": 24530
    },
    {
      "epoch": 1.972668810289389,
      "grad_norm": 2.0581512451171875,
      "learning_rate": 1.4124293785310734e-06,
      "loss": 1.0605,
      "step": 24540
    },
    {
      "epoch": 1.9734726688102895,
      "grad_norm": 1.8139474391937256,
      "learning_rate": 1.3720742534301856e-06,
      "loss": 0.9943,
      "step": 24550
    },
    {
      "epoch": 1.9742765273311897,
      "grad_norm": 1.691847562789917,
      "learning_rate": 1.3317191283292979e-06,
      "loss": 0.9957,
      "step": 24560
    },
    {
      "epoch": 1.97508038585209,
      "grad_norm": 3.390103578567505,
      "learning_rate": 1.29136400322841e-06,
      "loss": 0.9498,
      "step": 24570
    },
    {
      "epoch": 1.9758842443729905,
      "grad_norm": 1.823568344116211,
      "learning_rate": 1.2510088781275223e-06,
      "loss": 1.0667,
      "step": 24580
    },
    {
      "epoch": 1.9766881028938905,
      "grad_norm": 4.890821933746338,
      "learning_rate": 1.2106537530266345e-06,
      "loss": 0.9983,
      "step": 24590
    },
    {
      "epoch": 1.977491961414791,
      "grad_norm": 2.4225986003875732,
      "learning_rate": 1.1702986279257467e-06,
      "loss": 1.0648,
      "step": 24600
    },
    {
      "epoch": 1.9782958199356915,
      "grad_norm": 1.9177806377410889,
      "learning_rate": 1.129943502824859e-06,
      "loss": 1.1094,
      "step": 24610
    },
    {
      "epoch": 1.9790996784565915,
      "grad_norm": 2.4320592880249023,
      "learning_rate": 1.089588377723971e-06,
      "loss": 1.06,
      "step": 24620
    },
    {
      "epoch": 1.979903536977492,
      "grad_norm": 1.7749550342559814,
      "learning_rate": 1.0492332526230831e-06,
      "loss": 1.0337,
      "step": 24630
    },
    {
      "epoch": 1.9807073954983923,
      "grad_norm": 1.7348616123199463,
      "learning_rate": 1.0088781275221953e-06,
      "loss": 0.8639,
      "step": 24640
    },
    {
      "epoch": 1.9815112540192925,
      "grad_norm": 2.343806028366089,
      "learning_rate": 9.685230024213075e-07,
      "loss": 0.9784,
      "step": 24650
    },
    {
      "epoch": 1.982315112540193,
      "grad_norm": 1.7652950286865234,
      "learning_rate": 9.281678773204198e-07,
      "loss": 0.977,
      "step": 24660
    },
    {
      "epoch": 1.9831189710610932,
      "grad_norm": 1.9320908784866333,
      "learning_rate": 8.87812752219532e-07,
      "loss": 0.9649,
      "step": 24670
    },
    {
      "epoch": 1.9839228295819935,
      "grad_norm": 2.151548147201538,
      "learning_rate": 8.474576271186441e-07,
      "loss": 0.9741,
      "step": 24680
    },
    {
      "epoch": 1.984726688102894,
      "grad_norm": 1.9024114608764648,
      "learning_rate": 8.071025020177562e-07,
      "loss": 1.008,
      "step": 24690
    },
    {
      "epoch": 1.9855305466237942,
      "grad_norm": 1.5782983303070068,
      "learning_rate": 7.667473769168685e-07,
      "loss": 0.9779,
      "step": 24700
    },
    {
      "epoch": 1.9863344051446945,
      "grad_norm": 3.566256046295166,
      "learning_rate": 7.263922518159807e-07,
      "loss": 0.9926,
      "step": 24710
    },
    {
      "epoch": 1.987138263665595,
      "grad_norm": 1.6482263803482056,
      "learning_rate": 6.860371267150928e-07,
      "loss": 0.9734,
      "step": 24720
    },
    {
      "epoch": 1.987942122186495,
      "grad_norm": 2.0280840396881104,
      "learning_rate": 6.45682001614205e-07,
      "loss": 1.0214,
      "step": 24730
    },
    {
      "epoch": 1.9887459807073955,
      "grad_norm": 1.5748929977416992,
      "learning_rate": 6.053268765133172e-07,
      "loss": 1.018,
      "step": 24740
    },
    {
      "epoch": 1.989549839228296,
      "grad_norm": 1.5059864521026611,
      "learning_rate": 5.649717514124295e-07,
      "loss": 1.0083,
      "step": 24750
    },
    {
      "epoch": 1.990353697749196,
      "grad_norm": 1.619263768196106,
      "learning_rate": 5.246166263115416e-07,
      "loss": 0.9868,
      "step": 24760
    },
    {
      "epoch": 1.9911575562700965,
      "grad_norm": 1.563617467880249,
      "learning_rate": 4.842615012106538e-07,
      "loss": 0.952,
      "step": 24770
    },
    {
      "epoch": 1.9919614147909968,
      "grad_norm": 1.829676866531372,
      "learning_rate": 4.43906376109766e-07,
      "loss": 1.0386,
      "step": 24780
    },
    {
      "epoch": 1.992765273311897,
      "grad_norm": 3.145707130432129,
      "learning_rate": 4.035512510088781e-07,
      "loss": 0.9683,
      "step": 24790
    },
    {
      "epoch": 1.9935691318327975,
      "grad_norm": 1.8739086389541626,
      "learning_rate": 3.6319612590799036e-07,
      "loss": 0.9351,
      "step": 24800
    },
    {
      "epoch": 1.9943729903536977,
      "grad_norm": 1.6382272243499756,
      "learning_rate": 3.228410008071025e-07,
      "loss": 0.9973,
      "step": 24810
    },
    {
      "epoch": 1.995176848874598,
      "grad_norm": 1.8064838647842407,
      "learning_rate": 2.8248587570621473e-07,
      "loss": 1.0362,
      "step": 24820
    },
    {
      "epoch": 1.9959807073954985,
      "grad_norm": 1.9633867740631104,
      "learning_rate": 2.421307506053269e-07,
      "loss": 0.9698,
      "step": 24830
    },
    {
      "epoch": 1.9967845659163987,
      "grad_norm": 1.9776709079742432,
      "learning_rate": 2.0177562550443905e-07,
      "loss": 1.0884,
      "step": 24840
    },
    {
      "epoch": 1.997588424437299,
      "grad_norm": 2.630523681640625,
      "learning_rate": 1.6142050040355126e-07,
      "loss": 1.152,
      "step": 24850
    },
    {
      "epoch": 1.9983922829581995,
      "grad_norm": 1.5119197368621826,
      "learning_rate": 1.2106537530266344e-07,
      "loss": 1.0154,
      "step": 24860
    },
    {
      "epoch": 1.9991961414790995,
      "grad_norm": 2.289841651916504,
      "learning_rate": 8.071025020177563e-08,
      "loss": 1.0503,
      "step": 24870
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.790726661682129,
      "learning_rate": 4.0355125100887814e-08,
      "loss": 0.9707,
      "step": 24880
    }
  ],
  "logging_steps": 10,
  "max_steps": 24880,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.698693203135037e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}