{
  "best_metric": 0.9957587181903864,
  "best_model_checkpoint": "beit-large-patch16-224-finetuned-BreastCancer-Classification-BreakHis-AH-60-20-20-Shuffled/checkpoint-2985",
  "epoch": 15.0,
  "global_step": 2985,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 9.304056568663938e-09,
      "loss": 1.0321,
      "step": 5
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.8608113137327877e-08,
      "loss": 0.9905,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.7912169705991817e-08,
      "loss": 0.991,
      "step": 15
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.7216226274655753e-08,
      "loss": 1.1258,
      "step": 20
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.652028284331969e-08,
      "loss": 0.9302,
      "step": 25
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.5824339411983633e-08,
      "loss": 1.0139,
      "step": 30
    },
    {
      "epoch": 0.18,
      "learning_rate": 6.512839598064757e-08,
      "loss": 1.0124,
      "step": 35
    },
    {
      "epoch": 0.2,
      "learning_rate": 7.443245254931151e-08,
      "loss": 0.9788,
      "step": 40
    },
    {
      "epoch": 0.23,
      "learning_rate": 8.373650911797545e-08,
      "loss": 0.9479,
      "step": 45
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.304056568663937e-08,
      "loss": 0.9029,
      "step": 50
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.0234462225530333e-07,
      "loss": 0.9973,
      "step": 55
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.1164867882396727e-07,
      "loss": 0.8983,
      "step": 60
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.209527353926312e-07,
      "loss": 0.9177,
      "step": 65
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.3025679196129513e-07,
      "loss": 0.9102,
      "step": 70
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.3956084852995907e-07,
      "loss": 0.8791,
      "step": 75
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.4886490509862301e-07,
      "loss": 0.8303,
      "step": 80
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.5816896166728695e-07,
      "loss": 0.7997,
      "step": 85
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.674730182359509e-07,
      "loss": 0.8353,
      "step": 90
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.767770748046148e-07,
      "loss": 0.7583,
      "step": 95
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.8608113137327875e-07,
      "loss": 0.7995,
      "step": 100
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.953851879419427e-07,
      "loss": 0.7684,
      "step": 105
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.0468924451060665e-07,
      "loss": 0.7371,
      "step": 110
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.139933010792706e-07,
      "loss": 0.752,
      "step": 115
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.2329735764793453e-07,
      "loss": 0.8275,
      "step": 120
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.3260141421659845e-07,
      "loss": 0.7244,
      "step": 125
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.419054707852624e-07,
      "loss": 0.8018,
      "step": 130
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.5120952735392633e-07,
      "loss": 0.7265,
      "step": 135
    },
    {
      "epoch": 0.7,
      "learning_rate": 2.6051358392259027e-07,
      "loss": 0.7066,
      "step": 140
    },
    {
      "epoch": 0.73,
      "learning_rate": 2.698176404912542e-07,
      "loss": 0.6795,
      "step": 145
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.7912169705991815e-07,
      "loss": 0.6628,
      "step": 150
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.884257536285821e-07,
      "loss": 0.6541,
      "step": 155
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.9772981019724603e-07,
      "loss": 0.5732,
      "step": 160
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.0703386676590997e-07,
      "loss": 0.699,
      "step": 165
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.163379233345739e-07,
      "loss": 0.5896,
      "step": 170
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.2564197990323785e-07,
      "loss": 0.627,
      "step": 175
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.349460364719018e-07,
      "loss": 0.6103,
      "step": 180
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.4425009304056573e-07,
      "loss": 0.572,
      "step": 185
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.535541496092296e-07,
      "loss": 0.5898,
      "step": 190
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.628582061778936e-07,
      "loss": 0.5847,
      "step": 195
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8030160226201697,
      "eval_loss": 0.4639509618282318,
      "eval_runtime": 546.0635,
      "eval_samples_per_second": 3.886,
      "eval_steps_per_second": 0.244,
      "step": 199
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.721622627465575e-07,
      "loss": 0.5601,
      "step": 200
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.814663193152215e-07,
      "loss": 0.5478,
      "step": 205
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.907703758838854e-07,
      "loss": 0.5345,
      "step": 210
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.0007443245254937e-07,
      "loss": 0.4652,
      "step": 215
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.093784890212133e-07,
      "loss": 0.4762,
      "step": 220
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.186825455898772e-07,
      "loss": 0.5008,
      "step": 225
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.279866021585412e-07,
      "loss": 0.515,
      "step": 230
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.372906587272051e-07,
      "loss": 0.4674,
      "step": 235
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.4659471529586907e-07,
      "loss": 0.4464,
      "step": 240
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.5589877186453295e-07,
      "loss": 0.4493,
      "step": 245
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.652028284331969e-07,
      "loss": 0.4473,
      "step": 250
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.7450688500186083e-07,
      "loss": 0.4283,
      "step": 255
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.838109415705248e-07,
      "loss": 0.4304,
      "step": 260
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.931149981391888e-07,
      "loss": 0.3941,
      "step": 265
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.024190547078527e-07,
      "loss": 0.3848,
      "step": 270
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.117231112765166e-07,
      "loss": 0.4018,
      "step": 275
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.210271678451805e-07,
      "loss": 0.4228,
      "step": 280
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.303312244138445e-07,
      "loss": 0.3959,
      "step": 285
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.396352809825084e-07,
      "loss": 0.3868,
      "step": 290
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.489393375511724e-07,
      "loss": 0.4186,
      "step": 295
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.582433941198363e-07,
      "loss": 0.3898,
      "step": 300
    },
    {
      "epoch": 1.53,
      "learning_rate": 5.675474506885002e-07,
      "loss": 0.3181,
      "step": 305
    },
    {
      "epoch": 1.56,
      "learning_rate": 5.768515072571642e-07,
      "loss": 0.3409,
      "step": 310
    },
    {
      "epoch": 1.58,
      "learning_rate": 5.861555638258281e-07,
      "loss": 0.2985,
      "step": 315
    },
    {
      "epoch": 1.61,
      "learning_rate": 5.954596203944921e-07,
      "loss": 0.2872,
      "step": 320
    },
    {
      "epoch": 1.63,
      "learning_rate": 6.047636769631559e-07,
      "loss": 0.295,
      "step": 325
    },
    {
      "epoch": 1.66,
      "learning_rate": 6.140677335318199e-07,
      "loss": 0.2724,
      "step": 330
    },
    {
      "epoch": 1.68,
      "learning_rate": 6.233717901004838e-07,
      "loss": 0.258,
      "step": 335
    },
    {
      "epoch": 1.71,
      "learning_rate": 6.326758466691478e-07,
      "loss": 0.2898,
      "step": 340
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.419799032378117e-07,
      "loss": 0.2536,
      "step": 345
    },
    {
      "epoch": 1.76,
      "learning_rate": 6.512839598064757e-07,
      "loss": 0.2835,
      "step": 350
    },
    {
      "epoch": 1.78,
      "learning_rate": 6.605880163751397e-07,
      "loss": 0.2945,
      "step": 355
    },
    {
      "epoch": 1.81,
      "learning_rate": 6.698920729438036e-07,
      "loss": 0.2847,
      "step": 360
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.791961295124675e-07,
      "loss": 0.2896,
      "step": 365
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.885001860811315e-07,
      "loss": 0.2419,
      "step": 370
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.978042426497953e-07,
      "loss": 0.2376,
      "step": 375
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.071082992184592e-07,
      "loss": 0.2615,
      "step": 380
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.164123557871232e-07,
      "loss": 0.2602,
      "step": 385
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.257164123557872e-07,
      "loss": 0.2455,
      "step": 390
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.350204689244511e-07,
      "loss": 0.2856,
      "step": 395
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9354382657869934,
      "eval_loss": 0.17528533935546875,
      "eval_runtime": 86.4492,
      "eval_samples_per_second": 24.546,
      "eval_steps_per_second": 1.538,
      "step": 398
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.44324525493115e-07,
      "loss": 0.2498,
      "step": 400
    },
    {
      "epoch": 2.04,
      "learning_rate": 7.536285820617791e-07,
      "loss": 0.2269,
      "step": 405
    },
    {
      "epoch": 2.06,
      "learning_rate": 7.62932638630443e-07,
      "loss": 0.2345,
      "step": 410
    },
    {
      "epoch": 2.09,
      "learning_rate": 7.722366951991069e-07,
      "loss": 0.2149,
      "step": 415
    },
    {
      "epoch": 2.11,
      "learning_rate": 7.815407517677707e-07,
      "loss": 0.1939,
      "step": 420
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.908448083364347e-07,
      "loss": 0.2543,
      "step": 425
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.001488649050987e-07,
      "loss": 0.1779,
      "step": 430
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.094529214737626e-07,
      "loss": 0.1749,
      "step": 435
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.187569780424266e-07,
      "loss": 0.2132,
      "step": 440
    },
    {
      "epoch": 2.24,
      "learning_rate": 8.280610346110905e-07,
      "loss": 0.2808,
      "step": 445
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.373650911797544e-07,
      "loss": 0.1961,
      "step": 450
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.466691477484183e-07,
      "loss": 0.1777,
      "step": 455
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.559732043170824e-07,
      "loss": 0.1996,
      "step": 460
    },
    {
      "epoch": 2.34,
      "learning_rate": 8.652772608857463e-07,
      "loss": 0.1759,
      "step": 465
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.745813174544101e-07,
      "loss": 0.2076,
      "step": 470
    },
    {
      "epoch": 2.39,
      "learning_rate": 8.838853740230742e-07,
      "loss": 0.2393,
      "step": 475
    },
    {
      "epoch": 2.41,
      "learning_rate": 8.931894305917381e-07,
      "loss": 0.1882,
      "step": 480
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.02493487160402e-07,
      "loss": 0.1938,
      "step": 485
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.117975437290659e-07,
      "loss": 0.2034,
      "step": 490
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.211016002977299e-07,
      "loss": 0.2372,
      "step": 495
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.304056568663938e-07,
      "loss": 0.1723,
      "step": 500
    },
    {
      "epoch": 2.54,
      "learning_rate": 9.397097134350578e-07,
      "loss": 0.2127,
      "step": 505
    },
    {
      "epoch": 2.56,
      "learning_rate": 9.490137700037217e-07,
      "loss": 0.1834,
      "step": 510
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.583178265723858e-07,
      "loss": 0.1707,
      "step": 515
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.676218831410495e-07,
      "loss": 0.161,
      "step": 520
    },
    {
      "epoch": 2.64,
      "learning_rate": 9.769259397097135e-07,
      "loss": 0.3031,
      "step": 525
    },
    {
      "epoch": 2.66,
      "learning_rate": 9.862299962783775e-07,
      "loss": 0.1769,
      "step": 530
    },
    {
      "epoch": 2.69,
      "learning_rate": 9.955340528470413e-07,
      "loss": 0.1879,
      "step": 535
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.0048381094157053e-06,
      "loss": 0.1567,
      "step": 540
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.0141421659843693e-06,
      "loss": 0.2025,
      "step": 545
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.0234462225530333e-06,
      "loss": 0.2344,
      "step": 550
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.032750279121697e-06,
      "loss": 0.2667,
      "step": 555
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.042054335690361e-06,
      "loss": 0.175,
      "step": 560
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.051358392259025e-06,
      "loss": 0.2166,
      "step": 565
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.060662448827689e-06,
      "loss": 0.1611,
      "step": 570
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.0699665053963528e-06,
      "loss": 0.2097,
      "step": 575
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.0792705619650168e-06,
      "loss": 0.2383,
      "step": 580
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0885746185336808e-06,
      "loss": 0.1844,
      "step": 585
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.0978786751023448e-06,
      "loss": 0.1883,
      "step": 590
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.1071827316710086e-06,
      "loss": 0.156,
      "step": 595
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9552309142318567,
      "eval_loss": 0.11788399517536163,
      "eval_runtime": 86.2903,
      "eval_samples_per_second": 24.591,
      "eval_steps_per_second": 1.541,
      "step": 597
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.1164867882396726e-06,
      "loss": 0.1402,
      "step": 600
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.1257908448083366e-06,
      "loss": 0.2268,
      "step": 605
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.1350949013770004e-06,
      "loss": 0.1681,
      "step": 610
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.1443989579456644e-06,
      "loss": 0.1039,
      "step": 615
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.1537030145143283e-06,
      "loss": 0.099,
      "step": 620
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.1630070710829923e-06,
      "loss": 0.1558,
      "step": 625
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.1723111276516561e-06,
      "loss": 0.1635,
      "step": 630
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.1816151842203201e-06,
      "loss": 0.2037,
      "step": 635
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.1909192407889841e-06,
      "loss": 0.1673,
      "step": 640
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.200223297357648e-06,
      "loss": 0.1065,
      "step": 645
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.2095273539263119e-06,
      "loss": 0.1771,
      "step": 650
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.2188314104949759e-06,
      "loss": 0.1563,
      "step": 655
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.2281354670636399e-06,
      "loss": 0.178,
      "step": 660
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.2374395236323039e-06,
      "loss": 0.1881,
      "step": 665
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.2467435802009676e-06,
      "loss": 0.2181,
      "step": 670
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.2560476367696316e-06,
      "loss": 0.1146,
      "step": 675
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.2653516933382956e-06,
      "loss": 0.1517,
      "step": 680
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.2746557499069594e-06,
      "loss": 0.1578,
      "step": 685
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.2839598064756234e-06,
      "loss": 0.1616,
      "step": 690
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.2932638630442874e-06,
      "loss": 0.1427,
      "step": 695
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.3025679196129514e-06,
      "loss": 0.1389,
      "step": 700
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.3118719761816154e-06,
      "loss": 0.1678,
      "step": 705
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.3211760327502794e-06,
      "loss": 0.0795,
      "step": 710
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.3304800893189432e-06,
      "loss": 0.1571,
      "step": 715
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.3397841458876071e-06,
      "loss": 0.1405,
      "step": 720
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.349088202456271e-06,
      "loss": 0.1477,
      "step": 725
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.358392259024935e-06,
      "loss": 0.1592,
      "step": 730
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.3676963155935991e-06,
      "loss": 0.1094,
      "step": 735
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.377000372162263e-06,
      "loss": 0.1518,
      "step": 740
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.386304428730927e-06,
      "loss": 0.0673,
      "step": 745
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.3956084852995907e-06,
      "loss": 0.1078,
      "step": 750
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.4049125418682547e-06,
      "loss": 0.1396,
      "step": 755
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4142165984369185e-06,
      "loss": 0.1748,
      "step": 760
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.4235206550055824e-06,
      "loss": 0.1354,
      "step": 765
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.4328247115742464e-06,
      "loss": 0.1234,
      "step": 770
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.4421287681429104e-06,
      "loss": 0.117,
      "step": 775
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.4514328247115744e-06,
      "loss": 0.0658,
      "step": 780
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.4607368812802384e-06,
      "loss": 0.1081,
      "step": 785
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.4700409378489022e-06,
      "loss": 0.1429,
      "step": 790
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.4793449944175662e-06,
      "loss": 0.1049,
      "step": 795
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9585296889726673,
      "eval_loss": 0.10429327189922333,
      "eval_runtime": 86.1413,
      "eval_samples_per_second": 24.634,
      "eval_steps_per_second": 1.544,
      "step": 796
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.48864905098623e-06,
      "loss": 0.061,
      "step": 800
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.497953107554894e-06,
      "loss": 0.0704,
      "step": 805
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.5072571641235582e-06,
      "loss": 0.0991,
      "step": 810
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.516561220692222e-06,
      "loss": 0.0888,
      "step": 815
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.525865277260886e-06,
      "loss": 0.1584,
      "step": 820
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.5351693338295497e-06,
      "loss": 0.1265,
      "step": 825
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.5444733903982137e-06,
      "loss": 0.1607,
      "step": 830
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.5537774469668775e-06,
      "loss": 0.114,
      "step": 835
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.5630815035355415e-06,
      "loss": 0.1134,
      "step": 840
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.5723855601042057e-06,
      "loss": 0.1024,
      "step": 845
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.5816896166728695e-06,
      "loss": 0.1006,
      "step": 850
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.5909936732415335e-06,
      "loss": 0.1013,
      "step": 855
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.6002977298101975e-06,
      "loss": 0.1359,
      "step": 860
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.6096017863788612e-06,
      "loss": 0.075,
      "step": 865
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.6189058429475252e-06,
      "loss": 0.0676,
      "step": 870
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.628209899516189e-06,
      "loss": 0.1029,
      "step": 875
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.6375139560848532e-06,
      "loss": 0.0982,
      "step": 880
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.6468180126535172e-06,
      "loss": 0.0897,
      "step": 885
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.656122069222181e-06,
      "loss": 0.0857,
      "step": 890
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.665426125790845e-06,
      "loss": 0.0976,
      "step": 895
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.6747301823595088e-06,
      "loss": 0.0929,
      "step": 900
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.6840342389281728e-06,
      "loss": 0.1258,
      "step": 905
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.6933382954968365e-06,
      "loss": 0.1,
      "step": 910
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7026423520655008e-06,
      "loss": 0.0895,
      "step": 915
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.7119464086341647e-06,
      "loss": 0.0998,
      "step": 920
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.7212504652028285e-06,
      "loss": 0.0978,
      "step": 925
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.7305545217714925e-06,
      "loss": 0.1337,
      "step": 930
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.7398585783401565e-06,
      "loss": 0.1351,
      "step": 935
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.7491626349088203e-06,
      "loss": 0.1343,
      "step": 940
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.7584666914774843e-06,
      "loss": 0.0921,
      "step": 945
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.7677707480461485e-06,
      "loss": 0.1295,
      "step": 950
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.7770748046148123e-06,
      "loss": 0.1301,
      "step": 955
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.7863788611834763e-06,
      "loss": 0.1056,
      "step": 960
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.79568291775214e-06,
      "loss": 0.1165,
      "step": 965
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.804986974320804e-06,
      "loss": 0.0901,
      "step": 970
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.8142910308894678e-06,
      "loss": 0.0674,
      "step": 975
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.8235950874581318e-06,
      "loss": 0.1295,
      "step": 980
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.8328991440267956e-06,
      "loss": 0.1309,
      "step": 985
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.8422032005954598e-06,
      "loss": 0.1613,
      "step": 990
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.8515072571641238e-06,
      "loss": 0.1399,
      "step": 995
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9759660697455231,
      "eval_loss": 0.06734614074230194,
      "eval_runtime": 86.224,
      "eval_samples_per_second": 24.61,
      "eval_steps_per_second": 1.542,
      "step": 995
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.8608113137327876e-06,
      "loss": 0.0825,
      "step": 1000
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.8701153703014516e-06,
      "loss": 0.0851,
      "step": 1005
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.8794194268701156e-06,
      "loss": 0.1067,
      "step": 1010
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.8887234834387793e-06,
      "loss": 0.0964,
      "step": 1015
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.8980275400074433e-06,
      "loss": 0.097,
      "step": 1020
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.9073315965761075e-06,
      "loss": 0.1145,
      "step": 1025
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.9166356531447715e-06,
      "loss": 0.1044,
      "step": 1030
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.925939709713435e-06,
      "loss": 0.0721,
      "step": 1035
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.935243766282099e-06,
      "loss": 0.0841,
      "step": 1040
    },
    {
      "epoch": 5.25,
      "learning_rate": 1.944547822850763e-06,
      "loss": 0.0799,
      "step": 1045
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.953851879419427e-06,
      "loss": 0.0914,
      "step": 1050
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.9631559359880906e-06,
      "loss": 0.0887,
      "step": 1055
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.972459992556755e-06,
      "loss": 0.1541,
      "step": 1060
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.981764049125419e-06,
      "loss": 0.0694,
      "step": 1065
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.9910681056940826e-06,
      "loss": 0.0976,
      "step": 1070
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.0003721622627466e-06,
      "loss": 0.0959,
      "step": 1075
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.0096762188314106e-06,
      "loss": 0.152,
      "step": 1080
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.0189802754000746e-06,
      "loss": 0.0926,
      "step": 1085
    },
    {
      "epoch": 5.48,
      "learning_rate": 2.0282843319687386e-06,
      "loss": 0.0631,
      "step": 1090
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.0375883885374026e-06,
      "loss": 0.0836,
      "step": 1095
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.0468924451060666e-06,
      "loss": 0.0454,
      "step": 1100
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.0561965016747306e-06,
      "loss": 0.099,
      "step": 1105
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.065500558243394e-06,
      "loss": 0.1053,
      "step": 1110
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.074804614812058e-06,
      "loss": 0.0905,
      "step": 1115
    },
    {
      "epoch": 5.63,
      "learning_rate": 2.084108671380722e-06,
      "loss": 0.0572,
      "step": 1120
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.093412727949386e-06,
      "loss": 0.0706,
      "step": 1125
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.10271678451805e-06,
      "loss": 0.0367,
      "step": 1130
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.112020841086714e-06,
      "loss": 0.0911,
      "step": 1135
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.121324897655378e-06,
      "loss": 0.0173,
      "step": 1140
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.1306289542240417e-06,
      "loss": 0.0539,
      "step": 1145
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.1399330107927057e-06,
      "loss": 0.0476,
      "step": 1150
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.1492370673613697e-06,
      "loss": 0.081,
      "step": 1155
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.1585411239300337e-06,
      "loss": 0.1207,
      "step": 1160
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.1678451804986976e-06,
      "loss": 0.0899,
      "step": 1165
    },
    {
      "epoch": 5.88,
      "learning_rate": 2.1771492370673616e-06,
      "loss": 0.0756,
      "step": 1170
    },
    {
      "epoch": 5.9,
      "learning_rate": 2.1864532936360256e-06,
      "loss": 0.0723,
      "step": 1175
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.1957573502046896e-06,
      "loss": 0.0478,
      "step": 1180
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.205061406773353e-06,
      "loss": 0.0873,
      "step": 1185
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.214365463342017e-06,
      "loss": 0.0423,
      "step": 1190
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9802073515551367,
      "eval_loss": 0.04545706510543823,
      "eval_runtime": 86.1094,
      "eval_samples_per_second": 24.643,
      "eval_steps_per_second": 1.545,
      "step": 1194
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.223669519910681e-06,
      "loss": 0.1062,
      "step": 1195
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.232973576479345e-06,
      "loss": 0.1422,
      "step": 1200
    },
    {
      "epoch": 6.06,
      "learning_rate": 2.242277633048009e-06,
      "loss": 0.049,
      "step": 1205
    },
    {
      "epoch": 6.08,
      "learning_rate": 2.251581689616673e-06,
      "loss": 0.0623,
      "step": 1210
    },
    {
      "epoch": 6.11,
      "learning_rate": 2.260885746185337e-06,
      "loss": 0.0888,
      "step": 1215
    },
    {
      "epoch": 6.13,
      "learning_rate": 2.2701898027540007e-06,
      "loss": 0.0561,
      "step": 1220
    },
    {
      "epoch": 6.16,
      "learning_rate": 2.2794938593226647e-06,
      "loss": 0.0618,
      "step": 1225
    },
    {
      "epoch": 6.18,
      "learning_rate": 2.2887979158913287e-06,
      "loss": 0.0997,
      "step": 1230
    },
    {
      "epoch": 6.21,
      "learning_rate": 2.2981019724599927e-06,
      "loss": 0.0918,
      "step": 1235
    },
    {
      "epoch": 6.23,
      "learning_rate": 2.3074060290286567e-06,
      "loss": 0.066,
      "step": 1240
    },
    {
      "epoch": 6.26,
      "learning_rate": 2.3167100855973207e-06,
      "loss": 0.0655,
      "step": 1245
    },
    {
      "epoch": 6.28,
      "learning_rate": 2.3260141421659847e-06,
      "loss": 0.1025,
      "step": 1250
    },
    {
      "epoch": 6.31,
      "learning_rate": 2.3353181987346487e-06,
      "loss": 0.0507,
      "step": 1255
    },
    {
      "epoch": 6.33,
      "learning_rate": 2.3446222553033122e-06,
      "loss": 0.0949,
      "step": 1260
    },
    {
      "epoch": 6.36,
      "learning_rate": 2.3539263118719762e-06,
      "loss": 0.0609,
      "step": 1265
    },
    {
      "epoch": 6.38,
      "learning_rate": 2.3632303684406402e-06,
      "loss": 0.0126,
      "step": 1270
    },
    {
      "epoch": 6.41,
      "learning_rate": 2.3725344250093042e-06,
      "loss": 0.0593,
      "step": 1275
    },
    {
      "epoch": 6.43,
      "learning_rate": 2.3818384815779682e-06,
      "loss": 0.0683,
      "step": 1280
    },
    {
      "epoch": 6.46,
      "learning_rate": 2.391142538146632e-06,
      "loss": 0.0702,
      "step": 1285
    },
    {
      "epoch": 6.48,
      "learning_rate": 2.400446594715296e-06,
      "loss": 0.0247,
      "step": 1290
    },
    {
      "epoch": 6.51,
      "learning_rate": 2.4097506512839598e-06,
      "loss": 0.0749,
      "step": 1295
    },
    {
      "epoch": 6.53,
      "learning_rate": 2.4190547078526238e-06,
      "loss": 0.0453,
      "step": 1300
    },
    {
      "epoch": 6.56,
      "learning_rate": 2.4283587644212878e-06,
      "loss": 0.0527,
      "step": 1305
    },
    {
      "epoch": 6.58,
      "learning_rate": 2.4376628209899517e-06,
      "loss": 0.0513,
      "step": 1310
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.4469668775586157e-06,
      "loss": 0.0619,
      "step": 1315
    },
    {
      "epoch": 6.63,
      "learning_rate": 2.4562709341272797e-06,
      "loss": 0.104,
      "step": 1320
    },
    {
      "epoch": 6.66,
      "learning_rate": 2.4655749906959437e-06,
      "loss": 0.0743,
      "step": 1325
    },
    {
      "epoch": 6.68,
      "learning_rate": 2.4748790472646077e-06,
      "loss": 0.0758,
      "step": 1330
    },
    {
      "epoch": 6.71,
      "learning_rate": 2.4841831038332713e-06,
      "loss": 0.1288,
      "step": 1335
    },
    {
      "epoch": 6.73,
      "learning_rate": 2.4934871604019353e-06,
      "loss": 0.0781,
      "step": 1340
    },
    {
      "epoch": 6.76,
      "learning_rate": 2.5027912169705993e-06,
      "loss": 0.0546,
      "step": 1345
    },
    {
      "epoch": 6.78,
      "learning_rate": 2.5120952735392633e-06,
      "loss": 0.068,
      "step": 1350
    },
    {
      "epoch": 6.81,
      "learning_rate": 2.521399330107927e-06,
      "loss": 0.0633,
      "step": 1355
    },
    {
      "epoch": 6.83,
      "learning_rate": 2.5307033866765913e-06,
      "loss": 0.0726,
      "step": 1360
    },
    {
      "epoch": 6.86,
      "learning_rate": 2.5400074432452552e-06,
      "loss": 0.0661,
      "step": 1365
    },
    {
      "epoch": 6.88,
      "learning_rate": 2.549311499813919e-06,
      "loss": 0.0248,
      "step": 1370
    },
    {
      "epoch": 6.91,
      "learning_rate": 2.5586155563825832e-06,
      "loss": 0.1089,
      "step": 1375
    },
    {
      "epoch": 6.93,
      "learning_rate": 2.567919612951247e-06,
      "loss": 0.0472,
      "step": 1380
    },
    {
      "epoch": 6.96,
      "learning_rate": 2.577223669519911e-06,
      "loss": 0.0855,
      "step": 1385
    },
    {
      "epoch": 6.98,
      "learning_rate": 2.5865277260885748e-06,
      "loss": 0.078,
      "step": 1390
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9802073515551367,
      "eval_loss": 0.05543642118573189,
      "eval_runtime": 85.9916,
      "eval_samples_per_second": 24.677,
      "eval_steps_per_second": 1.547,
      "step": 1393
    },
    {
      "epoch": 7.01,
      "learning_rate": 2.5958317826572388e-06,
      "loss": 0.1323,
      "step": 1395
    },
    {
      "epoch": 7.04,
      "learning_rate": 2.6051358392259028e-06,
      "loss": 0.0732,
      "step": 1400
    },
    {
      "epoch": 7.06,
      "learning_rate": 2.6144398957945668e-06,
      "loss": 0.0742,
      "step": 1405
    },
    {
      "epoch": 7.09,
      "learning_rate": 2.6237439523632308e-06,
      "loss": 0.0807,
      "step": 1410
    },
    {
      "epoch": 7.11,
      "learning_rate": 2.6330480089318943e-06,
      "loss": 0.099,
      "step": 1415
    },
    {
      "epoch": 7.14,
      "learning_rate": 2.6423520655005587e-06,
      "loss": 0.0585,
      "step": 1420
    },
    {
      "epoch": 7.16,
      "learning_rate": 2.6516561220692223e-06,
      "loss": 0.0403,
      "step": 1425
    },
    {
      "epoch": 7.19,
      "learning_rate": 2.6609601786378863e-06,
      "loss": 0.0248,
      "step": 1430
    },
    {
      "epoch": 7.21,
      "learning_rate": 2.6702642352065507e-06,
      "loss": 0.0718,
      "step": 1435
    },
    {
      "epoch": 7.24,
      "learning_rate": 2.6795682917752143e-06,
      "loss": 0.0681,
      "step": 1440
    },
    {
      "epoch": 7.26,
      "learning_rate": 2.6888723483438783e-06,
      "loss": 0.0726,
      "step": 1445
    },
    {
      "epoch": 7.29,
      "learning_rate": 2.698176404912542e-06,
      "loss": 0.1098,
      "step": 1450
    },
    {
      "epoch": 7.31,
      "learning_rate": 2.7074804614812063e-06,
      "loss": 0.045,
      "step": 1455
    },
    {
      "epoch": 7.34,
      "learning_rate": 2.71678451804987e-06,
      "loss": 0.0613,
      "step": 1460
    },
    {
      "epoch": 7.36,
      "learning_rate": 2.726088574618534e-06,
      "loss": 0.0425,
      "step": 1465
    },
    {
      "epoch": 7.39,
      "learning_rate": 2.7353926311871983e-06,
      "loss": 0.0861,
      "step": 1470
    },
    {
      "epoch": 7.41,
      "learning_rate": 2.744696687755862e-06,
      "loss": 0.0627,
      "step": 1475
    },
    {
      "epoch": 7.44,
      "learning_rate": 2.754000744324526e-06,
      "loss": 0.1238,
      "step": 1480
    },
    {
      "epoch": 7.46,
      "learning_rate": 2.7633048008931894e-06,
      "loss": 0.0526,
      "step": 1485
    },
    {
      "epoch": 7.49,
      "learning_rate": 2.772608857461854e-06,
      "loss": 0.0445,
      "step": 1490
    },
    {
      "epoch": 7.51,
      "learning_rate": 2.7819129140305174e-06,
      "loss": 0.0784,
      "step": 1495
    },
    {
      "epoch": 7.54,
      "learning_rate": 2.7912169705991814e-06,
      "loss": 0.0499,
      "step": 1500
    },
    {
      "epoch": 7.56,
      "learning_rate": 2.8005210271678458e-06,
      "loss": 0.0332,
      "step": 1505
    },
    {
      "epoch": 7.59,
      "learning_rate": 2.8098250837365093e-06,
      "loss": 0.0724,
      "step": 1510
    },
    {
      "epoch": 7.61,
      "learning_rate": 2.8191291403051733e-06,
      "loss": 0.0389,
      "step": 1515
    },
    {
      "epoch": 7.64,
      "learning_rate": 2.828433196873837e-06,
      "loss": 0.0364,
      "step": 1520
    },
    {
      "epoch": 7.66,
      "learning_rate": 2.8377372534425013e-06,
      "loss": 0.0757,
      "step": 1525
    },
    {
      "epoch": 7.69,
      "learning_rate": 2.847041310011165e-06,
      "loss": 0.0601,
      "step": 1530
    },
    {
      "epoch": 7.71,
      "learning_rate": 2.856345366579829e-06,
      "loss": 0.0332,
      "step": 1535
    },
    {
      "epoch": 7.74,
      "learning_rate": 2.865649423148493e-06,
      "loss": 0.0605,
      "step": 1540
    },
    {
      "epoch": 7.76,
      "learning_rate": 2.874953479717157e-06,
      "loss": 0.0401,
      "step": 1545
    },
    {
      "epoch": 7.79,
      "learning_rate": 2.884257536285821e-06,
      "loss": 0.0583,
      "step": 1550
    },
    {
      "epoch": 7.81,
      "learning_rate": 2.893561592854485e-06,
      "loss": 0.0631,
      "step": 1555
    },
    {
      "epoch": 7.84,
      "learning_rate": 2.902865649423149e-06,
      "loss": 0.0335,
      "step": 1560
    },
    {
      "epoch": 7.86,
      "learning_rate": 2.9121697059918124e-06,
      "loss": 0.063,
      "step": 1565
    },
    {
      "epoch": 7.89,
      "learning_rate": 2.921473762560477e-06,
      "loss": 0.0186,
      "step": 1570
    },
    {
      "epoch": 7.91,
      "learning_rate": 2.9307778191291404e-06,
      "loss": 0.0476,
      "step": 1575
    },
    {
      "epoch": 7.94,
      "learning_rate": 2.9400818756978044e-06,
      "loss": 0.0782,
      "step": 1580
    },
    {
      "epoch": 7.96,
      "learning_rate": 2.949385932266469e-06,
      "loss": 0.0821,
      "step": 1585
    },
    {
      "epoch": 7.99,
      "learning_rate": 2.9586899888351324e-06,
      "loss": 0.1769,
      "step": 1590
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9764373232799246,
      "eval_loss": 0.05559385195374489,
      "eval_runtime": 86.1579,
      "eval_samples_per_second": 24.629,
      "eval_steps_per_second": 1.544,
      "step": 1592
    },
    {
      "epoch": 8.02,
      "learning_rate": 2.9679940454037964e-06,
      "loss": 0.1061,
      "step": 1595
    },
    {
      "epoch": 8.04,
      "learning_rate": 2.97729810197246e-06,
      "loss": 0.075,
      "step": 1600
    },
    {
      "epoch": 8.07,
      "learning_rate": 2.9866021585411244e-06,
      "loss": 0.0448,
      "step": 1605
    },
    {
      "epoch": 8.09,
      "learning_rate": 2.995906215109788e-06,
      "loss": 0.1074,
      "step": 1610
    },
    {
      "epoch": 8.12,
      "learning_rate": 3.005210271678452e-06,
      "loss": 0.0942,
      "step": 1615
    },
    {
      "epoch": 8.14,
      "learning_rate": 3.0145143282471163e-06,
      "loss": 0.0713,
      "step": 1620
    },
    {
      "epoch": 8.17,
      "learning_rate": 3.02381838481578e-06,
      "loss": 0.0686,
      "step": 1625
    },
    {
      "epoch": 8.19,
      "learning_rate": 3.033122441384444e-06,
      "loss": 0.0838,
      "step": 1630
    },
    {
      "epoch": 8.22,
      "learning_rate": 3.0424264979531075e-06,
      "loss": 0.0298,
      "step": 1635
    },
    {
      "epoch": 8.24,
      "learning_rate": 3.051730554521772e-06,
      "loss": 0.0271,
      "step": 1640
    },
    {
      "epoch": 8.27,
      "learning_rate": 3.0610346110904355e-06,
      "loss": 0.036,
      "step": 1645
    },
    {
      "epoch": 8.29,
      "learning_rate": 3.0703386676590995e-06,
      "loss": 0.032,
      "step": 1650
    },
    {
      "epoch": 8.32,
      "learning_rate": 3.079642724227764e-06,
      "loss": 0.0333,
      "step": 1655
    },
    {
      "epoch": 8.34,
      "learning_rate": 3.0889467807964274e-06,
      "loss": 0.064,
      "step": 1660
    },
    {
      "epoch": 8.37,
      "learning_rate": 3.0982508373650914e-06,
      "loss": 0.0551,
      "step": 1665
    },
    {
      "epoch": 8.39,
      "learning_rate": 3.107554893933755e-06,
      "loss": 0.0299,
      "step": 1670
    },
    {
      "epoch": 8.42,
      "learning_rate": 3.1168589505024194e-06,
      "loss": 0.046,
      "step": 1675
    },
    {
      "epoch": 8.44,
      "learning_rate": 3.126163007071083e-06,
      "loss": 0.0857,
      "step": 1680
    },
    {
      "epoch": 8.47,
      "learning_rate": 3.135467063639747e-06,
      "loss": 0.0668,
      "step": 1685
    },
    {
      "epoch": 8.49,
      "learning_rate": 3.1447711202084114e-06,
      "loss": 0.0187,
      "step": 1690
    },
    {
      "epoch": 8.52,
      "learning_rate": 3.154075176777075e-06,
      "loss": 0.0679,
      "step": 1695
    },
    {
      "epoch": 8.54,
      "learning_rate": 3.163379233345739e-06,
      "loss": 0.0122,
      "step": 1700
    },
    {
      "epoch": 8.57,
      "learning_rate": 3.172683289914403e-06,
      "loss": 0.0591,
      "step": 1705
    },
    {
      "epoch": 8.59,
      "learning_rate": 3.181987346483067e-06,
      "loss": 0.0681,
      "step": 1710
    },
    {
      "epoch": 8.62,
      "learning_rate": 3.1912914030517305e-06,
      "loss": 0.0199,
      "step": 1715
    },
    {
      "epoch": 8.64,
      "learning_rate": 3.200595459620395e-06,
      "loss": 0.0138,
      "step": 1720
    },
    {
      "epoch": 8.67,
      "learning_rate": 3.209899516189059e-06,
      "loss": 0.0181,
      "step": 1725
    },
    {
      "epoch": 8.69,
      "learning_rate": 3.2192035727577225e-06,
      "loss": 0.0347,
      "step": 1730
    },
    {
      "epoch": 8.72,
      "learning_rate": 3.228507629326387e-06,
      "loss": 0.0539,
      "step": 1735
    },
    {
      "epoch": 8.74,
      "learning_rate": 3.2378116858950505e-06,
      "loss": 0.0598,
      "step": 1740
    },
    {
      "epoch": 8.77,
      "learning_rate": 3.2471157424637145e-06,
      "loss": 0.0401,
      "step": 1745
    },
    {
      "epoch": 8.79,
      "learning_rate": 3.256419799032378e-06,
      "loss": 0.0286,
      "step": 1750
    },
    {
      "epoch": 8.82,
      "learning_rate": 3.2657238556010425e-06,
      "loss": 0.0956,
      "step": 1755
    },
    {
      "epoch": 8.84,
      "learning_rate": 3.2750279121697065e-06,
      "loss": 0.0232,
      "step": 1760
    },
    {
      "epoch": 8.87,
      "learning_rate": 3.28433196873837e-06,
      "loss": 0.0177,
      "step": 1765
    },
    {
      "epoch": 8.89,
      "learning_rate": 3.2936360253070344e-06,
      "loss": 0.063,
      "step": 1770
    },
    {
      "epoch": 8.92,
      "learning_rate": 3.302940081875698e-06,
      "loss": 0.0304,
      "step": 1775
    },
    {
      "epoch": 8.94,
      "learning_rate": 3.312244138444362e-06,
      "loss": 0.0377,
      "step": 1780
    },
    {
      "epoch": 8.97,
      "learning_rate": 3.3215481950130256e-06,
      "loss": 0.023,
      "step": 1785
    },
    {
      "epoch": 8.99,
      "learning_rate": 3.33085225158169e-06,
      "loss": 0.0568,
      "step": 1790
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9806786050895382,
      "eval_loss": 0.0568881630897522,
      "eval_runtime": 86.3133,
      "eval_samples_per_second": 24.585,
      "eval_steps_per_second": 1.541,
      "step": 1791
    },
    {
      "epoch": 9.02,
      "learning_rate": 3.340156308150354e-06,
      "loss": 0.0474,
      "step": 1795
    },
    {
      "epoch": 9.05,
      "learning_rate": 3.3494603647190175e-06,
      "loss": 0.0779,
      "step": 1800
    },
    {
      "epoch": 9.07,
      "learning_rate": 3.358764421287682e-06,
      "loss": 0.0458,
      "step": 1805
    },
    {
      "epoch": 9.1,
      "learning_rate": 3.3680684778563455e-06,
      "loss": 0.0285,
      "step": 1810
    },
    {
      "epoch": 9.12,
      "learning_rate": 3.3773725344250095e-06,
      "loss": 0.0147,
      "step": 1815
    },
    {
      "epoch": 9.15,
      "learning_rate": 3.386676590993673e-06,
      "loss": 0.0296,
      "step": 1820
    },
    {
      "epoch": 9.17,
      "learning_rate": 3.3959806475623375e-06,
      "loss": 0.0282,
      "step": 1825
    },
    {
      "epoch": 9.2,
      "learning_rate": 3.4052847041310015e-06,
      "loss": 0.0574,
      "step": 1830
    },
    {
      "epoch": 9.22,
      "learning_rate": 3.414588760699665e-06,
      "loss": 0.0424,
      "step": 1835
    },
    {
      "epoch": 9.25,
      "learning_rate": 3.4238928172683295e-06,
      "loss": 0.0438,
      "step": 1840
    },
    {
      "epoch": 9.27,
      "learning_rate": 3.433196873836993e-06,
      "loss": 0.024,
      "step": 1845
    },
    {
      "epoch": 9.3,
      "learning_rate": 3.442500930405657e-06,
      "loss": 0.0288,
      "step": 1850
    },
    {
      "epoch": 9.32,
      "learning_rate": 3.451804986974321e-06,
      "loss": 0.1919,
      "step": 1855
    },
    {
      "epoch": 9.35,
      "learning_rate": 3.461109043542985e-06,
      "loss": 0.0577,
      "step": 1860
    },
    {
      "epoch": 9.37,
      "learning_rate": 3.470413100111649e-06,
      "loss": 0.0688,
      "step": 1865
    },
    {
      "epoch": 9.4,
      "learning_rate": 3.479717156680313e-06,
      "loss": 0.038,
      "step": 1870
    },
    {
      "epoch": 9.42,
      "learning_rate": 3.489021213248977e-06,
      "loss": 0.0567,
      "step": 1875
    },
    {
      "epoch": 9.45,
      "learning_rate": 3.4983252698176406e-06,
      "loss": 0.0649,
      "step": 1880
    },
    {
      "epoch": 9.47,
      "learning_rate": 3.507629326386305e-06,
      "loss": 0.0546,
      "step": 1885
    },
    {
      "epoch": 9.5,
      "learning_rate": 3.5169333829549686e-06,
      "loss": 0.0062,
      "step": 1890
    },
    {
      "epoch": 9.52,
      "learning_rate": 3.5262374395236326e-06,
      "loss": 0.0863,
      "step": 1895
    },
    {
      "epoch": 9.55,
      "learning_rate": 3.535541496092297e-06,
      "loss": 0.0451,
      "step": 1900
    },
    {
      "epoch": 9.57,
      "learning_rate": 3.5448455526609606e-06,
      "loss": 0.0627,
      "step": 1905
    },
    {
      "epoch": 9.6,
      "learning_rate": 3.5541496092296245e-06,
      "loss": 0.1044,
      "step": 1910
    },
    {
      "epoch": 9.62,
      "learning_rate": 3.563453665798288e-06,
      "loss": 0.0658,
      "step": 1915
    },
    {
      "epoch": 9.65,
      "learning_rate": 3.5727577223669525e-06,
      "loss": 0.0683,
      "step": 1920
    },
    {
      "epoch": 9.67,
      "learning_rate": 3.582061778935616e-06,
      "loss": 0.025,
      "step": 1925
    },
    {
      "epoch": 9.7,
      "learning_rate": 3.59136583550428e-06,
      "loss": 0.0674,
      "step": 1930
    },
    {
      "epoch": 9.72,
      "learning_rate": 3.6006698920729445e-06,
      "loss": 0.066,
      "step": 1935
    },
    {
      "epoch": 9.75,
      "learning_rate": 3.609973948641608e-06,
      "loss": 0.0272,
      "step": 1940
    },
    {
      "epoch": 9.77,
      "learning_rate": 3.619278005210272e-06,
      "loss": 0.089,
      "step": 1945
    },
    {
      "epoch": 9.8,
      "learning_rate": 3.6285820617789356e-06,
      "loss": 0.0341,
      "step": 1950
    },
    {
      "epoch": 9.82,
      "learning_rate": 3.6378861183476e-06,
      "loss": 0.1069,
      "step": 1955
    },
    {
      "epoch": 9.85,
      "learning_rate": 3.6471901749162636e-06,
      "loss": 0.054,
      "step": 1960
    },
    {
      "epoch": 9.87,
      "learning_rate": 3.6564942314849276e-06,
      "loss": 0.1143,
      "step": 1965
    },
    {
      "epoch": 9.9,
      "learning_rate": 3.665798288053591e-06,
      "loss": 0.041,
      "step": 1970
    },
    {
      "epoch": 9.92,
      "learning_rate": 3.6751023446222556e-06,
      "loss": 0.0462,
      "step": 1975
    },
    {
      "epoch": 9.95,
      "learning_rate": 3.6844064011909196e-06,
      "loss": 0.0251,
      "step": 1980
    },
    {
      "epoch": 9.97,
      "learning_rate": 3.693710457759583e-06,
      "loss": 0.0185,
      "step": 1985
    },
    {
      "epoch": 10.0,
      "learning_rate": 3.7030145143282476e-06,
      "loss": 0.0728,
      "step": 1990
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9915174363807728,
      "eval_loss": 0.023358536884188652,
      "eval_runtime": 86.3411,
      "eval_samples_per_second": 24.577,
      "eval_steps_per_second": 1.54,
      "step": 1990
    },
    {
      "epoch": 10.03,
      "learning_rate": 3.712318570896911e-06,
      "loss": 0.0272,
      "step": 1995
    },
    {
      "epoch": 10.05,
      "learning_rate": 3.721622627465575e-06,
      "loss": 0.0349,
      "step": 2000
    },
    {
      "epoch": 10.08,
      "learning_rate": 3.730926684034239e-06,
      "loss": 0.0286,
      "step": 2005
    },
    {
      "epoch": 10.1,
      "learning_rate": 3.740230740602903e-06,
      "loss": 0.0442,
      "step": 2010
    },
    {
      "epoch": 10.13,
      "learning_rate": 3.749534797171567e-06,
      "loss": 0.0424,
      "step": 2015
    },
    {
      "epoch": 10.15,
      "learning_rate": 3.758838853740231e-06,
      "loss": 0.0509,
      "step": 2020
    },
    {
      "epoch": 10.18,
      "learning_rate": 3.768142910308895e-06,
      "loss": 0.0246,
      "step": 2025
    },
    {
      "epoch": 10.2,
      "learning_rate": 3.7774469668775587e-06,
      "loss": 0.1116,
      "step": 2030
    },
    {
      "epoch": 10.23,
      "learning_rate": 3.786751023446223e-06,
      "loss": 0.0309,
      "step": 2035
    },
    {
      "epoch": 10.25,
      "learning_rate": 3.7960550800148867e-06,
      "loss": 0.092,
      "step": 2040
    },
    {
      "epoch": 10.28,
      "learning_rate": 3.8053591365835507e-06,
      "loss": 0.106,
      "step": 2045
    },
    {
      "epoch": 10.3,
      "learning_rate": 3.814663193152215e-06,
      "loss": 0.0147,
      "step": 2050
    },
    {
      "epoch": 10.33,
      "learning_rate": 3.823967249720879e-06,
      "loss": 0.0562,
      "step": 2055
    },
    {
      "epoch": 10.35,
      "learning_rate": 3.833271306289543e-06,
      "loss": 0.0882,
      "step": 2060
    },
    {
      "epoch": 10.38,
      "learning_rate": 3.842575362858207e-06,
      "loss": 0.0367,
      "step": 2065
    },
    {
      "epoch": 10.4,
      "learning_rate": 3.85187941942687e-06,
      "loss": 0.0086,
      "step": 2070
    },
    {
      "epoch": 10.43,
      "learning_rate": 3.861183475995534e-06,
      "loss": 0.0887,
      "step": 2075
    },
    {
      "epoch": 10.45,
      "learning_rate": 3.870487532564198e-06,
      "loss": 0.0313,
      "step": 2080
    },
    {
      "epoch": 10.48,
      "learning_rate": 3.879791589132863e-06,
      "loss": 0.0233,
      "step": 2085
    },
    {
      "epoch": 10.5,
      "learning_rate": 3.889095645701526e-06,
      "loss": 0.0827,
      "step": 2090
    },
    {
      "epoch": 10.53,
      "learning_rate": 3.898399702270191e-06,
      "loss": 0.0678,
      "step": 2095
    },
    {
      "epoch": 10.55,
      "learning_rate": 3.907703758838854e-06,
      "loss": 0.0669,
      "step": 2100
    },
    {
      "epoch": 10.58,
      "learning_rate": 3.917007815407518e-06,
      "loss": 0.1371,
      "step": 2105
    },
    {
      "epoch": 10.6,
      "learning_rate": 3.926311871976181e-06,
      "loss": 0.0447,
      "step": 2110
    },
    {
      "epoch": 10.63,
      "learning_rate": 3.935615928544846e-06,
      "loss": 0.0131,
      "step": 2115
    },
    {
      "epoch": 10.65,
      "learning_rate": 3.94491998511351e-06,
      "loss": 0.0519,
      "step": 2120
    },
    {
      "epoch": 10.68,
      "learning_rate": 3.954224041682174e-06,
      "loss": 0.0282,
      "step": 2125
    },
    {
      "epoch": 10.7,
      "learning_rate": 3.963528098250838e-06,
      "loss": 0.028,
      "step": 2130
    },
    {
      "epoch": 10.73,
      "learning_rate": 3.972832154819502e-06,
      "loss": 0.0219,
      "step": 2135
    },
    {
      "epoch": 10.75,
      "learning_rate": 3.982136211388165e-06,
      "loss": 0.0348,
      "step": 2140
    },
    {
      "epoch": 10.78,
      "learning_rate": 3.99144026795683e-06,
      "loss": 0.0388,
      "step": 2145
    },
    {
      "epoch": 10.8,
      "learning_rate": 4.000744324525493e-06,
      "loss": 0.0292,
      "step": 2150
    },
    {
      "epoch": 10.83,
      "learning_rate": 4.010048381094158e-06,
      "loss": 0.0478,
      "step": 2155
    },
    {
      "epoch": 10.85,
      "learning_rate": 4.019352437662821e-06,
      "loss": 0.1067,
      "step": 2160
    },
    {
      "epoch": 10.88,
      "learning_rate": 4.028656494231486e-06,
      "loss": 0.015,
      "step": 2165
    },
    {
      "epoch": 10.9,
      "learning_rate": 4.037960550800149e-06,
      "loss": 0.0167,
      "step": 2170
    },
    {
      "epoch": 10.93,
      "learning_rate": 4.047264607368814e-06,
      "loss": 0.0488,
      "step": 2175
    },
    {
      "epoch": 10.95,
      "learning_rate": 4.056568663937477e-06,
      "loss": 0.0104,
      "step": 2180
    },
    {
      "epoch": 10.98,
      "learning_rate": 4.065872720506141e-06,
      "loss": 0.0229,
      "step": 2185
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.9910461828463714,
      "eval_loss": 0.02395106852054596,
      "eval_runtime": 86.1738,
      "eval_samples_per_second": 24.625,
      "eval_steps_per_second": 1.543,
      "step": 2189
    },
    {
      "epoch": 11.01,
      "learning_rate": 4.075176777074805e-06,
      "loss": 0.015,
      "step": 2190
    },
    {
      "epoch": 11.03,
      "learning_rate": 4.084480833643469e-06,
      "loss": 0.0155,
      "step": 2195
    },
    {
      "epoch": 11.06,
      "learning_rate": 4.093784890212133e-06,
      "loss": 0.01,
      "step": 2200
    },
    {
      "epoch": 11.08,
      "learning_rate": 4.103088946780797e-06,
      "loss": 0.0297,
      "step": 2205
    },
    {
      "epoch": 11.11,
      "learning_rate": 4.112393003349461e-06,
      "loss": 0.0736,
      "step": 2210
    },
    {
      "epoch": 11.13,
      "learning_rate": 4.121697059918125e-06,
      "loss": 0.0356,
      "step": 2215
    },
    {
      "epoch": 11.16,
      "learning_rate": 4.131001116486788e-06,
      "loss": 0.0379,
      "step": 2220
    },
    {
      "epoch": 11.18,
      "learning_rate": 4.140305173055453e-06,
      "loss": 0.0781,
      "step": 2225
    },
    {
      "epoch": 11.21,
      "learning_rate": 4.149609229624116e-06,
      "loss": 0.0089,
      "step": 2230
    },
    {
      "epoch": 11.23,
      "learning_rate": 4.158913286192781e-06,
      "loss": 0.0279,
      "step": 2235
    },
    {
      "epoch": 11.26,
      "learning_rate": 4.168217342761444e-06,
      "loss": 0.015,
      "step": 2240
    },
    {
      "epoch": 11.28,
      "learning_rate": 4.177521399330109e-06,
      "loss": 0.027,
      "step": 2245
    },
    {
      "epoch": 11.31,
      "learning_rate": 4.186825455898772e-06,
      "loss": 0.0151,
      "step": 2250
    },
    {
      "epoch": 11.33,
      "learning_rate": 4.196129512467436e-06,
      "loss": 0.0567,
      "step": 2255
    },
    {
      "epoch": 11.36,
      "learning_rate": 4.2054335690361e-06,
      "loss": 0.0407,
      "step": 2260
    },
    {
      "epoch": 11.38,
      "learning_rate": 4.214737625604764e-06,
      "loss": 0.0416,
      "step": 2265
    },
    {
      "epoch": 11.41,
      "learning_rate": 4.224041682173428e-06,
      "loss": 0.0762,
      "step": 2270
    },
    {
      "epoch": 11.43,
      "learning_rate": 4.233345738742092e-06,
      "loss": 0.0331,
      "step": 2275
    },
    {
      "epoch": 11.46,
      "learning_rate": 4.242649795310756e-06,
      "loss": 0.0082,
      "step": 2280
    },
    {
      "epoch": 11.48,
      "learning_rate": 4.25195385187942e-06,
      "loss": 0.0784,
      "step": 2285
    },
    {
      "epoch": 11.51,
      "learning_rate": 4.261257908448083e-06,
      "loss": 0.0373,
      "step": 2290
    },
    {
      "epoch": 11.53,
      "learning_rate": 4.270561965016748e-06,
      "loss": 0.0136,
      "step": 2295
    },
    {
      "epoch": 11.56,
      "learning_rate": 4.279866021585411e-06,
      "loss": 0.005,
      "step": 2300
    },
    {
      "epoch": 11.58,
      "learning_rate": 4.289170078154076e-06,
      "loss": 0.0523,
      "step": 2305
    },
    {
      "epoch": 11.61,
      "learning_rate": 4.298474134722739e-06,
      "loss": 0.0233,
      "step": 2310
    },
    {
      "epoch": 11.63,
      "learning_rate": 4.307778191291404e-06,
      "loss": 0.056,
      "step": 2315
    },
    {
      "epoch": 11.66,
      "learning_rate": 4.317082247860067e-06,
      "loss": 0.0013,
      "step": 2320
    },
    {
      "epoch": 11.68,
      "learning_rate": 4.326386304428732e-06,
      "loss": 0.031,
      "step": 2325
    },
    {
      "epoch": 11.71,
      "learning_rate": 4.335690360997395e-06,
      "loss": 0.033,
      "step": 2330
    },
    {
      "epoch": 11.73,
      "learning_rate": 4.344994417566059e-06,
      "loss": 0.1066,
      "step": 2335
    },
    {
      "epoch": 11.76,
      "learning_rate": 4.354298474134723e-06,
      "loss": 0.0108,
      "step": 2340
    },
    {
      "epoch": 11.78,
      "learning_rate": 4.363602530703387e-06,
      "loss": 0.0127,
      "step": 2345
    },
    {
      "epoch": 11.81,
      "learning_rate": 4.372906587272051e-06,
      "loss": 0.0044,
      "step": 2350
    },
    {
      "epoch": 11.83,
      "learning_rate": 4.382210643840715e-06,
      "loss": 0.0484,
      "step": 2355
    },
    {
      "epoch": 11.86,
      "learning_rate": 4.391514700409379e-06,
      "loss": 0.069,
      "step": 2360
    },
    {
      "epoch": 11.88,
      "learning_rate": 4.400818756978043e-06,
      "loss": 0.0609,
      "step": 2365
    },
    {
      "epoch": 11.91,
      "learning_rate": 4.410122813546706e-06,
      "loss": 0.0397,
      "step": 2370
    },
    {
      "epoch": 11.93,
      "learning_rate": 4.419426870115371e-06,
      "loss": 0.0869,
      "step": 2375
    },
    {
      "epoch": 11.96,
      "learning_rate": 4.428730926684034e-06,
      "loss": 0.1044,
      "step": 2380
    },
    {
      "epoch": 11.98,
      "learning_rate": 4.438034983252699e-06,
      "loss": 0.0561,
      "step": 2385
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9901036757775683,
      "eval_loss": 0.03518503159284592,
      "eval_runtime": 86.2049,
      "eval_samples_per_second": 24.616,
      "eval_steps_per_second": 1.543,
      "step": 2388
    },
    {
      "epoch": 12.01,
      "learning_rate": 4.447339039821362e-06,
      "loss": 0.0209,
      "step": 2390
    },
    {
      "epoch": 12.04,
      "learning_rate": 4.456643096390027e-06,
      "loss": 0.0749,
      "step": 2395
    },
    {
      "epoch": 12.06,
      "learning_rate": 4.46594715295869e-06,
      "loss": 0.048,
      "step": 2400
    },
    {
      "epoch": 12.09,
      "learning_rate": 4.475251209527354e-06,
      "loss": 0.0166,
      "step": 2405
    },
    {
      "epoch": 12.11,
      "learning_rate": 4.484555266096018e-06,
      "loss": 0.022,
      "step": 2410
    },
    {
      "epoch": 12.14,
      "learning_rate": 4.493859322664682e-06,
      "loss": 0.0056,
      "step": 2415
    },
    {
      "epoch": 12.16,
      "learning_rate": 4.503163379233346e-06,
      "loss": 0.0242,
      "step": 2420
    },
    {
      "epoch": 12.19,
      "learning_rate": 4.51246743580201e-06,
      "loss": 0.0028,
      "step": 2425
    },
    {
      "epoch": 12.21,
      "learning_rate": 4.521771492370674e-06,
      "loss": 0.0664,
      "step": 2430
    },
    {
      "epoch": 12.24,
      "learning_rate": 4.531075548939338e-06,
      "loss": 0.0128,
      "step": 2435
    },
    {
      "epoch": 12.26,
      "learning_rate": 4.5403796055080014e-06,
      "loss": 0.0426,
      "step": 2440
    },
    {
      "epoch": 12.29,
      "learning_rate": 4.549683662076666e-06,
      "loss": 0.0398,
      "step": 2445
    },
    {
      "epoch": 12.31,
      "learning_rate": 4.5589877186453294e-06,
      "loss": 0.0308,
      "step": 2450
    },
    {
      "epoch": 12.34,
      "learning_rate": 4.568291775213994e-06,
      "loss": 0.0321,
      "step": 2455
    },
    {
      "epoch": 12.36,
      "learning_rate": 4.577595831782657e-06,
      "loss": 0.0018,
      "step": 2460
    },
    {
      "epoch": 12.39,
      "learning_rate": 4.586899888351322e-06,
      "loss": 0.006,
      "step": 2465
    },
    {
      "epoch": 12.41,
      "learning_rate": 4.596203944919985e-06,
      "loss": 0.0453,
      "step": 2470
    },
    {
      "epoch": 12.44,
      "learning_rate": 4.60550800148865e-06,
      "loss": 0.01,
      "step": 2475
    },
    {
      "epoch": 12.46,
      "learning_rate": 4.614812058057313e-06,
      "loss": 0.0006,
      "step": 2480
    },
    {
      "epoch": 12.49,
      "learning_rate": 4.624116114625977e-06,
      "loss": 0.0134,
      "step": 2485
    },
    {
      "epoch": 12.51,
      "learning_rate": 4.633420171194641e-06,
      "loss": 0.0589,
      "step": 2490
    },
    {
      "epoch": 12.54,
      "learning_rate": 4.642724227763305e-06,
      "loss": 0.0715,
      "step": 2495
    },
    {
      "epoch": 12.56,
      "learning_rate": 4.652028284331969e-06,
      "loss": 0.1123,
      "step": 2500
    },
    {
      "epoch": 12.59,
      "learning_rate": 4.661332340900633e-06,
      "loss": 0.0317,
      "step": 2505
    },
    {
      "epoch": 12.61,
      "learning_rate": 4.670636397469297e-06,
      "loss": 0.1303,
      "step": 2510
    },
    {
      "epoch": 12.64,
      "learning_rate": 4.679940454037961e-06,
      "loss": 0.0684,
      "step": 2515
    },
    {
      "epoch": 12.66,
      "learning_rate": 4.6892445106066245e-06,
      "loss": 0.0028,
      "step": 2520
    },
    {
      "epoch": 12.69,
      "learning_rate": 4.698548567175289e-06,
      "loss": 0.0125,
      "step": 2525
    },
    {
      "epoch": 12.71,
      "learning_rate": 4.7078526237439525e-06,
      "loss": 0.0309,
      "step": 2530
    },
    {
      "epoch": 12.74,
      "learning_rate": 4.717156680312617e-06,
      "loss": 0.0301,
      "step": 2535
    },
    {
      "epoch": 12.76,
      "learning_rate": 4.7264607368812805e-06,
      "loss": 0.0347,
      "step": 2540
    },
    {
      "epoch": 12.79,
      "learning_rate": 4.735764793449945e-06,
      "loss": 0.0419,
      "step": 2545
    },
    {
      "epoch": 12.81,
      "learning_rate": 4.7450688500186084e-06,
      "loss": 0.0664,
      "step": 2550
    },
    {
      "epoch": 12.84,
      "learning_rate": 4.754372906587272e-06,
      "loss": 0.0961,
      "step": 2555
    },
    {
      "epoch": 12.86,
      "learning_rate": 4.7636769631559364e-06,
      "loss": 0.0475,
      "step": 2560
    },
    {
      "epoch": 12.89,
      "learning_rate": 4.7729810197246e-06,
      "loss": 0.0016,
      "step": 2565
    },
    {
      "epoch": 12.91,
      "learning_rate": 4.782285076293264e-06,
      "loss": 0.0427,
      "step": 2570
    },
    {
      "epoch": 12.94,
      "learning_rate": 4.791589132861928e-06,
      "loss": 0.0263,
      "step": 2575
    },
    {
      "epoch": 12.96,
      "learning_rate": 4.800893189430592e-06,
      "loss": 0.0381,
      "step": 2580
    },
    {
      "epoch": 12.99,
      "learning_rate": 4.810197245999256e-06,
      "loss": 0.014,
      "step": 2585
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.9797360980207351,
      "eval_loss": 0.07485808432102203,
      "eval_runtime": 86.0191,
      "eval_samples_per_second": 24.669,
      "eval_steps_per_second": 1.546,
      "step": 2587
    },
    {
      "epoch": 13.02,
      "learning_rate": 4.8195013025679195e-06,
      "loss": 0.102,
      "step": 2590
    },
    {
      "epoch": 13.04,
      "learning_rate": 4.828805359136584e-06,
      "loss": 0.0077,
      "step": 2595
    },
    {
      "epoch": 13.07,
      "learning_rate": 4.8381094157052475e-06,
      "loss": 0.05,
      "step": 2600
    },
    {
      "epoch": 13.09,
      "learning_rate": 4.847413472273912e-06,
      "loss": 0.0908,
      "step": 2605
    },
    {
      "epoch": 13.12,
      "learning_rate": 4.8567175288425755e-06,
      "loss": 0.0791,
      "step": 2610
    },
    {
      "epoch": 13.14,
      "learning_rate": 4.86602158541124e-06,
      "loss": 0.0168,
      "step": 2615
    },
    {
      "epoch": 13.17,
      "learning_rate": 4.8753256419799035e-06,
      "loss": 0.0008,
      "step": 2620
    },
    {
      "epoch": 13.19,
      "learning_rate": 4.884629698548568e-06,
      "loss": 0.0439,
      "step": 2625
    },
    {
      "epoch": 13.22,
      "learning_rate": 4.8939337551172315e-06,
      "loss": 0.0391,
      "step": 2630
    },
    {
      "epoch": 13.24,
      "learning_rate": 4.903237811685895e-06,
      "loss": 0.0288,
      "step": 2635
    },
    {
      "epoch": 13.27,
      "learning_rate": 4.9125418682545595e-06,
      "loss": 0.0544,
      "step": 2640
    },
    {
      "epoch": 13.29,
      "learning_rate": 4.921845924823223e-06,
      "loss": 0.0194,
      "step": 2645
    },
    {
      "epoch": 13.32,
      "learning_rate": 4.9311499813918875e-06,
      "loss": 0.0714,
      "step": 2650
    },
    {
      "epoch": 13.34,
      "learning_rate": 4.940454037960552e-06,
      "loss": 0.0032,
      "step": 2655
    },
    {
      "epoch": 13.37,
      "learning_rate": 4.9497580945292154e-06,
      "loss": 0.0144,
      "step": 2660
    },
    {
      "epoch": 13.39,
      "learning_rate": 4.959062151097879e-06,
      "loss": 0.0291,
      "step": 2665
    },
    {
      "epoch": 13.42,
      "learning_rate": 4.968366207666543e-06,
      "loss": 0.0368,
      "step": 2670
    },
    {
      "epoch": 13.44,
      "learning_rate": 4.977670264235207e-06,
      "loss": 0.0323,
      "step": 2675
    },
    {
      "epoch": 13.47,
      "learning_rate": 4.9869743208038706e-06,
      "loss": 0.0196,
      "step": 2680
    },
    {
      "epoch": 13.49,
      "learning_rate": 4.996278377372535e-06,
      "loss": 0.0567,
      "step": 2685
    },
    {
      "epoch": 13.52,
      "learning_rate": 4.949664429530202e-06,
      "loss": 0.0885,
      "step": 2690
    },
    {
      "epoch": 13.54,
      "learning_rate": 4.8657718120805375e-06,
      "loss": 0.0077,
      "step": 2695
    },
    {
      "epoch": 13.57,
      "learning_rate": 4.781879194630873e-06,
      "loss": 0.0844,
      "step": 2700
    },
    {
      "epoch": 13.59,
      "learning_rate": 4.697986577181208e-06,
      "loss": 0.0623,
      "step": 2705
    },
    {
      "epoch": 13.62,
      "learning_rate": 4.614093959731544e-06,
      "loss": 0.0506,
      "step": 2710
    },
    {
      "epoch": 13.64,
      "learning_rate": 4.530201342281879e-06,
      "loss": 0.0869,
      "step": 2715
    },
    {
      "epoch": 13.67,
      "learning_rate": 4.446308724832215e-06,
      "loss": 0.0554,
      "step": 2720
    },
    {
      "epoch": 13.69,
      "learning_rate": 4.362416107382551e-06,
      "loss": 0.0215,
      "step": 2725
    },
    {
      "epoch": 13.72,
      "learning_rate": 4.278523489932886e-06,
      "loss": 0.0298,
      "step": 2730
    },
    {
      "epoch": 13.74,
      "learning_rate": 4.194630872483222e-06,
      "loss": 0.0226,
      "step": 2735
    },
    {
      "epoch": 13.77,
      "learning_rate": 4.110738255033558e-06,
      "loss": 0.0192,
      "step": 2740
    },
    {
      "epoch": 13.79,
      "learning_rate": 4.026845637583892e-06,
      "loss": 0.0975,
      "step": 2745
    },
    {
      "epoch": 13.82,
      "learning_rate": 3.942953020134229e-06,
      "loss": 0.0455,
      "step": 2750
    },
    {
      "epoch": 13.84,
      "learning_rate": 3.859060402684564e-06,
      "loss": 0.0314,
      "step": 2755
    },
    {
      "epoch": 13.87,
      "learning_rate": 3.7751677852348994e-06,
      "loss": 0.0163,
      "step": 2760
    },
    {
      "epoch": 13.89,
      "learning_rate": 3.6912751677852355e-06,
      "loss": 0.0283,
      "step": 2765
    },
    {
      "epoch": 13.92,
      "learning_rate": 3.6073825503355707e-06,
      "loss": 0.0044,
      "step": 2770
    },
    {
      "epoch": 13.94,
      "learning_rate": 3.5234899328859064e-06,
      "loss": 0.0078,
      "step": 2775
    },
    {
      "epoch": 13.97,
      "learning_rate": 3.4395973154362416e-06,
      "loss": 0.0201,
      "step": 2780
    },
    {
      "epoch": 13.99,
      "learning_rate": 3.3557046979865777e-06,
      "loss": 0.096,
      "step": 2785
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.9934024505183789,
      "eval_loss": 0.02680634893476963,
      "eval_runtime": 86.0801,
      "eval_samples_per_second": 24.651,
      "eval_steps_per_second": 1.545,
      "step": 2786
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.271812080536913e-06,
      "loss": 0.0893,
      "step": 2790
    },
    {
      "epoch": 14.05,
      "learning_rate": 3.1879194630872486e-06,
      "loss": 0.0148,
      "step": 2795
    },
    {
      "epoch": 14.07,
      "learning_rate": 3.1040268456375843e-06,
      "loss": 0.0098,
      "step": 2800
    },
    {
      "epoch": 14.1,
      "learning_rate": 3.02013422818792e-06,
      "loss": 0.0079,
      "step": 2805
    },
    {
      "epoch": 14.12,
      "learning_rate": 2.936241610738255e-06,
      "loss": 0.0873,
      "step": 2810
    },
    {
      "epoch": 14.15,
      "learning_rate": 2.8523489932885904e-06,
      "loss": 0.0396,
      "step": 2815
    },
    {
      "epoch": 14.17,
      "learning_rate": 2.7684563758389265e-06,
      "loss": 0.0012,
      "step": 2820
    },
    {
      "epoch": 14.2,
      "learning_rate": 2.684563758389262e-06,
      "loss": 0.0012,
      "step": 2825
    },
    {
      "epoch": 14.22,
      "learning_rate": 2.6006711409395974e-06,
      "loss": 0.0507,
      "step": 2830
    },
    {
      "epoch": 14.25,
      "learning_rate": 2.5167785234899335e-06,
      "loss": 0.055,
      "step": 2835
    },
    {
      "epoch": 14.27,
      "learning_rate": 2.4328859060402687e-06,
      "loss": 0.003,
      "step": 2840
    },
    {
      "epoch": 14.3,
      "learning_rate": 2.348993288590604e-06,
      "loss": 0.014,
      "step": 2845
    },
    {
      "epoch": 14.32,
      "learning_rate": 2.2651006711409396e-06,
      "loss": 0.0065,
      "step": 2850
    },
    {
      "epoch": 14.35,
      "learning_rate": 2.1812080536912753e-06,
      "loss": 0.0195,
      "step": 2855
    },
    {
      "epoch": 14.37,
      "learning_rate": 2.097315436241611e-06,
      "loss": 0.0167,
      "step": 2860
    },
    {
      "epoch": 14.4,
      "learning_rate": 2.013422818791946e-06,
      "loss": 0.0397,
      "step": 2865
    },
    {
      "epoch": 14.42,
      "learning_rate": 1.929530201342282e-06,
      "loss": 0.0008,
      "step": 2870
    },
    {
      "epoch": 14.45,
      "learning_rate": 1.8456375838926177e-06,
      "loss": 0.0175,
      "step": 2875
    },
    {
      "epoch": 14.47,
      "learning_rate": 1.7617449664429532e-06,
      "loss": 0.0257,
      "step": 2880
    },
    {
      "epoch": 14.5,
      "learning_rate": 1.6778523489932889e-06,
      "loss": 0.0033,
      "step": 2885
    },
    {
      "epoch": 14.52,
      "learning_rate": 1.5939597315436243e-06,
      "loss": 0.0356,
      "step": 2890
    },
    {
      "epoch": 14.55,
      "learning_rate": 1.51006711409396e-06,
      "loss": 0.0244,
      "step": 2895
    },
    {
      "epoch": 14.57,
      "learning_rate": 1.4261744966442952e-06,
      "loss": 0.0103,
      "step": 2900
    },
    {
      "epoch": 14.6,
      "learning_rate": 1.342281879194631e-06,
      "loss": 0.0255,
      "step": 2905
    },
    {
      "epoch": 14.62,
      "learning_rate": 1.2583892617449667e-06,
      "loss": 0.003,
      "step": 2910
    },
    {
      "epoch": 14.65,
      "learning_rate": 1.174496644295302e-06,
      "loss": 0.0697,
      "step": 2915
    },
    {
      "epoch": 14.67,
      "learning_rate": 1.0906040268456377e-06,
      "loss": 0.0591,
      "step": 2920
    },
    {
      "epoch": 14.7,
      "learning_rate": 1.006711409395973e-06,
      "loss": 0.007,
      "step": 2925
    },
    {
      "epoch": 14.72,
      "learning_rate": 9.228187919463089e-07,
      "loss": 0.0063,
      "step": 2930
    },
    {
      "epoch": 14.75,
      "learning_rate": 8.389261744966444e-07,
      "loss": 0.0301,
      "step": 2935
    },
    {
      "epoch": 14.77,
      "learning_rate": 7.5503355704698e-07,
      "loss": 0.0056,
      "step": 2940
    },
    {
      "epoch": 14.8,
      "learning_rate": 6.711409395973155e-07,
      "loss": 0.0406,
      "step": 2945
    },
    {
      "epoch": 14.82,
      "learning_rate": 5.87248322147651e-07,
      "loss": 0.0192,
      "step": 2950
    },
    {
      "epoch": 14.85,
      "learning_rate": 5.033557046979866e-07,
      "loss": 0.045,
      "step": 2955
    },
    {
      "epoch": 14.87,
      "learning_rate": 4.194630872483222e-07,
      "loss": 0.0156,
      "step": 2960
    },
    {
      "epoch": 14.9,
      "learning_rate": 3.3557046979865777e-07,
      "loss": 0.0183,
      "step": 2965
    },
    {
      "epoch": 14.92,
      "learning_rate": 2.516778523489933e-07,
      "loss": 0.0013,
      "step": 2970
    },
    {
      "epoch": 14.95,
      "learning_rate": 1.6778523489932889e-07,
      "loss": 0.0081,
      "step": 2975
    },
    {
      "epoch": 14.97,
      "learning_rate": 8.389261744966444e-08,
      "loss": 0.0115,
      "step": 2980
    },
    {
      "epoch": 15.0,
      "learning_rate": 0.0,
      "loss": 0.0005,
      "step": 2985
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9957587181903864,
      "eval_loss": 0.014596104621887207,
      "eval_runtime": 85.895,
      "eval_samples_per_second": 24.705,
      "eval_steps_per_second": 1.548,
      "step": 2985
    },
    {
      "epoch": 15.0,
      "step": 2985,
      "total_flos": 2.6189408941190627e+19,
      "train_loss": 0.0015452209050444202,
      "train_runtime": 893.2543,
      "train_samples_per_second": 106.868,
      "train_steps_per_second": 3.342
    }
  ],
  "max_steps": 2985,
  "num_train_epochs": 15,
  "total_flos": 2.6189408941190627e+19,
  "trial_name": null,
  "trial_params": null
}