{
  "best_global_step": 7075,
  "best_metric": 0.957730518736978,
  "best_model_checkpoint": "/kaggle/working/sparse_lora/checkpoint-7075",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 7075,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0353356890459364,
      "grad_norm": 87870.6484375,
      "learning_rate": 4.9653710247349825e-05,
      "loss": 4.5998,
      "step": 50
    },
    {
      "epoch": 0.0706713780918728,
      "grad_norm": 671269.4375,
      "learning_rate": 4.930035335689046e-05,
      "loss": 3.6646,
      "step": 100
    },
    {
      "epoch": 0.10600706713780919,
      "grad_norm": 103706.0,
      "learning_rate": 4.89469964664311e-05,
      "loss": 3.0394,
      "step": 150
    },
    {
      "epoch": 0.1413427561837456,
      "grad_norm": 60311.45703125,
      "learning_rate": 4.859363957597173e-05,
      "loss": 2.1607,
      "step": 200
    },
    {
      "epoch": 0.17667844522968199,
      "grad_norm": 56506.35546875,
      "learning_rate": 4.8240282685512374e-05,
      "loss": 2.0396,
      "step": 250
    },
    {
      "epoch": 0.21201413427561838,
      "grad_norm": 64563.61328125,
      "learning_rate": 4.788692579505301e-05,
      "loss": 1.8784,
      "step": 300
    },
    {
      "epoch": 0.24734982332155478,
      "grad_norm": 84701.9765625,
      "learning_rate": 4.753356890459364e-05,
      "loss": 1.823,
      "step": 350
    },
    {
      "epoch": 0.2826855123674912,
      "grad_norm": 80899.9765625,
      "learning_rate": 4.718021201413428e-05,
      "loss": 1.825,
      "step": 400
    },
    {
      "epoch": 0.31802120141342755,
      "grad_norm": 83637.8671875,
      "learning_rate": 4.682685512367491e-05,
      "loss": 1.6937,
      "step": 450
    },
    {
      "epoch": 0.35335689045936397,
      "grad_norm": 172655.1875,
      "learning_rate": 4.647349823321555e-05,
      "loss": 1.6309,
      "step": 500
    },
    {
      "epoch": 0.38869257950530034,
      "grad_norm": 212648.234375,
      "learning_rate": 4.612014134275618e-05,
      "loss": 1.5667,
      "step": 550
    },
    {
      "epoch": 0.42402826855123676,
      "grad_norm": 85414.8125,
      "learning_rate": 4.5766784452296823e-05,
      "loss": 1.5433,
      "step": 600
    },
    {
      "epoch": 0.45936395759717313,
      "grad_norm": 84665.4296875,
      "learning_rate": 4.541342756183746e-05,
      "loss": 1.5225,
      "step": 650
    },
    {
      "epoch": 0.49469964664310956,
      "grad_norm": 87850.546875,
      "learning_rate": 4.50600706713781e-05,
      "loss": 1.4696,
      "step": 700
    },
    {
      "epoch": 0.5300353356890459,
      "grad_norm": 105920.03125,
      "learning_rate": 4.470671378091873e-05,
      "loss": 1.5046,
      "step": 750
    },
    {
      "epoch": 0.5653710247349824,
      "grad_norm": 128489.2734375,
      "learning_rate": 4.435335689045937e-05,
      "loss": 1.4268,
      "step": 800
    },
    {
      "epoch": 0.6007067137809188,
      "grad_norm": 126656.5390625,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 1.3749,
      "step": 850
    },
    {
      "epoch": 0.6360424028268551,
      "grad_norm": 90037.234375,
      "learning_rate": 4.364664310954063e-05,
      "loss": 1.4043,
      "step": 900
    },
    {
      "epoch": 0.6713780918727915,
      "grad_norm": 123195.953125,
      "learning_rate": 4.329328621908127e-05,
      "loss": 1.3655,
      "step": 950
    },
    {
      "epoch": 0.7067137809187279,
      "grad_norm": 106699.8984375,
      "learning_rate": 4.293992932862191e-05,
      "loss": 1.3125,
      "step": 1000
    },
    {
      "epoch": 0.7420494699646644,
      "grad_norm": 141920.890625,
      "learning_rate": 4.258657243816255e-05,
      "loss": 1.2784,
      "step": 1050
    },
    {
      "epoch": 0.7773851590106007,
      "grad_norm": 155978.375,
      "learning_rate": 4.223321554770318e-05,
      "loss": 1.307,
      "step": 1100
    },
    {
      "epoch": 0.8127208480565371,
      "grad_norm": 151003.53125,
      "learning_rate": 4.187985865724382e-05,
      "loss": 1.2263,
      "step": 1150
    },
    {
      "epoch": 0.8480565371024735,
      "grad_norm": 122458.796875,
      "learning_rate": 4.1526501766784455e-05,
      "loss": 1.1795,
      "step": 1200
    },
    {
      "epoch": 0.8833922261484098,
      "grad_norm": 137286.90625,
      "learning_rate": 4.1173144876325096e-05,
      "loss": 1.2285,
      "step": 1250
    },
    {
      "epoch": 0.9187279151943463,
      "grad_norm": 112117.640625,
      "learning_rate": 4.081978798586573e-05,
      "loss": 1.2263,
      "step": 1300
    },
    {
      "epoch": 0.9540636042402827,
      "grad_norm": 127897.46875,
      "learning_rate": 4.0466431095406356e-05,
      "loss": 1.1636,
      "step": 1350
    },
    {
      "epoch": 0.9893992932862191,
      "grad_norm": 121189.921875,
      "learning_rate": 4.0113074204947e-05,
      "loss": 1.1762,
      "step": 1400
    },
    {
      "epoch": 1.0,
      "eval_bertscore_f1": 0.9435882283983255,
      "eval_bleu": 0.34081729524428495,
      "eval_loss": 0.8283334970474243,
      "eval_meteor": 0.5505336031382576,
      "eval_rouge1": 0.741807000391657,
      "eval_rouge2": 0.561792451504467,
      "eval_runtime": 458.3629,
      "eval_samples_per_second": 14.102,
      "eval_steps_per_second": 0.881,
      "step": 1415
    },
    {
      "epoch": 1.0247349823321554,
      "grad_norm": 128025.5703125,
      "learning_rate": 3.975971731448763e-05,
      "loss": 1.1689,
      "step": 1450
    },
    {
      "epoch": 1.0600706713780919,
      "grad_norm": 117022.6953125,
      "learning_rate": 3.940636042402827e-05,
      "loss": 1.1721,
      "step": 1500
    },
    {
      "epoch": 1.0954063604240283,
      "grad_norm": 127291.71875,
      "learning_rate": 3.9053003533568905e-05,
      "loss": 1.1144,
      "step": 1550
    },
    {
      "epoch": 1.1307420494699647,
      "grad_norm": 432520.5,
      "learning_rate": 3.8699646643109545e-05,
      "loss": 1.1173,
      "step": 1600
    },
    {
      "epoch": 1.1660777385159011,
      "grad_norm": 133514.078125,
      "learning_rate": 3.834628975265018e-05,
      "loss": 1.1349,
      "step": 1650
    },
    {
      "epoch": 1.2014134275618376,
      "grad_norm": 196108.515625,
      "learning_rate": 3.799293286219082e-05,
      "loss": 1.1105,
      "step": 1700
    },
    {
      "epoch": 1.2367491166077738,
      "grad_norm": 141071.890625,
      "learning_rate": 3.763957597173145e-05,
      "loss": 1.0843,
      "step": 1750
    },
    {
      "epoch": 1.2720848056537102,
      "grad_norm": 132118.53125,
      "learning_rate": 3.728621908127209e-05,
      "loss": 1.1424,
      "step": 1800
    },
    {
      "epoch": 1.3074204946996466,
      "grad_norm": 160950.84375,
      "learning_rate": 3.693286219081272e-05,
      "loss": 1.1462,
      "step": 1850
    },
    {
      "epoch": 1.342756183745583,
      "grad_norm": 128804.078125,
      "learning_rate": 3.6579505300353355e-05,
      "loss": 1.0572,
      "step": 1900
    },
    {
      "epoch": 1.3780918727915195,
      "grad_norm": 157611.953125,
      "learning_rate": 3.6226148409893995e-05,
      "loss": 1.0987,
      "step": 1950
    },
    {
      "epoch": 1.4134275618374559,
      "grad_norm": 141798.90625,
      "learning_rate": 3.587279151943463e-05,
      "loss": 1.0541,
      "step": 2000
    },
    {
      "epoch": 1.4487632508833923,
      "grad_norm": 171915.71875,
      "learning_rate": 3.551943462897527e-05,
      "loss": 1.0487,
      "step": 2050
    },
    {
      "epoch": 1.4840989399293285,
      "grad_norm": 127521.7109375,
      "learning_rate": 3.51660777385159e-05,
      "loss": 0.9808,
      "step": 2100
    },
    {
      "epoch": 1.5194346289752652,
      "grad_norm": 144515.828125,
      "learning_rate": 3.4812720848056544e-05,
      "loss": 0.9844,
      "step": 2150
    },
    {
      "epoch": 1.5547703180212014,
      "grad_norm": 135642.578125,
      "learning_rate": 3.445936395759717e-05,
      "loss": 0.9523,
      "step": 2200
    },
    {
      "epoch": 1.5901060070671378,
      "grad_norm": 211087.34375,
      "learning_rate": 3.410600706713781e-05,
      "loss": 0.9904,
      "step": 2250
    },
    {
      "epoch": 1.6254416961130742,
      "grad_norm": 157448.421875,
      "learning_rate": 3.3752650176678445e-05,
      "loss": 1.0143,
      "step": 2300
    },
    {
      "epoch": 1.6607773851590106,
      "grad_norm": 185968.609375,
      "learning_rate": 3.3399293286219085e-05,
      "loss": 1.0191,
      "step": 2350
    },
    {
      "epoch": 1.696113074204947,
      "grad_norm": 138438.078125,
      "learning_rate": 3.304593639575972e-05,
      "loss": 1.0524,
      "step": 2400
    },
    {
      "epoch": 1.7314487632508833,
      "grad_norm": 151158.640625,
      "learning_rate": 3.269257950530035e-05,
      "loss": 0.9748,
      "step": 2450
    },
    {
      "epoch": 1.76678445229682,
      "grad_norm": 146153.25,
      "learning_rate": 3.233922261484099e-05,
      "loss": 0.9595,
      "step": 2500
    },
    {
      "epoch": 1.802120141342756,
      "grad_norm": 145571.171875,
      "learning_rate": 3.198586572438163e-05,
      "loss": 0.9884,
      "step": 2550
    },
    {
      "epoch": 1.8374558303886925,
      "grad_norm": 200024.28125,
      "learning_rate": 3.163250883392227e-05,
      "loss": 0.9961,
      "step": 2600
    },
    {
      "epoch": 1.872791519434629,
      "grad_norm": 1255579.375,
      "learning_rate": 3.1279151943462894e-05,
      "loss": 0.9381,
      "step": 2650
    },
    {
      "epoch": 1.9081272084805654,
      "grad_norm": 191306.1875,
      "learning_rate": 3.0925795053003535e-05,
      "loss": 0.9179,
      "step": 2700
    },
    {
      "epoch": 1.9434628975265018,
      "grad_norm": 139976.265625,
      "learning_rate": 3.057243816254417e-05,
      "loss": 0.9741,
      "step": 2750
    },
    {
      "epoch": 1.978798586572438,
      "grad_norm": 254590.484375,
      "learning_rate": 3.021908127208481e-05,
      "loss": 0.97,
      "step": 2800
    },
    {
      "epoch": 2.0,
      "eval_bertscore_f1": 0.9522204981784862,
      "eval_bleu": 0.4391766527339286,
      "eval_loss": 0.6113855838775635,
      "eval_meteor": 0.6141492660226205,
      "eval_rouge1": 0.7802485332799334,
      "eval_rouge2": 0.6456575876833133,
      "eval_runtime": 442.4344,
      "eval_samples_per_second": 14.61,
      "eval_steps_per_second": 0.913,
      "step": 2830
    },
    {
      "epoch": 2.0141342756183747,
      "grad_norm": 251050.734375,
      "learning_rate": 2.9865724381625443e-05,
      "loss": 0.9362,
      "step": 2850
    },
    {
      "epoch": 2.049469964664311,
      "grad_norm": 165134.1875,
      "learning_rate": 2.9512367491166083e-05,
      "loss": 0.9642,
      "step": 2900
    },
    {
      "epoch": 2.0848056537102475,
      "grad_norm": 167977.453125,
      "learning_rate": 2.9159010600706717e-05,
      "loss": 0.899,
      "step": 2950
    },
    {
      "epoch": 2.1201413427561837,
      "grad_norm": 137508.921875,
      "learning_rate": 2.8805653710247347e-05,
      "loss": 0.8828,
      "step": 3000
    },
    {
      "epoch": 2.1554770318021204,
      "grad_norm": 155117.59375,
      "learning_rate": 2.8452296819787988e-05,
      "loss": 0.8958,
      "step": 3050
    },
    {
      "epoch": 2.1908127208480566,
      "grad_norm": 155701.109375,
      "learning_rate": 2.809893992932862e-05,
      "loss": 0.9106,
      "step": 3100
    },
    {
      "epoch": 2.2261484098939928,
      "grad_norm": 148495.125,
      "learning_rate": 2.7745583038869262e-05,
      "loss": 0.9755,
      "step": 3150
    },
    {
      "epoch": 2.2614840989399294,
      "grad_norm": 179815.203125,
      "learning_rate": 2.7392226148409892e-05,
      "loss": 0.9842,
      "step": 3200
    },
    {
      "epoch": 2.2968197879858656,
      "grad_norm": 193006.0625,
      "learning_rate": 2.7038869257950533e-05,
      "loss": 0.8794,
      "step": 3250
    },
    {
      "epoch": 2.3321554770318023,
      "grad_norm": 181638.421875,
      "learning_rate": 2.6685512367491167e-05,
      "loss": 0.9415,
      "step": 3300
    },
    {
      "epoch": 2.3674911660777385,
      "grad_norm": 172265.5,
      "learning_rate": 2.6332155477031807e-05,
      "loss": 0.9453,
      "step": 3350
    },
    {
      "epoch": 2.402826855123675,
      "grad_norm": 142922.71875,
      "learning_rate": 2.597879858657244e-05,
      "loss": 0.8669,
      "step": 3400
    },
    {
      "epoch": 2.4381625441696113,
      "grad_norm": 153024.515625,
      "learning_rate": 2.562544169611307e-05,
      "loss": 0.9168,
      "step": 3450
    },
    {
      "epoch": 2.4734982332155475,
      "grad_norm": 325320.28125,
      "learning_rate": 2.5272084805653712e-05,
      "loss": 0.8523,
      "step": 3500
    },
    {
      "epoch": 2.508833922261484,
      "grad_norm": 180908.15625,
      "learning_rate": 2.491872791519435e-05,
      "loss": 0.8853,
      "step": 3550
    },
    {
      "epoch": 2.5441696113074204,
      "grad_norm": 171192.765625,
      "learning_rate": 2.4565371024734986e-05,
      "loss": 0.9041,
      "step": 3600
    },
    {
      "epoch": 2.579505300353357,
      "grad_norm": 148674.578125,
      "learning_rate": 2.421201413427562e-05,
      "loss": 0.865,
      "step": 3650
    },
    {
      "epoch": 2.614840989399293,
      "grad_norm": 189248.390625,
      "learning_rate": 2.3858657243816253e-05,
      "loss": 0.9255,
      "step": 3700
    },
    {
      "epoch": 2.65017667844523,
      "grad_norm": 159843.796875,
      "learning_rate": 2.350530035335689e-05,
      "loss": 0.8886,
      "step": 3750
    },
    {
      "epoch": 2.685512367491166,
      "grad_norm": 162496.359375,
      "learning_rate": 2.3151943462897528e-05,
      "loss": 0.8689,
      "step": 3800
    },
    {
      "epoch": 2.7208480565371023,
      "grad_norm": 135053.5625,
      "learning_rate": 2.2798586572438165e-05,
      "loss": 0.8275,
      "step": 3850
    },
    {
      "epoch": 2.756183745583039,
      "grad_norm": 249403.953125,
      "learning_rate": 2.24452296819788e-05,
      "loss": 0.8319,
      "step": 3900
    },
    {
      "epoch": 2.791519434628975,
      "grad_norm": 181885.328125,
      "learning_rate": 2.2091872791519436e-05,
      "loss": 0.8719,
      "step": 3950
    },
    {
      "epoch": 2.8268551236749118,
      "grad_norm": 156204.75,
      "learning_rate": 2.1738515901060073e-05,
      "loss": 0.8592,
      "step": 4000
    },
    {
      "epoch": 2.862190812720848,
      "grad_norm": 159614.03125,
      "learning_rate": 2.138515901060071e-05,
      "loss": 0.8381,
      "step": 4050
    },
    {
      "epoch": 2.8975265017667846,
      "grad_norm": 162379.78125,
      "learning_rate": 2.1031802120141344e-05,
      "loss": 0.8612,
      "step": 4100
    },
    {
      "epoch": 2.932862190812721,
      "grad_norm": 149608.25,
      "learning_rate": 2.067844522968198e-05,
      "loss": 0.838,
      "step": 4150
    },
    {
      "epoch": 2.968197879858657,
      "grad_norm": 145233.40625,
      "learning_rate": 2.0325088339222618e-05,
      "loss": 0.8411,
      "step": 4200
    },
    {
      "epoch": 3.0,
      "eval_bertscore_f1": 0.955309976823628,
      "eval_bleu": 0.47422911395097656,
      "eval_loss": 0.5273156762123108,
      "eval_meteor": 0.6439931966505934,
      "eval_rouge1": 0.7960674625670938,
      "eval_rouge2": 0.6807866947718111,
      "eval_runtime": 440.5737,
      "eval_samples_per_second": 14.672,
      "eval_steps_per_second": 0.917,
      "step": 4245
    },
    {
      "epoch": 3.0035335689045937,
      "grad_norm": 146904.109375,
      "learning_rate": 1.997173144876325e-05,
      "loss": 0.8212,
      "step": 4250
    },
    {
      "epoch": 3.03886925795053,
      "grad_norm": 141972.140625,
      "learning_rate": 1.9618374558303885e-05,
      "loss": 0.8332,
      "step": 4300
    },
    {
      "epoch": 3.0742049469964665,
      "grad_norm": 155438.34375,
      "learning_rate": 1.9265017667844522e-05,
      "loss": 0.8314,
      "step": 4350
    },
    {
      "epoch": 3.1095406360424027,
      "grad_norm": 157805.5,
      "learning_rate": 1.891166077738516e-05,
      "loss": 0.8432,
      "step": 4400
    },
    {
      "epoch": 3.1448763250883394,
      "grad_norm": 171890.296875,
      "learning_rate": 1.8558303886925797e-05,
      "loss": 0.8171,
      "step": 4450
    },
    {
      "epoch": 3.1802120141342756,
      "grad_norm": 159529.09375,
      "learning_rate": 1.8204946996466434e-05,
      "loss": 0.8462,
      "step": 4500
    },
    {
      "epoch": 3.215547703180212,
      "grad_norm": 291361.3125,
      "learning_rate": 1.7851590106007067e-05,
      "loss": 0.8525,
      "step": 4550
    },
    {
      "epoch": 3.2508833922261484,
      "grad_norm": 146320.515625,
      "learning_rate": 1.7498233215547705e-05,
      "loss": 0.8691,
      "step": 4600
    },
    {
      "epoch": 3.2862190812720846,
      "grad_norm": 142328.234375,
      "learning_rate": 1.714487632508834e-05,
      "loss": 0.8152,
      "step": 4650
    },
    {
      "epoch": 3.3215547703180213,
      "grad_norm": 164441.828125,
      "learning_rate": 1.679151943462898e-05,
      "loss": 0.7897,
      "step": 4700
    },
    {
      "epoch": 3.3568904593639575,
      "grad_norm": 181799.328125,
      "learning_rate": 1.643816254416961e-05,
      "loss": 0.89,
      "step": 4750
    },
    {
      "epoch": 3.392226148409894,
      "grad_norm": 161868.921875,
      "learning_rate": 1.6084805653710246e-05,
      "loss": 0.8342,
      "step": 4800
    },
    {
      "epoch": 3.4275618374558303,
      "grad_norm": 168403.046875,
      "learning_rate": 1.5731448763250883e-05,
      "loss": 0.7708,
      "step": 4850
    },
    {
      "epoch": 3.462897526501767,
      "grad_norm": 180798.71875,
      "learning_rate": 1.537809187279152e-05,
      "loss": 0.8452,
      "step": 4900
    },
    {
      "epoch": 3.498233215547703,
      "grad_norm": 194585.109375,
      "learning_rate": 1.5024734982332156e-05,
      "loss": 0.8543,
      "step": 4950
    },
    {
      "epoch": 3.53356890459364,
      "grad_norm": 162353.515625,
      "learning_rate": 1.4671378091872793e-05,
      "loss": 0.7865,
      "step": 5000
    },
    {
      "epoch": 3.568904593639576,
      "grad_norm": 161337.140625,
      "learning_rate": 1.4318021201413428e-05,
      "loss": 0.8459,
      "step": 5050
    },
    {
      "epoch": 3.604240282685512,
      "grad_norm": 174313.953125,
      "learning_rate": 1.3964664310954066e-05,
      "loss": 0.8914,
      "step": 5100
    },
    {
      "epoch": 3.639575971731449,
      "grad_norm": 163668.421875,
      "learning_rate": 1.3611307420494701e-05,
      "loss": 0.8457,
      "step": 5150
    },
    {
      "epoch": 3.674911660777385,
      "grad_norm": 164355.765625,
      "learning_rate": 1.3257950530035338e-05,
      "loss": 0.8266,
      "step": 5200
    },
    {
      "epoch": 3.7102473498233217,
      "grad_norm": 159919.046875,
      "learning_rate": 1.2904593639575973e-05,
      "loss": 0.7511,
      "step": 5250
    },
    {
      "epoch": 3.745583038869258,
      "grad_norm": 131121.03125,
      "learning_rate": 1.2551236749116607e-05,
      "loss": 0.8333,
      "step": 5300
    },
    {
      "epoch": 3.7809187279151946,
      "grad_norm": 160561.640625,
      "learning_rate": 1.2197879858657244e-05,
      "loss": 0.7992,
      "step": 5350
    },
    {
      "epoch": 3.8162544169611308,
      "grad_norm": 200662.859375,
      "learning_rate": 1.184452296819788e-05,
      "loss": 0.8494,
      "step": 5400
    },
    {
      "epoch": 3.851590106007067,
      "grad_norm": 158871.84375,
      "learning_rate": 1.1491166077738517e-05,
      "loss": 0.7689,
      "step": 5450
    },
    {
      "epoch": 3.8869257950530036,
      "grad_norm": 233291.453125,
      "learning_rate": 1.1137809187279152e-05,
      "loss": 0.7717,
      "step": 5500
    },
    {
      "epoch": 3.92226148409894,
      "grad_norm": 152039.84375,
      "learning_rate": 1.078445229681979e-05,
      "loss": 0.7991,
      "step": 5550
    },
    {
      "epoch": 3.9575971731448765,
      "grad_norm": 174898.390625,
      "learning_rate": 1.0431095406360425e-05,
      "loss": 0.7713,
      "step": 5600
    },
    {
      "epoch": 3.9929328621908127,
      "grad_norm": 185653.25,
      "learning_rate": 1.0077738515901062e-05,
      "loss": 0.8271,
      "step": 5650
    },
    {
      "epoch": 4.0,
      "eval_bertscore_f1": 0.9574022400700072,
      "eval_bleu": 0.49288907686192807,
      "eval_loss": 0.4886058270931244,
      "eval_meteor": 0.6607653874114606,
      "eval_rouge1": 0.8043075620747033,
      "eval_rouge2": 0.698973717724064,
      "eval_runtime": 437.782,
      "eval_samples_per_second": 14.765,
      "eval_steps_per_second": 0.923,
      "step": 5660
    },
    {
      "epoch": 4.028268551236749,
      "grad_norm": 189546.390625,
      "learning_rate": 9.724381625441696e-06,
      "loss": 0.8009,
      "step": 5700
    },
    {
      "epoch": 4.063604240282685,
      "grad_norm": 179902.28125,
      "learning_rate": 9.371024734982333e-06,
      "loss": 0.7955,
      "step": 5750
    },
    {
      "epoch": 4.098939929328622,
      "grad_norm": 134316.078125,
      "learning_rate": 9.017667844522968e-06,
      "loss": 0.8093,
      "step": 5800
    },
    {
      "epoch": 4.134275618374558,
      "grad_norm": 113333.8203125,
      "learning_rate": 8.664310954063605e-06,
      "loss": 0.7959,
      "step": 5850
    },
    {
      "epoch": 4.169611307420495,
      "grad_norm": 153999.46875,
      "learning_rate": 8.31095406360424e-06,
      "loss": 0.7883,
      "step": 5900
    },
    {
      "epoch": 4.204946996466431,
      "grad_norm": 284284.375,
      "learning_rate": 7.957597173144876e-06,
      "loss": 0.7858,
      "step": 5950
    },
    {
      "epoch": 4.240282685512367,
      "grad_norm": 139182.28125,
      "learning_rate": 7.6042402826855124e-06,
      "loss": 0.804,
      "step": 6000
    },
    {
      "epoch": 4.275618374558304,
      "grad_norm": 169185.328125,
      "learning_rate": 7.250883392226149e-06,
      "loss": 0.7745,
      "step": 6050
    },
    {
      "epoch": 4.310954063604241,
      "grad_norm": 200688.453125,
      "learning_rate": 6.897526501766785e-06,
      "loss": 0.7795,
      "step": 6100
    },
    {
      "epoch": 4.3462897526501765,
      "grad_norm": 135357.96875,
      "learning_rate": 6.544169611307421e-06,
      "loss": 0.8412,
      "step": 6150
    },
    {
      "epoch": 4.381625441696113,
      "grad_norm": 137914.25,
      "learning_rate": 6.190812720848057e-06,
      "loss": 0.811,
      "step": 6200
    },
    {
      "epoch": 4.41696113074205,
      "grad_norm": 179002.140625,
      "learning_rate": 5.837455830388693e-06,
      "loss": 0.7859,
      "step": 6250
    },
    {
      "epoch": 4.4522968197879855,
      "grad_norm": 139846.9375,
      "learning_rate": 5.484098939929329e-06,
      "loss": 0.801,
      "step": 6300
    },
    {
      "epoch": 4.487632508833922,
      "grad_norm": 161350.671875,
      "learning_rate": 5.1307420494699655e-06,
      "loss": 0.8203,
      "step": 6350
    },
    {
      "epoch": 4.522968197879859,
      "grad_norm": 153958.46875,
      "learning_rate": 4.777385159010601e-06,
      "loss": 0.7693,
      "step": 6400
    },
    {
      "epoch": 4.5583038869257955,
      "grad_norm": 154714.984375,
      "learning_rate": 4.424028268551237e-06,
      "loss": 0.7808,
      "step": 6450
    },
    {
      "epoch": 4.593639575971731,
      "grad_norm": 294880.1875,
      "learning_rate": 4.070671378091873e-06,
      "loss": 0.8194,
      "step": 6500
    },
    {
      "epoch": 4.628975265017668,
      "grad_norm": 480907.34375,
      "learning_rate": 3.717314487632509e-06,
      "loss": 0.8096,
      "step": 6550
    },
    {
      "epoch": 4.6643109540636045,
      "grad_norm": 319982.71875,
      "learning_rate": 3.363957597173145e-06,
      "loss": 0.7987,
      "step": 6600
    },
    {
      "epoch": 4.69964664310954,
      "grad_norm": 188937.359375,
      "learning_rate": 3.010600706713781e-06,
      "loss": 0.8162,
      "step": 6650
    },
    {
      "epoch": 4.734982332155477,
      "grad_norm": 166750.71875,
      "learning_rate": 2.6572438162544172e-06,
      "loss": 0.7726,
      "step": 6700
    },
    {
      "epoch": 4.770318021201414,
      "grad_norm": 179732.5625,
      "learning_rate": 2.303886925795053e-06,
      "loss": 0.7848,
      "step": 6750
    },
    {
      "epoch": 4.80565371024735,
      "grad_norm": 135768.640625,
      "learning_rate": 1.9505300353356893e-06,
      "loss": 0.8204,
      "step": 6800
    },
    {
      "epoch": 4.840989399293286,
      "grad_norm": 194958.0625,
      "learning_rate": 1.5971731448763252e-06,
      "loss": 0.8214,
      "step": 6850
    },
    {
      "epoch": 4.876325088339223,
      "grad_norm": 205979.40625,
      "learning_rate": 1.2438162544169612e-06,
      "loss": 0.7658,
      "step": 6900
    },
    {
      "epoch": 4.911660777385159,
      "grad_norm": 140958.078125,
      "learning_rate": 8.904593639575972e-07,
      "loss": 0.777,
      "step": 6950
    },
    {
      "epoch": 4.946996466431095,
      "grad_norm": 149027.953125,
      "learning_rate": 5.371024734982332e-07,
      "loss": 0.8286,
      "step": 7000
    },
    {
      "epoch": 4.982332155477032,
      "grad_norm": 161736.4375,
      "learning_rate": 1.8374558303886926e-07,
      "loss": 0.7922,
      "step": 7050
    },
    {
      "epoch": 5.0,
      "eval_bertscore_f1": 0.957730518736978,
      "eval_bleu": 0.49562010024897835,
      "eval_loss": 0.47903409600257874,
      "eval_meteor": 0.6644886697409358,
      "eval_rouge1": 0.8053692499714378,
      "eval_rouge2": 0.7018033389816531,
      "eval_runtime": 438.2906,
      "eval_samples_per_second": 14.748,
      "eval_steps_per_second": 0.922,
      "step": 7075
    }
  ],
  "logging_steps": 50,
  "max_steps": 7075,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.730048539557888e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}