{
  "best_global_step": 2000,
  "best_metric": 1.2956099257632903,
  "best_model_checkpoint": "./SALAMA_NEW6/checkpoint-2000",
  "epoch": 0.798881565807869,
  "eval_steps": 2000,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003994407829039345,
      "grad_norm": 3.5260863304138184,
      "learning_rate": 1.8e-07,
      "loss": 0.0171,
      "step": 10
    },
    {
      "epoch": 0.00798881565807869,
      "grad_norm": 3.8677453994750977,
      "learning_rate": 3.8e-07,
      "loss": 0.0168,
      "step": 20
    },
    {
      "epoch": 0.011983223487118035,
      "grad_norm": 2.2200398445129395,
      "learning_rate": 5.800000000000001e-07,
      "loss": 0.0143,
      "step": 30
    },
    {
      "epoch": 0.01597763131615738,
      "grad_norm": 2.8865885734558105,
      "learning_rate": 7.8e-07,
      "loss": 0.0148,
      "step": 40
    },
    {
      "epoch": 0.019972039145196723,
      "grad_norm": 2.37961483001709,
      "learning_rate": 9.800000000000001e-07,
      "loss": 0.0278,
      "step": 50
    },
    {
      "epoch": 0.02396644697423607,
      "grad_norm": 0.550510585308075,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 0.01,
      "step": 60
    },
    {
      "epoch": 0.027960854803275415,
      "grad_norm": 0.5346644520759583,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 0.0095,
      "step": 70
    },
    {
      "epoch": 0.03195526263231476,
      "grad_norm": 0.3471826910972595,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 0.0102,
      "step": 80
    },
    {
      "epoch": 0.035949670461354104,
      "grad_norm": 0.6894211769104004,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 0.0097,
      "step": 90
    },
    {
      "epoch": 0.039944078290393446,
      "grad_norm": 1.4083952903747559,
      "learning_rate": 1.98e-06,
      "loss": 0.0145,
      "step": 100
    },
    {
      "epoch": 0.043938486119432796,
      "grad_norm": 0.9561022520065308,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 0.0099,
      "step": 110
    },
    {
      "epoch": 0.04793289394847214,
      "grad_norm": 0.5757145285606384,
      "learning_rate": 2.38e-06,
      "loss": 0.0078,
      "step": 120
    },
    {
      "epoch": 0.05192730177751148,
      "grad_norm": 0.9376910924911499,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 0.0135,
      "step": 130
    },
    {
      "epoch": 0.05592170960655083,
      "grad_norm": 1.6390388011932373,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 0.0157,
      "step": 140
    },
    {
      "epoch": 0.05991611743559017,
      "grad_norm": 1.0386844873428345,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 0.0099,
      "step": 150
    },
    {
      "epoch": 0.06391052526462952,
      "grad_norm": 1.7207807302474976,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 0.0096,
      "step": 160
    },
    {
      "epoch": 0.06790493309366886,
      "grad_norm": 0.7114217877388,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 0.0137,
      "step": 170
    },
    {
      "epoch": 0.07189934092270821,
      "grad_norm": 2.062455654144287,
      "learning_rate": 3.58e-06,
      "loss": 0.0181,
      "step": 180
    },
    {
      "epoch": 0.07589374875174755,
      "grad_norm": 3.8748466968536377,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 0.0155,
      "step": 190
    },
    {
      "epoch": 0.07988815658078689,
      "grad_norm": 2.8697941303253174,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.0206,
      "step": 200
    },
    {
      "epoch": 0.08388256440982625,
      "grad_norm": 1.5111762285232544,
      "learning_rate": 4.18e-06,
      "loss": 0.014,
      "step": 210
    },
    {
      "epoch": 0.08787697223886559,
      "grad_norm": 2.7273306846618652,
      "learning_rate": 4.38e-06,
      "loss": 0.0115,
      "step": 220
    },
    {
      "epoch": 0.09187138006790493,
      "grad_norm": 1.604333758354187,
      "learning_rate": 4.58e-06,
      "loss": 0.0111,
      "step": 230
    },
    {
      "epoch": 0.09586578789694428,
      "grad_norm": 1.6656010150909424,
      "learning_rate": 4.78e-06,
      "loss": 0.0103,
      "step": 240
    },
    {
      "epoch": 0.09986019572598362,
      "grad_norm": 1.9466540813446045,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.0118,
      "step": 250
    },
    {
      "epoch": 0.10385460355502296,
      "grad_norm": 0.7741212844848633,
      "learning_rate": 5.18e-06,
      "loss": 0.0203,
      "step": 260
    },
    {
      "epoch": 0.10784901138406232,
      "grad_norm": 1.5594321489334106,
      "learning_rate": 5.380000000000001e-06,
      "loss": 0.0132,
      "step": 270
    },
    {
      "epoch": 0.11184341921310166,
      "grad_norm": 2.1854960918426514,
      "learning_rate": 5.580000000000001e-06,
      "loss": 0.0227,
      "step": 280
    },
    {
      "epoch": 0.115837827042141,
      "grad_norm": 1.809208869934082,
      "learning_rate": 5.78e-06,
      "loss": 0.014,
      "step": 290
    },
    {
      "epoch": 0.11983223487118035,
      "grad_norm": 2.003018617630005,
      "learning_rate": 5.98e-06,
      "loss": 0.0253,
      "step": 300
    },
    {
      "epoch": 0.12382664270021969,
      "grad_norm": 2.226447582244873,
      "learning_rate": 6.18e-06,
      "loss": 0.0141,
      "step": 310
    },
    {
      "epoch": 0.12782105052925904,
      "grad_norm": 3.213719367980957,
      "learning_rate": 6.380000000000001e-06,
      "loss": 0.0222,
      "step": 320
    },
    {
      "epoch": 0.1318154583582984,
      "grad_norm": 2.7083969116210938,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 0.0165,
      "step": 330
    },
    {
      "epoch": 0.13580986618733773,
      "grad_norm": 2.327775001525879,
      "learning_rate": 6.780000000000001e-06,
      "loss": 0.0127,
      "step": 340
    },
    {
      "epoch": 0.13980427401637707,
      "grad_norm": 2.199615478515625,
      "learning_rate": 6.98e-06,
      "loss": 0.0092,
      "step": 350
    },
    {
      "epoch": 0.14379868184541642,
      "grad_norm": 2.6200337409973145,
      "learning_rate": 7.180000000000001e-06,
      "loss": 0.0229,
      "step": 360
    },
    {
      "epoch": 0.14779308967445576,
      "grad_norm": 2.0899276733398438,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 0.0123,
      "step": 370
    },
    {
      "epoch": 0.1517874975034951,
      "grad_norm": 3.313321113586426,
      "learning_rate": 7.58e-06,
      "loss": 0.0225,
      "step": 380
    },
    {
      "epoch": 0.15578190533253444,
      "grad_norm": 2.0713999271392822,
      "learning_rate": 7.78e-06,
      "loss": 0.0166,
      "step": 390
    },
    {
      "epoch": 0.15977631316157379,
      "grad_norm": 3.2806386947631836,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.0143,
      "step": 400
    },
    {
      "epoch": 0.16377072099061313,
      "grad_norm": 2.053285598754883,
      "learning_rate": 8.18e-06,
      "loss": 0.0183,
      "step": 410
    },
    {
      "epoch": 0.1677651288196525,
      "grad_norm": 1.4635854959487915,
      "learning_rate": 8.380000000000001e-06,
      "loss": 0.0198,
      "step": 420
    },
    {
      "epoch": 0.17175953664869184,
      "grad_norm": 1.461671233177185,
      "learning_rate": 8.580000000000001e-06,
      "loss": 0.0223,
      "step": 430
    },
    {
      "epoch": 0.17575394447773118,
      "grad_norm": 1.910568118095398,
      "learning_rate": 8.78e-06,
      "loss": 0.0127,
      "step": 440
    },
    {
      "epoch": 0.17974835230677053,
      "grad_norm": 2.9844930171966553,
      "learning_rate": 8.98e-06,
      "loss": 0.0198,
      "step": 450
    },
    {
      "epoch": 0.18374276013580987,
      "grad_norm": 3.3782155513763428,
      "learning_rate": 9.180000000000002e-06,
      "loss": 0.0277,
      "step": 460
    },
    {
      "epoch": 0.1877371679648492,
      "grad_norm": 2.6012775897979736,
      "learning_rate": 9.38e-06,
      "loss": 0.0154,
      "step": 470
    },
    {
      "epoch": 0.19173157579388855,
      "grad_norm": 2.880096197128296,
      "learning_rate": 9.58e-06,
      "loss": 0.0219,
      "step": 480
    },
    {
      "epoch": 0.1957259836229279,
      "grad_norm": 3.002333879470825,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.0179,
      "step": 490
    },
    {
      "epoch": 0.19972039145196724,
      "grad_norm": 6.162044525146484,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.0165,
      "step": 500
    },
    {
      "epoch": 0.20371479928100658,
      "grad_norm": 3.4659605026245117,
      "learning_rate": 9.98716486023959e-06,
      "loss": 0.0239,
      "step": 510
    },
    {
      "epoch": 0.20770920711004592,
      "grad_norm": 3.2337770462036133,
      "learning_rate": 9.972903593839133e-06,
      "loss": 0.017,
      "step": 520
    },
    {
      "epoch": 0.2117036149390853,
      "grad_norm": 1.7297512292861938,
      "learning_rate": 9.958642327438678e-06,
      "loss": 0.0197,
      "step": 530
    },
    {
      "epoch": 0.21569802276812464,
      "grad_norm": 3.336660385131836,
      "learning_rate": 9.944381061038221e-06,
      "loss": 0.0215,
      "step": 540
    },
    {
      "epoch": 0.21969243059716398,
      "grad_norm": 1.850265383720398,
      "learning_rate": 9.930119794637765e-06,
      "loss": 0.0259,
      "step": 550
    },
    {
      "epoch": 0.22368683842620332,
      "grad_norm": 2.2153711318969727,
      "learning_rate": 9.91585852823731e-06,
      "loss": 0.0192,
      "step": 560
    },
    {
      "epoch": 0.22768124625524266,
      "grad_norm": 1.2140111923217773,
      "learning_rate": 9.901597261836851e-06,
      "loss": 0.0214,
      "step": 570
    },
    {
      "epoch": 0.231675654084282,
      "grad_norm": 2.983187198638916,
      "learning_rate": 9.887335995436396e-06,
      "loss": 0.0212,
      "step": 580
    },
    {
      "epoch": 0.23567006191332135,
      "grad_norm": 1.3873775005340576,
      "learning_rate": 9.87307472903594e-06,
      "loss": 0.0278,
      "step": 590
    },
    {
      "epoch": 0.2396644697423607,
      "grad_norm": 2.8896610736846924,
      "learning_rate": 9.858813462635483e-06,
      "loss": 0.0188,
      "step": 600
    },
    {
      "epoch": 0.24365887757140003,
      "grad_norm": 3.1558310985565186,
      "learning_rate": 9.844552196235026e-06,
      "loss": 0.0268,
      "step": 610
    },
    {
      "epoch": 0.24765328540043938,
      "grad_norm": 1.7268836498260498,
      "learning_rate": 9.83029092983457e-06,
      "loss": 0.0214,
      "step": 620
    },
    {
      "epoch": 0.2516476932294787,
      "grad_norm": 3.3516862392425537,
      "learning_rate": 9.816029663434114e-06,
      "loss": 0.0264,
      "step": 630
    },
    {
      "epoch": 0.2556421010585181,
      "grad_norm": 2.8272225856781006,
      "learning_rate": 9.801768397033657e-06,
      "loss": 0.0279,
      "step": 640
    },
    {
      "epoch": 0.2596365088875574,
      "grad_norm": 2.9199869632720947,
      "learning_rate": 9.787507130633202e-06,
      "loss": 0.0225,
      "step": 650
    },
    {
      "epoch": 0.2636309167165968,
      "grad_norm": 2.80952787399292,
      "learning_rate": 9.773245864232744e-06,
      "loss": 0.0188,
      "step": 660
    },
    {
      "epoch": 0.2676253245456361,
      "grad_norm": 2.8590474128723145,
      "learning_rate": 9.758984597832289e-06,
      "loss": 0.0214,
      "step": 670
    },
    {
      "epoch": 0.27161973237467546,
      "grad_norm": 2.1354668140411377,
      "learning_rate": 9.744723331431832e-06,
      "loss": 0.0241,
      "step": 680
    },
    {
      "epoch": 0.2756141402037148,
      "grad_norm": 1.54200279712677,
      "learning_rate": 9.730462065031375e-06,
      "loss": 0.0185,
      "step": 690
    },
    {
      "epoch": 0.27960854803275415,
      "grad_norm": 2.3145532608032227,
      "learning_rate": 9.71620079863092e-06,
      "loss": 0.0162,
      "step": 700
    },
    {
      "epoch": 0.2836029558617935,
      "grad_norm": 2.7305352687835693,
      "learning_rate": 9.701939532230463e-06,
      "loss": 0.0246,
      "step": 710
    },
    {
      "epoch": 0.28759736369083283,
      "grad_norm": 1.983603596687317,
      "learning_rate": 9.687678265830007e-06,
      "loss": 0.0236,
      "step": 720
    },
    {
      "epoch": 0.2915917715198722,
      "grad_norm": 1.5119189023971558,
      "learning_rate": 9.67341699942955e-06,
      "loss": 0.02,
      "step": 730
    },
    {
      "epoch": 0.2955861793489115,
      "grad_norm": 2.9574663639068604,
      "learning_rate": 9.659155733029095e-06,
      "loss": 0.024,
      "step": 740
    },
    {
      "epoch": 0.2995805871779509,
      "grad_norm": 2.4619083404541016,
      "learning_rate": 9.644894466628636e-06,
      "loss": 0.0267,
      "step": 750
    },
    {
      "epoch": 0.3035749950069902,
      "grad_norm": 1.9842376708984375,
      "learning_rate": 9.630633200228181e-06,
      "loss": 0.0183,
      "step": 760
    },
    {
      "epoch": 0.30756940283602957,
      "grad_norm": 2.300018310546875,
      "learning_rate": 9.616371933827725e-06,
      "loss": 0.022,
      "step": 770
    },
    {
      "epoch": 0.3115638106650689,
      "grad_norm": 1.9513983726501465,
      "learning_rate": 9.602110667427268e-06,
      "loss": 0.0183,
      "step": 780
    },
    {
      "epoch": 0.31555821849410826,
      "grad_norm": 1.9509272575378418,
      "learning_rate": 9.587849401026813e-06,
      "loss": 0.0239,
      "step": 790
    },
    {
      "epoch": 0.31955262632314757,
      "grad_norm": 2.2186226844787598,
      "learning_rate": 9.573588134626356e-06,
      "loss": 0.0244,
      "step": 800
    },
    {
      "epoch": 0.32354703415218694,
      "grad_norm": 3.058155059814453,
      "learning_rate": 9.5593268682259e-06,
      "loss": 0.0225,
      "step": 810
    },
    {
      "epoch": 0.32754144198122626,
      "grad_norm": 2.130378484725952,
      "learning_rate": 9.545065601825442e-06,
      "loss": 0.0302,
      "step": 820
    },
    {
      "epoch": 0.3315358498102656,
      "grad_norm": 2.9102509021759033,
      "learning_rate": 9.530804335424987e-06,
      "loss": 0.0223,
      "step": 830
    },
    {
      "epoch": 0.335530257639305,
      "grad_norm": 3.1255991458892822,
      "learning_rate": 9.516543069024529e-06,
      "loss": 0.0199,
      "step": 840
    },
    {
      "epoch": 0.3395246654683443,
      "grad_norm": 1.9874292612075806,
      "learning_rate": 9.502281802624074e-06,
      "loss": 0.0214,
      "step": 850
    },
    {
      "epoch": 0.3435190732973837,
      "grad_norm": 1.9070899486541748,
      "learning_rate": 9.488020536223617e-06,
      "loss": 0.0283,
      "step": 860
    },
    {
      "epoch": 0.347513481126423,
      "grad_norm": 2.783308744430542,
      "learning_rate": 9.47375926982316e-06,
      "loss": 0.0172,
      "step": 870
    },
    {
      "epoch": 0.35150788895546237,
      "grad_norm": 1.1049232482910156,
      "learning_rate": 9.459498003422705e-06,
      "loss": 0.0261,
      "step": 880
    },
    {
      "epoch": 0.3555022967845017,
      "grad_norm": 3.386366844177246,
      "learning_rate": 9.445236737022249e-06,
      "loss": 0.0214,
      "step": 890
    },
    {
      "epoch": 0.35949670461354105,
      "grad_norm": 1.623195767402649,
      "learning_rate": 9.430975470621792e-06,
      "loss": 0.0221,
      "step": 900
    },
    {
      "epoch": 0.36349111244258037,
      "grad_norm": 3.33766508102417,
      "learning_rate": 9.416714204221335e-06,
      "loss": 0.0245,
      "step": 910
    },
    {
      "epoch": 0.36748552027161974,
      "grad_norm": 3.1946351528167725,
      "learning_rate": 9.40245293782088e-06,
      "loss": 0.0225,
      "step": 920
    },
    {
      "epoch": 0.37147992810065905,
      "grad_norm": 2.7116000652313232,
      "learning_rate": 9.388191671420423e-06,
      "loss": 0.0224,
      "step": 930
    },
    {
      "epoch": 0.3754743359296984,
      "grad_norm": 1.7920254468917847,
      "learning_rate": 9.373930405019966e-06,
      "loss": 0.024,
      "step": 940
    },
    {
      "epoch": 0.3794687437587378,
      "grad_norm": 3.112621545791626,
      "learning_rate": 9.35966913861951e-06,
      "loss": 0.0179,
      "step": 950
    },
    {
      "epoch": 0.3834631515877771,
      "grad_norm": 2.081601142883301,
      "learning_rate": 9.345407872219053e-06,
      "loss": 0.0237,
      "step": 960
    },
    {
      "epoch": 0.3874575594168165,
      "grad_norm": 2.1594202518463135,
      "learning_rate": 9.331146605818598e-06,
      "loss": 0.0286,
      "step": 970
    },
    {
      "epoch": 0.3914519672458558,
      "grad_norm": 2.277575969696045,
      "learning_rate": 9.316885339418141e-06,
      "loss": 0.0199,
      "step": 980
    },
    {
      "epoch": 0.39544637507489516,
      "grad_norm": 2.457400321960449,
      "learning_rate": 9.302624073017684e-06,
      "loss": 0.0148,
      "step": 990
    },
    {
      "epoch": 0.3994407829039345,
      "grad_norm": 2.731627941131592,
      "learning_rate": 9.288362806617228e-06,
      "loss": 0.026,
      "step": 1000
    },
    {
      "epoch": 0.40343519073297385,
      "grad_norm": 2.26240873336792,
      "learning_rate": 9.274101540216773e-06,
      "loss": 0.0205,
      "step": 1010
    },
    {
      "epoch": 0.40742959856201316,
      "grad_norm": 3.270246982574463,
      "learning_rate": 9.259840273816316e-06,
      "loss": 0.0245,
      "step": 1020
    },
    {
      "epoch": 0.41142400639105253,
      "grad_norm": 2.118734836578369,
      "learning_rate": 9.245579007415859e-06,
      "loss": 0.0244,
      "step": 1030
    },
    {
      "epoch": 0.41541841422009185,
      "grad_norm": 2.5819344520568848,
      "learning_rate": 9.231317741015402e-06,
      "loss": 0.0204,
      "step": 1040
    },
    {
      "epoch": 0.4194128220491312,
      "grad_norm": 3.1689200401306152,
      "learning_rate": 9.217056474614946e-06,
      "loss": 0.0283,
      "step": 1050
    },
    {
      "epoch": 0.4234072298781706,
      "grad_norm": 2.6276655197143555,
      "learning_rate": 9.20279520821449e-06,
      "loss": 0.022,
      "step": 1060
    },
    {
      "epoch": 0.4274016377072099,
      "grad_norm": 3.010298013687134,
      "learning_rate": 9.188533941814034e-06,
      "loss": 0.0355,
      "step": 1070
    },
    {
      "epoch": 0.4313960455362493,
      "grad_norm": 2.780543327331543,
      "learning_rate": 9.174272675413579e-06,
      "loss": 0.0341,
      "step": 1080
    },
    {
      "epoch": 0.4353904533652886,
      "grad_norm": 2.277057647705078,
      "learning_rate": 9.16001140901312e-06,
      "loss": 0.0239,
      "step": 1090
    },
    {
      "epoch": 0.43938486119432796,
      "grad_norm": 1.640129566192627,
      "learning_rate": 9.145750142612665e-06,
      "loss": 0.0228,
      "step": 1100
    },
    {
      "epoch": 0.4433792690233673,
      "grad_norm": 3.9390289783477783,
      "learning_rate": 9.131488876212208e-06,
      "loss": 0.0299,
      "step": 1110
    },
    {
      "epoch": 0.44737367685240664,
      "grad_norm": 1.726510763168335,
      "learning_rate": 9.117227609811752e-06,
      "loss": 0.0207,
      "step": 1120
    },
    {
      "epoch": 0.45136808468144596,
      "grad_norm": 1.583237648010254,
      "learning_rate": 9.102966343411297e-06,
      "loss": 0.018,
      "step": 1130
    },
    {
      "epoch": 0.45536249251048533,
      "grad_norm": 3.3998148441314697,
      "learning_rate": 9.08870507701084e-06,
      "loss": 0.0333,
      "step": 1140
    },
    {
      "epoch": 0.45935690033952464,
      "grad_norm": 1.7249724864959717,
      "learning_rate": 9.074443810610383e-06,
      "loss": 0.0244,
      "step": 1150
    },
    {
      "epoch": 0.463351308168564,
      "grad_norm": 2.0827536582946777,
      "learning_rate": 9.060182544209926e-06,
      "loss": 0.0295,
      "step": 1160
    },
    {
      "epoch": 0.46734571599760333,
      "grad_norm": 2.095423698425293,
      "learning_rate": 9.045921277809471e-06,
      "loss": 0.025,
      "step": 1170
    },
    {
      "epoch": 0.4713401238266427,
      "grad_norm": 2.7629899978637695,
      "learning_rate": 9.031660011409013e-06,
      "loss": 0.0139,
      "step": 1180
    },
    {
      "epoch": 0.47533453165568207,
      "grad_norm": 1.6056689023971558,
      "learning_rate": 9.017398745008558e-06,
      "loss": 0.0379,
      "step": 1190
    },
    {
      "epoch": 0.4793289394847214,
      "grad_norm": 2.5653021335601807,
      "learning_rate": 9.003137478608101e-06,
      "loss": 0.0236,
      "step": 1200
    },
    {
      "epoch": 0.48332334731376075,
      "grad_norm": 2.9385814666748047,
      "learning_rate": 8.988876212207644e-06,
      "loss": 0.035,
      "step": 1210
    },
    {
      "epoch": 0.48731775514280007,
      "grad_norm": 3.0189478397369385,
      "learning_rate": 8.97461494580719e-06,
      "loss": 0.0187,
      "step": 1220
    },
    {
      "epoch": 0.49131216297183944,
      "grad_norm": 1.1368498802185059,
      "learning_rate": 8.960353679406733e-06,
      "loss": 0.0205,
      "step": 1230
    },
    {
      "epoch": 0.49530657080087875,
      "grad_norm": 2.0715737342834473,
      "learning_rate": 8.946092413006276e-06,
      "loss": 0.021,
      "step": 1240
    },
    {
      "epoch": 0.4993009786299181,
      "grad_norm": 2.2337212562561035,
      "learning_rate": 8.931831146605819e-06,
      "loss": 0.0216,
      "step": 1250
    },
    {
      "epoch": 0.5032953864589574,
      "grad_norm": 1.4322441816329956,
      "learning_rate": 8.917569880205364e-06,
      "loss": 0.0188,
      "step": 1260
    },
    {
      "epoch": 0.5072897942879968,
      "grad_norm": 1.9030108451843262,
      "learning_rate": 8.903308613804906e-06,
      "loss": 0.0177,
      "step": 1270
    },
    {
      "epoch": 0.5112842021170362,
      "grad_norm": 2.806227684020996,
      "learning_rate": 8.88904734740445e-06,
      "loss": 0.0166,
      "step": 1280
    },
    {
      "epoch": 0.5152786099460755,
      "grad_norm": 3.692470073699951,
      "learning_rate": 8.874786081003994e-06,
      "loss": 0.032,
      "step": 1290
    },
    {
      "epoch": 0.5192730177751148,
      "grad_norm": 2.1846556663513184,
      "learning_rate": 8.860524814603537e-06,
      "loss": 0.0203,
      "step": 1300
    },
    {
      "epoch": 0.5232674256041542,
      "grad_norm": 2.7069506645202637,
      "learning_rate": 8.846263548203082e-06,
      "loss": 0.0216,
      "step": 1310
    },
    {
      "epoch": 0.5272618334331935,
      "grad_norm": 3.010519027709961,
      "learning_rate": 8.832002281802625e-06,
      "loss": 0.0291,
      "step": 1320
    },
    {
      "epoch": 0.5312562412622329,
      "grad_norm": 2.396578073501587,
      "learning_rate": 8.817741015402168e-06,
      "loss": 0.0315,
      "step": 1330
    },
    {
      "epoch": 0.5352506490912722,
      "grad_norm": 3.0465903282165527,
      "learning_rate": 8.803479749001712e-06,
      "loss": 0.0229,
      "step": 1340
    },
    {
      "epoch": 0.5392450569203115,
      "grad_norm": 3.0167880058288574,
      "learning_rate": 8.789218482601257e-06,
      "loss": 0.0237,
      "step": 1350
    },
    {
      "epoch": 0.5432394647493509,
      "grad_norm": 2.1383025646209717,
      "learning_rate": 8.7749572162008e-06,
      "loss": 0.0204,
      "step": 1360
    },
    {
      "epoch": 0.5472338725783903,
      "grad_norm": 4.080812931060791,
      "learning_rate": 8.760695949800343e-06,
      "loss": 0.0289,
      "step": 1370
    },
    {
      "epoch": 0.5512282804074295,
      "grad_norm": 1.5765025615692139,
      "learning_rate": 8.746434683399886e-06,
      "loss": 0.0266,
      "step": 1380
    },
    {
      "epoch": 0.5552226882364689,
      "grad_norm": 2.26448917388916,
      "learning_rate": 8.73217341699943e-06,
      "loss": 0.0209,
      "step": 1390
    },
    {
      "epoch": 0.5592170960655083,
      "grad_norm": 2.3659119606018066,
      "learning_rate": 8.717912150598975e-06,
      "loss": 0.027,
      "step": 1400
    },
    {
      "epoch": 0.5632115038945477,
      "grad_norm": 4.025057315826416,
      "learning_rate": 8.703650884198518e-06,
      "loss": 0.0349,
      "step": 1410
    },
    {
      "epoch": 0.567205911723587,
      "grad_norm": 1.6939404010772705,
      "learning_rate": 8.689389617798061e-06,
      "loss": 0.0164,
      "step": 1420
    },
    {
      "epoch": 0.5712003195526263,
      "grad_norm": 2.2709405422210693,
      "learning_rate": 8.675128351397604e-06,
      "loss": 0.0239,
      "step": 1430
    },
    {
      "epoch": 0.5751947273816657,
      "grad_norm": 1.9596635103225708,
      "learning_rate": 8.66086708499715e-06,
      "loss": 0.0223,
      "step": 1440
    },
    {
      "epoch": 0.579189135210705,
      "grad_norm": 1.6292943954467773,
      "learning_rate": 8.646605818596692e-06,
      "loss": 0.0211,
      "step": 1450
    },
    {
      "epoch": 0.5831835430397444,
      "grad_norm": 2.063108205795288,
      "learning_rate": 8.632344552196236e-06,
      "loss": 0.0187,
      "step": 1460
    },
    {
      "epoch": 0.5871779508687837,
      "grad_norm": 2.243898868560791,
      "learning_rate": 8.618083285795779e-06,
      "loss": 0.0201,
      "step": 1470
    },
    {
      "epoch": 0.591172358697823,
      "grad_norm": 2.1789815425872803,
      "learning_rate": 8.603822019395322e-06,
      "loss": 0.0275,
      "step": 1480
    },
    {
      "epoch": 0.5951667665268624,
      "grad_norm": 2.9697136878967285,
      "learning_rate": 8.589560752994867e-06,
      "loss": 0.0265,
      "step": 1490
    },
    {
      "epoch": 0.5991611743559018,
      "grad_norm": 2.9582135677337646,
      "learning_rate": 8.57529948659441e-06,
      "loss": 0.0222,
      "step": 1500
    },
    {
      "epoch": 0.603155582184941,
      "grad_norm": 3.5698182582855225,
      "learning_rate": 8.561038220193954e-06,
      "loss": 0.0356,
      "step": 1510
    },
    {
      "epoch": 0.6071499900139804,
      "grad_norm": 1.6024861335754395,
      "learning_rate": 8.546776953793497e-06,
      "loss": 0.0262,
      "step": 1520
    },
    {
      "epoch": 0.6111443978430198,
      "grad_norm": 3.0098986625671387,
      "learning_rate": 8.532515687393042e-06,
      "loss": 0.0193,
      "step": 1530
    },
    {
      "epoch": 0.6151388056720591,
      "grad_norm": 2.4644837379455566,
      "learning_rate": 8.518254420992585e-06,
      "loss": 0.0228,
      "step": 1540
    },
    {
      "epoch": 0.6191332135010985,
      "grad_norm": 1.8468703031539917,
      "learning_rate": 8.503993154592128e-06,
      "loss": 0.0205,
      "step": 1550
    },
    {
      "epoch": 0.6231276213301378,
      "grad_norm": 1.8541923761367798,
      "learning_rate": 8.489731888191672e-06,
      "loss": 0.0288,
      "step": 1560
    },
    {
      "epoch": 0.6271220291591771,
      "grad_norm": 2.5148401260375977,
      "learning_rate": 8.475470621791215e-06,
      "loss": 0.0236,
      "step": 1570
    },
    {
      "epoch": 0.6311164369882165,
      "grad_norm": 1.897991418838501,
      "learning_rate": 8.46120935539076e-06,
      "loss": 0.0223,
      "step": 1580
    },
    {
      "epoch": 0.6351108448172559,
      "grad_norm": 2.4308152198791504,
      "learning_rate": 8.446948088990303e-06,
      "loss": 0.0212,
      "step": 1590
    },
    {
      "epoch": 0.6391052526462951,
      "grad_norm": 2.6001315116882324,
      "learning_rate": 8.432686822589846e-06,
      "loss": 0.0305,
      "step": 1600
    },
    {
      "epoch": 0.6430996604753345,
      "grad_norm": 2.6471099853515625,
      "learning_rate": 8.41842555618939e-06,
      "loss": 0.0217,
      "step": 1610
    },
    {
      "epoch": 0.6470940683043739,
      "grad_norm": 1.336553692817688,
      "learning_rate": 8.404164289788934e-06,
      "loss": 0.0219,
      "step": 1620
    },
    {
      "epoch": 0.6510884761334133,
      "grad_norm": 2.2210710048675537,
      "learning_rate": 8.389903023388478e-06,
      "loss": 0.0229,
      "step": 1630
    },
    {
      "epoch": 0.6550828839624525,
      "grad_norm": 1.4983543157577515,
      "learning_rate": 8.375641756988021e-06,
      "loss": 0.0229,
      "step": 1640
    },
    {
      "epoch": 0.6590772917914919,
      "grad_norm": 4.39826774597168,
      "learning_rate": 8.361380490587566e-06,
      "loss": 0.0258,
      "step": 1650
    },
    {
      "epoch": 0.6630716996205313,
      "grad_norm": 1.8835957050323486,
      "learning_rate": 8.347119224187107e-06,
      "loss": 0.0252,
      "step": 1660
    },
    {
      "epoch": 0.6670661074495706,
      "grad_norm": 2.090482473373413,
      "learning_rate": 8.332857957786652e-06,
      "loss": 0.0249,
      "step": 1670
    },
    {
      "epoch": 0.67106051527861,
      "grad_norm": 4.0251336097717285,
      "learning_rate": 8.318596691386196e-06,
      "loss": 0.0358,
      "step": 1680
    },
    {
      "epoch": 0.6750549231076493,
      "grad_norm": 2.3550221920013428,
      "learning_rate": 8.304335424985739e-06,
      "loss": 0.0257,
      "step": 1690
    },
    {
      "epoch": 0.6790493309366886,
      "grad_norm": 3.605212926864624,
      "learning_rate": 8.290074158585282e-06,
      "loss": 0.0243,
      "step": 1700
    },
    {
      "epoch": 0.683043738765728,
      "grad_norm": 3.9534785747528076,
      "learning_rate": 8.275812892184827e-06,
      "loss": 0.021,
      "step": 1710
    },
    {
      "epoch": 0.6870381465947674,
      "grad_norm": 2.5969812870025635,
      "learning_rate": 8.26155162578437e-06,
      "loss": 0.0202,
      "step": 1720
    },
    {
      "epoch": 0.6910325544238066,
      "grad_norm": 2.922255277633667,
      "learning_rate": 8.247290359383914e-06,
      "loss": 0.0183,
      "step": 1730
    },
    {
      "epoch": 0.695026962252846,
      "grad_norm": 2.6963930130004883,
      "learning_rate": 8.233029092983458e-06,
      "loss": 0.0288,
      "step": 1740
    },
    {
      "epoch": 0.6990213700818854,
      "grad_norm": 1.4426485300064087,
      "learning_rate": 8.218767826583002e-06,
      "loss": 0.0267,
      "step": 1750
    },
    {
      "epoch": 0.7030157779109247,
      "grad_norm": 1.6854075193405151,
      "learning_rate": 8.204506560182545e-06,
      "loss": 0.0189,
      "step": 1760
    },
    {
      "epoch": 0.7070101857399641,
      "grad_norm": 2.4257256984710693,
      "learning_rate": 8.190245293782088e-06,
      "loss": 0.0256,
      "step": 1770
    },
    {
      "epoch": 0.7110045935690034,
      "grad_norm": 2.0894641876220703,
      "learning_rate": 8.175984027381633e-06,
      "loss": 0.02,
      "step": 1780
    },
    {
      "epoch": 0.7149990013980427,
      "grad_norm": 2.5761284828186035,
      "learning_rate": 8.161722760981175e-06,
      "loss": 0.0205,
      "step": 1790
    },
    {
      "epoch": 0.7189934092270821,
      "grad_norm": 4.715319633483887,
      "learning_rate": 8.14746149458072e-06,
      "loss": 0.0219,
      "step": 1800
    },
    {
      "epoch": 0.7229878170561215,
      "grad_norm": 2.933713674545288,
      "learning_rate": 8.133200228180263e-06,
      "loss": 0.0215,
      "step": 1810
    },
    {
      "epoch": 0.7269822248851607,
      "grad_norm": 2.0981826782226562,
      "learning_rate": 8.118938961779806e-06,
      "loss": 0.0193,
      "step": 1820
    },
    {
      "epoch": 0.7309766327142001,
      "grad_norm": 3.5366568565368652,
      "learning_rate": 8.104677695379351e-06,
      "loss": 0.0287,
      "step": 1830
    },
    {
      "epoch": 0.7349710405432395,
      "grad_norm": 2.165713310241699,
      "learning_rate": 8.090416428978894e-06,
      "loss": 0.0183,
      "step": 1840
    },
    {
      "epoch": 0.7389654483722788,
      "grad_norm": 3.0615482330322266,
      "learning_rate": 8.076155162578438e-06,
      "loss": 0.0233,
      "step": 1850
    },
    {
      "epoch": 0.7429598562013181,
      "grad_norm": 2.1526572704315186,
      "learning_rate": 8.06189389617798e-06,
      "loss": 0.0316,
      "step": 1860
    },
    {
      "epoch": 0.7469542640303575,
      "grad_norm": 1.9894167184829712,
      "learning_rate": 8.047632629777526e-06,
      "loss": 0.0214,
      "step": 1870
    },
    {
      "epoch": 0.7509486718593968,
      "grad_norm": 1.5561473369598389,
      "learning_rate": 8.033371363377069e-06,
      "loss": 0.0246,
      "step": 1880
    },
    {
      "epoch": 0.7549430796884362,
      "grad_norm": 2.030789375305176,
      "learning_rate": 8.019110096976612e-06,
      "loss": 0.028,
      "step": 1890
    },
    {
      "epoch": 0.7589374875174756,
      "grad_norm": 2.139389991760254,
      "learning_rate": 8.004848830576156e-06,
      "loss": 0.0192,
      "step": 1900
    },
    {
      "epoch": 0.7629318953465148,
      "grad_norm": 1.4104957580566406,
      "learning_rate": 7.990587564175699e-06,
      "loss": 0.0214,
      "step": 1910
    },
    {
      "epoch": 0.7669263031755542,
      "grad_norm": 3.027559280395508,
      "learning_rate": 7.976326297775244e-06,
      "loss": 0.0225,
      "step": 1920
    },
    {
      "epoch": 0.7709207110045936,
      "grad_norm": 2.3780980110168457,
      "learning_rate": 7.962065031374787e-06,
      "loss": 0.0231,
      "step": 1930
    },
    {
      "epoch": 0.774915118833633,
      "grad_norm": 2.3442487716674805,
      "learning_rate": 7.94780376497433e-06,
      "loss": 0.0223,
      "step": 1940
    },
    {
      "epoch": 0.7789095266626722,
      "grad_norm": 2.0246195793151855,
      "learning_rate": 7.933542498573873e-06,
      "loss": 0.0164,
      "step": 1950
    },
    {
      "epoch": 0.7829039344917116,
      "grad_norm": 1.8316524028778076,
      "learning_rate": 7.919281232173418e-06,
      "loss": 0.019,
      "step": 1960
    },
    {
      "epoch": 0.786898342320751,
      "grad_norm": 1.7914390563964844,
      "learning_rate": 7.905019965772962e-06,
      "loss": 0.0178,
      "step": 1970
    },
    {
      "epoch": 0.7908927501497903,
      "grad_norm": 1.9735348224639893,
      "learning_rate": 7.890758699372505e-06,
      "loss": 0.0144,
      "step": 1980
    },
    {
      "epoch": 0.7948871579788296,
      "grad_norm": 1.8981409072875977,
      "learning_rate": 7.876497432972048e-06,
      "loss": 0.0244,
      "step": 1990
    },
    {
      "epoch": 0.798881565807869,
      "grad_norm": 1.4974066019058228,
      "learning_rate": 7.862236166571591e-06,
      "loss": 0.0239,
      "step": 2000
    },
    {
      "epoch": 0.798881565807869,
      "eval_loss": 0.014320253394544125,
      "eval_runtime": 7268.4159,
      "eval_samples_per_second": 2.755,
      "eval_steps_per_second": 0.345,
      "eval_wer": 1.2956099257632903,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 7512,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.846946562048e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}