{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.103125664469488,
  "global_step": 24000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.771793054571226e-06,
      "loss": 11.8316,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.543586109142452e-06,
      "loss": 8.5077,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.3153791637136785e-06,
      "loss": 6.6343,
      "step": 150
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.087172218284904e-06,
      "loss": 6.0709,
      "step": 200
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.85896527285613e-06,
      "loss": 5.5047,
      "step": 250
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0630758327427357e-05,
      "loss": 4.9657,
      "step": 300
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2402551381998583e-05,
      "loss": 4.5669,
      "step": 350
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4174344436569808e-05,
      "loss": 4.0142,
      "step": 400
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.5946137491141037e-05,
      "loss": 3.5509,
      "step": 450
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.771793054571226e-05,
      "loss": 2.9313,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.948972360028349e-05,
      "loss": 2.3465,
      "step": 550
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.1261516654854714e-05,
      "loss": 1.7314,
      "step": 600
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.303330970942594e-05,
      "loss": 1.2661,
      "step": 650
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4805102763997165e-05,
      "loss": 1.0417,
      "step": 700
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.6576895818568394e-05,
      "loss": 0.9334,
      "step": 750
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.8348688873139616e-05,
      "loss": 0.8874,
      "step": 800
    },
    {
      "epoch": 0.18,
      "learning_rate": 3.012048192771085e-05,
      "loss": 0.8745,
      "step": 850
    },
    {
      "epoch": 0.19,
      "learning_rate": 3.1892274982282074e-05,
      "loss": 0.8545,
      "step": 900
    },
    {
      "epoch": 0.2,
      "learning_rate": 3.36640680368533e-05,
      "loss": 0.8327,
      "step": 950
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.543586109142452e-05,
      "loss": 0.7961,
      "step": 1000
    },
    {
      "epoch": 0.22,
      "learning_rate": 3.720765414599575e-05,
      "loss": 0.7932,
      "step": 1050
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.897944720056698e-05,
      "loss": 0.7677,
      "step": 1100
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.07512402551382e-05,
      "loss": 0.7398,
      "step": 1150
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.252303330970943e-05,
      "loss": 0.719,
      "step": 1200
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.4294826364280654e-05,
      "loss": 0.7188,
      "step": 1250
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.606661941885188e-05,
      "loss": 0.7027,
      "step": 1300
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.7838412473423105e-05,
      "loss": 0.688,
      "step": 1350
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.961020552799433e-05,
      "loss": 0.6751,
      "step": 1400
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.992725780579699e-05,
      "loss": 0.6552,
      "step": 1450
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.9833998582459806e-05,
      "loss": 0.6749,
      "step": 1500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.974073935912262e-05,
      "loss": 0.6607,
      "step": 1550
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9647480135785426e-05,
      "loss": 0.6949,
      "step": 1600
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.955422091244825e-05,
      "loss": 0.6694,
      "step": 1650
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.946096168911106e-05,
      "loss": 0.6725,
      "step": 1700
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.936770246577387e-05,
      "loss": 0.6419,
      "step": 1750
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.927444324243668e-05,
      "loss": 0.6336,
      "step": 1800
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.9181184019099494e-05,
      "loss": 0.6478,
      "step": 1850
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.90879247957623e-05,
      "loss": 0.6395,
      "step": 1900
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.8994665572425114e-05,
      "loss": 0.6237,
      "step": 1950
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.890140634908793e-05,
      "loss": 0.623,
      "step": 2000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.880814712575074e-05,
      "loss": 0.6267,
      "step": 2050
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.8714887902413555e-05,
      "loss": 0.6271,
      "step": 2100
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.862162867907636e-05,
      "loss": 0.6038,
      "step": 2150
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.8528369455739175e-05,
      "loss": 0.5901,
      "step": 2200
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.843511023240199e-05,
      "loss": 0.6193,
      "step": 2250
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.8341851009064796e-05,
      "loss": 0.5823,
      "step": 2300
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.824859178572761e-05,
      "loss": 0.5939,
      "step": 2350
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.815533256239042e-05,
      "loss": 0.5954,
      "step": 2400
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.8062073339053236e-05,
      "loss": 0.5818,
      "step": 2450
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.796881411571605e-05,
      "loss": 0.5803,
      "step": 2500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.787555489237886e-05,
      "loss": 0.5652,
      "step": 2550
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.778229566904167e-05,
      "loss": 0.5845,
      "step": 2600
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.7689036445704483e-05,
      "loss": 0.5669,
      "step": 2650
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.75957772223673e-05,
      "loss": 0.5835,
      "step": 2700
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.7502517999030104e-05,
      "loss": 0.5977,
      "step": 2750
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.740925877569292e-05,
      "loss": 0.5749,
      "step": 2800
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.731599955235573e-05,
      "loss": 0.5512,
      "step": 2850
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.7222740329018544e-05,
      "loss": 0.5496,
      "step": 2900
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.712948110568136e-05,
      "loss": 0.5624,
      "step": 2950
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.7036221882344165e-05,
      "loss": 0.5474,
      "step": 3000
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.694296265900698e-05,
      "loss": 0.5573,
      "step": 3050
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.684970343566979e-05,
      "loss": 0.5258,
      "step": 3100
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.67564442123326e-05,
      "loss": 0.5418,
      "step": 3150
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.666318498899541e-05,
      "loss": 0.5513,
      "step": 3200
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.6569925765658226e-05,
      "loss": 0.5413,
      "step": 3250
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.647666654232104e-05,
      "loss": 0.557,
      "step": 3300
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.638340731898385e-05,
      "loss": 0.5425,
      "step": 3350
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.629014809564666e-05,
      "loss": 0.5292,
      "step": 3400
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.619688887230947e-05,
      "loss": 0.5297,
      "step": 3450
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.6103629648972286e-05,
      "loss": 0.5469,
      "step": 3500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.601037042563509e-05,
      "loss": 0.5485,
      "step": 3550
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.591711120229791e-05,
      "loss": 0.5543,
      "step": 3600
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.582385197896072e-05,
      "loss": 0.5603,
      "step": 3650
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.5730592755623534e-05,
      "loss": 0.5547,
      "step": 3700
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.563733353228635e-05,
      "loss": 0.5245,
      "step": 3750
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.554407430894916e-05,
      "loss": 0.5307,
      "step": 3800
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.545081508561197e-05,
      "loss": 0.5329,
      "step": 3850
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.535755586227478e-05,
      "loss": 0.5214,
      "step": 3900
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.5264296638937595e-05,
      "loss": 0.5135,
      "step": 3950
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.51710374156004e-05,
      "loss": 0.5261,
      "step": 4000
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.5077778192263215e-05,
      "loss": 0.5193,
      "step": 4050
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.498451896892603e-05,
      "loss": 0.5196,
      "step": 4100
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.489125974558884e-05,
      "loss": 0.5249,
      "step": 4150
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.4798000522251656e-05,
      "loss": 0.5147,
      "step": 4200
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.470474129891446e-05,
      "loss": 0.5277,
      "step": 4250
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.4611482075577276e-05,
      "loss": 0.509,
      "step": 4300
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.451822285224009e-05,
      "loss": 0.5099,
      "step": 4350
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.4424963628902896e-05,
      "loss": 0.5178,
      "step": 4400
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.433170440556571e-05,
      "loss": 0.49,
      "step": 4450
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.423844518222852e-05,
      "loss": 0.4989,
      "step": 4500
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.414518595889134e-05,
      "loss": 0.5077,
      "step": 4550
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.405192673555415e-05,
      "loss": 0.4866,
      "step": 4600
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.3958667512216964e-05,
      "loss": 0.5038,
      "step": 4650
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.386540828887977e-05,
      "loss": 0.4938,
      "step": 4700
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3772149065542584e-05,
      "loss": 0.516,
      "step": 4750
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.36788898422054e-05,
      "loss": 0.4992,
      "step": 4800
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.3585630618868204e-05,
      "loss": 0.4833,
      "step": 4850
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.349237139553102e-05,
      "loss": 0.5047,
      "step": 4900
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.339911217219383e-05,
      "loss": 0.4872,
      "step": 4950
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.3305852948856645e-05,
      "loss": 0.5096,
      "step": 5000
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.321259372551946e-05,
      "loss": 0.4837,
      "step": 5050
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.3119334502182265e-05,
      "loss": 0.5084,
      "step": 5100
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.302607527884508e-05,
      "loss": 0.5009,
      "step": 5150
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.293281605550789e-05,
      "loss": 0.4796,
      "step": 5200
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.28395568321707e-05,
      "loss": 0.4956,
      "step": 5250
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.274629760883351e-05,
      "loss": 0.495,
      "step": 5300
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.265303838549633e-05,
      "loss": 0.4797,
      "step": 5350
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.255977916215914e-05,
      "loss": 0.5058,
      "step": 5400
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.246651993882195e-05,
      "loss": 0.5026,
      "step": 5450
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.237326071548477e-05,
      "loss": 0.4923,
      "step": 5500
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.2280001492147573e-05,
      "loss": 0.48,
      "step": 5550
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.218674226881039e-05,
      "loss": 0.4825,
      "step": 5600
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.20934830454732e-05,
      "loss": 0.4714,
      "step": 5650
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.200022382213601e-05,
      "loss": 0.4785,
      "step": 5700
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.190696459879883e-05,
      "loss": 0.4794,
      "step": 5750
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.1813705375461634e-05,
      "loss": 0.4608,
      "step": 5800
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.172044615212445e-05,
      "loss": 0.4848,
      "step": 5850
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.162718692878726e-05,
      "loss": 0.4667,
      "step": 5900
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.153392770545007e-05,
      "loss": 0.4831,
      "step": 5950
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.144066848211288e-05,
      "loss": 0.4827,
      "step": 6000
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.1347409258775695e-05,
      "loss": 0.4781,
      "step": 6050
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.12541500354385e-05,
      "loss": 0.4832,
      "step": 6100
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.116089081210132e-05,
      "loss": 0.4759,
      "step": 6150
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.1067631588764136e-05,
      "loss": 0.4902,
      "step": 6200
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.097437236542694e-05,
      "loss": 0.4718,
      "step": 6250
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.0881113142089756e-05,
      "loss": 0.4789,
      "step": 6300
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.078785391875257e-05,
      "loss": 0.4927,
      "step": 6350
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.0694594695415376e-05,
      "loss": 0.4849,
      "step": 6400
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.060133547207819e-05,
      "loss": 0.4829,
      "step": 6450
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.0508076248741003e-05,
      "loss": 0.4608,
      "step": 6500
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.041481702540381e-05,
      "loss": 0.4457,
      "step": 6550
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.032155780206663e-05,
      "loss": 0.4656,
      "step": 6600
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.022829857872944e-05,
      "loss": 0.4575,
      "step": 6650
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.013503935539225e-05,
      "loss": 0.4783,
      "step": 6700
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.0041780132055064e-05,
      "loss": 0.4552,
      "step": 6750
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.994852090871787e-05,
      "loss": 0.471,
      "step": 6800
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.9855261685380685e-05,
      "loss": 0.4664,
      "step": 6850
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.97620024620435e-05,
      "loss": 0.4632,
      "step": 6900
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.9668743238706305e-05,
      "loss": 0.4694,
      "step": 6950
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.9575484015369125e-05,
      "loss": 0.467,
      "step": 7000
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.948222479203194e-05,
      "loss": 0.4647,
      "step": 7050
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.9388965568694745e-05,
      "loss": 0.4709,
      "step": 7100
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.929570634535756e-05,
      "loss": 0.4626,
      "step": 7150
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.920244712202037e-05,
      "loss": 0.4739,
      "step": 7200
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.910918789868318e-05,
      "loss": 0.4663,
      "step": 7250
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.901592867534599e-05,
      "loss": 0.4595,
      "step": 7300
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.8922669452008806e-05,
      "loss": 0.4559,
      "step": 7350
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.882941022867162e-05,
      "loss": 0.4445,
      "step": 7400
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.8736151005334433e-05,
      "loss": 0.4566,
      "step": 7450
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.864289178199724e-05,
      "loss": 0.463,
      "step": 7500
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.8549632558660054e-05,
      "loss": 0.4581,
      "step": 7550
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.845637333532287e-05,
      "loss": 0.4501,
      "step": 7600
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.8363114111985674e-05,
      "loss": 0.4532,
      "step": 7650
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.826985488864849e-05,
      "loss": 0.4509,
      "step": 7700
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.81765956653113e-05,
      "loss": 0.4495,
      "step": 7750
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.8083336441974115e-05,
      "loss": 0.456,
      "step": 7800
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.799007721863693e-05,
      "loss": 0.4626,
      "step": 7850
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.7896817995299735e-05,
      "loss": 0.4543,
      "step": 7900
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.780355877196255e-05,
      "loss": 0.4517,
      "step": 7950
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.771029954862536e-05,
      "loss": 0.4389,
      "step": 8000
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.761704032528817e-05,
      "loss": 0.4449,
      "step": 8050
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.752378110195098e-05,
      "loss": 0.449,
      "step": 8100
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.7430521878613796e-05,
      "loss": 0.4399,
      "step": 8150
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.733726265527661e-05,
      "loss": 0.4434,
      "step": 8200
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.724400343193942e-05,
      "loss": 0.4484,
      "step": 8250
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7150744208602236e-05,
      "loss": 0.4711,
      "step": 8300
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.705748498526504e-05,
      "loss": 0.4502,
      "step": 8350
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.696422576192786e-05,
      "loss": 0.4454,
      "step": 8400
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.687096653859067e-05,
      "loss": 0.4556,
      "step": 8450
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.677770731525348e-05,
      "loss": 0.4426,
      "step": 8500
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.668444809191629e-05,
      "loss": 0.4454,
      "step": 8550
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.6591188868579104e-05,
      "loss": 0.4474,
      "step": 8600
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.649792964524192e-05,
      "loss": 0.4523,
      "step": 8650
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.640467042190473e-05,
      "loss": 0.4426,
      "step": 8700
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.631141119856754e-05,
      "loss": 0.4329,
      "step": 8750
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.621815197523035e-05,
      "loss": 0.4364,
      "step": 8800
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.6124892751893165e-05,
      "loss": 0.4405,
      "step": 8850
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.603163352855597e-05,
      "loss": 0.4351,
      "step": 8900
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.5938374305218785e-05,
      "loss": 0.4426,
      "step": 8950
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.58451150818816e-05,
      "loss": 0.4389,
      "step": 9000
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.575185585854441e-05,
      "loss": 0.433,
      "step": 9050
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.5658596635207226e-05,
      "loss": 0.4389,
      "step": 9100
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.556533741187004e-05,
      "loss": 0.4375,
      "step": 9150
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.5472078188532846e-05,
      "loss": 0.4467,
      "step": 9200
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.537881896519566e-05,
      "loss": 0.4402,
      "step": 9250
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.528555974185847e-05,
      "loss": 0.4522,
      "step": 9300
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.519230051852128e-05,
      "loss": 0.4408,
      "step": 9350
    },
    {
      "epoch": 2.0,
      "learning_rate": 3.5099041295184093e-05,
      "loss": 0.4273,
      "step": 9400
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.500578207184691e-05,
      "loss": 0.4579,
      "step": 9450
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.491252284850972e-05,
      "loss": 0.4349,
      "step": 9500
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.4819263625172534e-05,
      "loss": 0.434,
      "step": 9550
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.472600440183534e-05,
      "loss": 0.4396,
      "step": 9600
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.4632745178498154e-05,
      "loss": 0.4251,
      "step": 9650
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.453948595516097e-05,
      "loss": 0.4495,
      "step": 9700
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.4446226731823775e-05,
      "loss": 0.4418,
      "step": 9750
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.435296750848659e-05,
      "loss": 0.4367,
      "step": 9800
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.425970828514941e-05,
      "loss": 0.4233,
      "step": 9850
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.4166449061812215e-05,
      "loss": 0.4383,
      "step": 9900
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.407318983847503e-05,
      "loss": 0.4368,
      "step": 9950
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.397993061513784e-05,
      "loss": 0.4389,
      "step": 10000
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.388667139180065e-05,
      "loss": 0.4226,
      "step": 10050
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.379341216846346e-05,
      "loss": 0.4251,
      "step": 10100
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.3700152945126276e-05,
      "loss": 0.4266,
      "step": 10150
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.360689372178908e-05,
      "loss": 0.418,
      "step": 10200
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.3513634498451896e-05,
      "loss": 0.4367,
      "step": 10250
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.342037527511471e-05,
      "loss": 0.4311,
      "step": 10300
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.3327116051777523e-05,
      "loss": 0.4308,
      "step": 10350
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.323385682844034e-05,
      "loss": 0.4272,
      "step": 10400
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3140597605103144e-05,
      "loss": 0.4381,
      "step": 10450
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.304733838176596e-05,
      "loss": 0.4253,
      "step": 10500
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.295407915842877e-05,
      "loss": 0.4238,
      "step": 10550
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.286081993509158e-05,
      "loss": 0.4157,
      "step": 10600
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.276756071175439e-05,
      "loss": 0.4425,
      "step": 10650
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.267430148841721e-05,
      "loss": 0.4352,
      "step": 10700
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.258104226508002e-05,
      "loss": 0.4191,
      "step": 10750
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.248778304174283e-05,
      "loss": 0.4153,
      "step": 10800
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.2394523818405645e-05,
      "loss": 0.4139,
      "step": 10850
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.230126459506845e-05,
      "loss": 0.4284,
      "step": 10900
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.2208005371731265e-05,
      "loss": 0.4046,
      "step": 10950
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.211474614839408e-05,
      "loss": 0.4049,
      "step": 11000
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.2021486925056886e-05,
      "loss": 0.4325,
      "step": 11050
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.1928227701719706e-05,
      "loss": 0.4308,
      "step": 11100
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.183496847838251e-05,
      "loss": 0.4257,
      "step": 11150
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.1741709255045326e-05,
      "loss": 0.4305,
      "step": 11200
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.164845003170814e-05,
      "loss": 0.4296,
      "step": 11250
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.155519080837095e-05,
      "loss": 0.4145,
      "step": 11300
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.146193158503376e-05,
      "loss": 0.4175,
      "step": 11350
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.1368672361696574e-05,
      "loss": 0.4275,
      "step": 11400
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.127541313835938e-05,
      "loss": 0.4268,
      "step": 11450
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.11821539150222e-05,
      "loss": 0.4368,
      "step": 11500
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.1088894691685014e-05,
      "loss": 0.4141,
      "step": 11550
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.099563546834782e-05,
      "loss": 0.4194,
      "step": 11600
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.0902376245010635e-05,
      "loss": 0.4042,
      "step": 11650
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.080911702167345e-05,
      "loss": 0.4298,
      "step": 11700
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.0715857798336255e-05,
      "loss": 0.4079,
      "step": 11750
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.062259857499907e-05,
      "loss": 0.3971,
      "step": 11800
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.052933935166188e-05,
      "loss": 0.408,
      "step": 11850
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.0436080128324695e-05,
      "loss": 0.4218,
      "step": 11900
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.0342820904987506e-05,
      "loss": 0.416,
      "step": 11950
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.024956168165032e-05,
      "loss": 0.4176,
      "step": 12000
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.015630245831313e-05,
      "loss": 0.4255,
      "step": 12050
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.006304323497594e-05,
      "loss": 0.4243,
      "step": 12100
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.9969784011638753e-05,
      "loss": 0.4186,
      "step": 12150
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.9876524788301563e-05,
      "loss": 0.4252,
      "step": 12200
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.9783265564964373e-05,
      "loss": 0.4235,
      "step": 12250
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.9690006341627187e-05,
      "loss": 0.4192,
      "step": 12300
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.9596747118290004e-05,
      "loss": 0.4281,
      "step": 12350
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.9503487894952814e-05,
      "loss": 0.4263,
      "step": 12400
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.9410228671615624e-05,
      "loss": 0.416,
      "step": 12450
    },
    {
      "epoch": 2.66,
      "learning_rate": 2.9316969448278437e-05,
      "loss": 0.4239,
      "step": 12500
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.9223710224941248e-05,
      "loss": 0.4011,
      "step": 12550
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.9130451001604058e-05,
      "loss": 0.4067,
      "step": 12600
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.903719177826687e-05,
      "loss": 0.4131,
      "step": 12650
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.894393255492968e-05,
      "loss": 0.4166,
      "step": 12700
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.88506733315925e-05,
      "loss": 0.3927,
      "step": 12750
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.875741410825531e-05,
      "loss": 0.4247,
      "step": 12800
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.8664154884918122e-05,
      "loss": 0.4248,
      "step": 12850
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.8570895661580932e-05,
      "loss": 0.4083,
      "step": 12900
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.8477636438243742e-05,
      "loss": 0.4248,
      "step": 12950
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.8384377214906556e-05,
      "loss": 0.4268,
      "step": 13000
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.8291117991569366e-05,
      "loss": 0.4194,
      "step": 13050
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.8197858768232176e-05,
      "loss": 0.43,
      "step": 13100
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.8104599544894993e-05,
      "loss": 0.4088,
      "step": 13150
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.8011340321557807e-05,
      "loss": 0.4109,
      "step": 13200
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.7918081098220617e-05,
      "loss": 0.4114,
      "step": 13250
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7824821874883427e-05,
      "loss": 0.4075,
      "step": 13300
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.773156265154624e-05,
      "loss": 0.4106,
      "step": 13350
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.763830342820905e-05,
      "loss": 0.4024,
      "step": 13400
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.754504420487186e-05,
      "loss": 0.4268,
      "step": 13450
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.7451784981534674e-05,
      "loss": 0.4031,
      "step": 13500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.735852575819749e-05,
      "loss": 0.4139,
      "step": 13550
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.72652665348603e-05,
      "loss": 0.4121,
      "step": 13600
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.717200731152311e-05,
      "loss": 0.4,
      "step": 13650
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.7078748088185925e-05,
      "loss": 0.3982,
      "step": 13700
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.6985488864848735e-05,
      "loss": 0.4128,
      "step": 13750
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.6892229641511545e-05,
      "loss": 0.4064,
      "step": 13800
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.679897041817436e-05,
      "loss": 0.4245,
      "step": 13850
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.670571119483717e-05,
      "loss": 0.4022,
      "step": 13900
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.6612451971499986e-05,
      "loss": 0.4249,
      "step": 13950
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.6519192748162796e-05,
      "loss": 0.4185,
      "step": 14000
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.642593352482561e-05,
      "loss": 0.3955,
      "step": 14050
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.633267430148842e-05,
      "loss": 0.4135,
      "step": 14100
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.623941507815123e-05,
      "loss": 0.4045,
      "step": 14150
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.6146155854814043e-05,
      "loss": 0.4041,
      "step": 14200
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.6052896631476854e-05,
      "loss": 0.4,
      "step": 14250
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.5959637408139664e-05,
      "loss": 0.4053,
      "step": 14300
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.5866378184802477e-05,
      "loss": 0.3975,
      "step": 14350
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.577311896146529e-05,
      "loss": 0.4006,
      "step": 14400
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.5679859738128104e-05,
      "loss": 0.407,
      "step": 14450
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.5586600514790914e-05,
      "loss": 0.4176,
      "step": 14500
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.5493341291453725e-05,
      "loss": 0.4014,
      "step": 14550
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.5400082068116538e-05,
      "loss": 0.4111,
      "step": 14600
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.5306822844779348e-05,
      "loss": 0.3951,
      "step": 14650
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.521356362144216e-05,
      "loss": 0.403,
      "step": 14700
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.5120304398104972e-05,
      "loss": 0.4006,
      "step": 14750
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.502704517476779e-05,
      "loss": 0.4062,
      "step": 14800
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.4933785951430596e-05,
      "loss": 0.4242,
      "step": 14850
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.484052672809341e-05,
      "loss": 0.3955,
      "step": 14900
    },
    {
      "epoch": 3.18,
      "learning_rate": 2.4747267504756223e-05,
      "loss": 0.4014,
      "step": 14950
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.4654008281419033e-05,
      "loss": 0.4073,
      "step": 15000
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.4560749058081843e-05,
      "loss": 0.4224,
      "step": 15050
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.446748983474466e-05,
      "loss": 0.396,
      "step": 15100
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.437423061140747e-05,
      "loss": 0.3879,
      "step": 15150
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.428097138807028e-05,
      "loss": 0.3986,
      "step": 15200
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.4187712164733094e-05,
      "loss": 0.4013,
      "step": 15250
    },
    {
      "epoch": 3.25,
      "learning_rate": 2.4094452941395907e-05,
      "loss": 0.4129,
      "step": 15300
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.4001193718058717e-05,
      "loss": 0.385,
      "step": 15350
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.3907934494721527e-05,
      "loss": 0.3973,
      "step": 15400
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.381467527138434e-05,
      "loss": 0.4077,
      "step": 15450
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.3721416048047155e-05,
      "loss": 0.3829,
      "step": 15500
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.3628156824709965e-05,
      "loss": 0.3952,
      "step": 15550
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.3534897601372778e-05,
      "loss": 0.4166,
      "step": 15600
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.344163837803559e-05,
      "loss": 0.3802,
      "step": 15650
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.3348379154698402e-05,
      "loss": 0.402,
      "step": 15700
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.3255119931361212e-05,
      "loss": 0.3891,
      "step": 15750
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.3161860708024026e-05,
      "loss": 0.3961,
      "step": 15800
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.3068601484686836e-05,
      "loss": 0.4022,
      "step": 15850
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.297534226134965e-05,
      "loss": 0.3854,
      "step": 15900
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.2882083038012463e-05,
      "loss": 0.3939,
      "step": 15950
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.2788823814675273e-05,
      "loss": 0.4001,
      "step": 16000
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.2695564591338083e-05,
      "loss": 0.3946,
      "step": 16050
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.2602305368000897e-05,
      "loss": 0.4052,
      "step": 16100
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.250904614466371e-05,
      "loss": 0.3869,
      "step": 16150
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.241578692132652e-05,
      "loss": 0.4003,
      "step": 16200
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.232252769798933e-05,
      "loss": 0.3748,
      "step": 16250
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.2229268474652144e-05,
      "loss": 0.4055,
      "step": 16300
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.2136009251314957e-05,
      "loss": 0.4013,
      "step": 16350
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.2042750027977768e-05,
      "loss": 0.4041,
      "step": 16400
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.194949080464058e-05,
      "loss": 0.394,
      "step": 16450
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.185623158130339e-05,
      "loss": 0.4041,
      "step": 16500
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.1762972357966205e-05,
      "loss": 0.3931,
      "step": 16550
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.1669713134629015e-05,
      "loss": 0.3857,
      "step": 16600
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.157645391129183e-05,
      "loss": 0.3884,
      "step": 16650
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.148319468795464e-05,
      "loss": 0.3984,
      "step": 16700
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.1389935464617452e-05,
      "loss": 0.3918,
      "step": 16750
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.1296676241280262e-05,
      "loss": 0.4041,
      "step": 16800
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.1203417017943076e-05,
      "loss": 0.3781,
      "step": 16850
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.1110157794605886e-05,
      "loss": 0.3917,
      "step": 16900
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.10168985712687e-05,
      "loss": 0.4061,
      "step": 16950
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.0923639347931513e-05,
      "loss": 0.3916,
      "step": 17000
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.0830380124594323e-05,
      "loss": 0.3893,
      "step": 17050
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.0737120901257133e-05,
      "loss": 0.3926,
      "step": 17100
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.0643861677919947e-05,
      "loss": 0.391,
      "step": 17150
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.055060245458276e-05,
      "loss": 0.4,
      "step": 17200
    },
    {
      "epoch": 3.67,
      "learning_rate": 2.045734323124557e-05,
      "loss": 0.3907,
      "step": 17250
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.036408400790838e-05,
      "loss": 0.4068,
      "step": 17300
    },
    {
      "epoch": 3.69,
      "learning_rate": 2.0270824784571198e-05,
      "loss": 0.3918,
      "step": 17350
    },
    {
      "epoch": 3.7,
      "learning_rate": 2.0177565561234008e-05,
      "loss": 0.3924,
      "step": 17400
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.0084306337896818e-05,
      "loss": 0.3948,
      "step": 17450
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.999104711455963e-05,
      "loss": 0.3926,
      "step": 17500
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.9897787891222445e-05,
      "loss": 0.4048,
      "step": 17550
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.9804528667885255e-05,
      "loss": 0.3878,
      "step": 17600
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.9711269444548065e-05,
      "loss": 0.3833,
      "step": 17650
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.961801022121088e-05,
      "loss": 0.385,
      "step": 17700
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.9524750997873692e-05,
      "loss": 0.3816,
      "step": 17750
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.9431491774536502e-05,
      "loss": 0.389,
      "step": 17800
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.9338232551199316e-05,
      "loss": 0.3926,
      "step": 17850
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.9244973327862126e-05,
      "loss": 0.3939,
      "step": 17900
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.9151714104524936e-05,
      "loss": 0.3958,
      "step": 17950
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.905845488118775e-05,
      "loss": 0.3962,
      "step": 18000
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.8965195657850563e-05,
      "loss": 0.4012,
      "step": 18050
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.8871936434513373e-05,
      "loss": 0.4,
      "step": 18100
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.8778677211176184e-05,
      "loss": 0.3816,
      "step": 18150
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.8685417987839e-05,
      "loss": 0.4022,
      "step": 18200
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.859215876450181e-05,
      "loss": 0.3865,
      "step": 18250
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.849889954116462e-05,
      "loss": 0.3896,
      "step": 18300
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.8405640317827434e-05,
      "loss": 0.4095,
      "step": 18350
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.8312381094490248e-05,
      "loss": 0.3945,
      "step": 18400
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.8219121871153058e-05,
      "loss": 0.4106,
      "step": 18450
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.8125862647815868e-05,
      "loss": 0.3846,
      "step": 18500
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.803260342447868e-05,
      "loss": 0.3833,
      "step": 18550
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.7939344201141495e-05,
      "loss": 0.3827,
      "step": 18600
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.7846084977804305e-05,
      "loss": 0.3851,
      "step": 18650
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.775282575446712e-05,
      "loss": 0.3882,
      "step": 18700
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.765956653112993e-05,
      "loss": 0.3743,
      "step": 18750
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.7566307307792743e-05,
      "loss": 0.3919,
      "step": 18800
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.7473048084455553e-05,
      "loss": 0.3906,
      "step": 18850
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.7379788861118366e-05,
      "loss": 0.3767,
      "step": 18900
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.7286529637781176e-05,
      "loss": 0.3705,
      "step": 18950
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.719327041444399e-05,
      "loss": 0.3727,
      "step": 19000
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.71000111911068e-05,
      "loss": 0.3846,
      "step": 19050
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.7006751967769614e-05,
      "loss": 0.3694,
      "step": 19100
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.6913492744432424e-05,
      "loss": 0.3776,
      "step": 19150
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.6820233521095237e-05,
      "loss": 0.3787,
      "step": 19200
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.672697429775805e-05,
      "loss": 0.3818,
      "step": 19250
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.663371507442086e-05,
      "loss": 0.3862,
      "step": 19300
    },
    {
      "epoch": 4.11,
      "learning_rate": 1.654045585108367e-05,
      "loss": 0.3697,
      "step": 19350
    },
    {
      "epoch": 4.13,
      "learning_rate": 1.6447196627746485e-05,
      "loss": 0.3797,
      "step": 19400
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.6353937404409298e-05,
      "loss": 0.3881,
      "step": 19450
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.6260678181072108e-05,
      "loss": 0.3924,
      "step": 19500
    },
    {
      "epoch": 4.16,
      "learning_rate": 1.616741895773492e-05,
      "loss": 0.3804,
      "step": 19550
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.6074159734397735e-05,
      "loss": 0.3899,
      "step": 19600
    },
    {
      "epoch": 4.18,
      "learning_rate": 1.5980900511060546e-05,
      "loss": 0.3864,
      "step": 19650
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.5887641287723356e-05,
      "loss": 0.3918,
      "step": 19700
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.579438206438617e-05,
      "loss": 0.387,
      "step": 19750
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.570112284104898e-05,
      "loss": 0.3936,
      "step": 19800
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.5607863617711793e-05,
      "loss": 0.3697,
      "step": 19850
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.5514604394374603e-05,
      "loss": 0.381,
      "step": 19900
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.5421345171037417e-05,
      "loss": 0.3829,
      "step": 19950
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.5328085947700227e-05,
      "loss": 0.3891,
      "step": 20000
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.523482672436304e-05,
      "loss": 0.3946,
      "step": 20050
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.5141567501025852e-05,
      "loss": 0.3913,
      "step": 20100
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.5048308277688664e-05,
      "loss": 0.3778,
      "step": 20150
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.4955049054351474e-05,
      "loss": 0.3867,
      "step": 20200
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.486178983101429e-05,
      "loss": 0.4,
      "step": 20250
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.47685306076771e-05,
      "loss": 0.3632,
      "step": 20300
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.4675271384339911e-05,
      "loss": 0.3735,
      "step": 20350
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.4582012161002723e-05,
      "loss": 0.3829,
      "step": 20400
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.4488752937665537e-05,
      "loss": 0.3725,
      "step": 20450
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.4395493714328348e-05,
      "loss": 0.4004,
      "step": 20500
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.4302234490991159e-05,
      "loss": 0.3915,
      "step": 20550
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.420897526765397e-05,
      "loss": 0.3941,
      "step": 20600
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.4115716044316784e-05,
      "loss": 0.3845,
      "step": 20650
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.4022456820979596e-05,
      "loss": 0.3779,
      "step": 20700
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.3929197597642408e-05,
      "loss": 0.3843,
      "step": 20750
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.3835938374305218e-05,
      "loss": 0.4076,
      "step": 20800
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.3742679150968033e-05,
      "loss": 0.3701,
      "step": 20850
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.3649419927630843e-05,
      "loss": 0.3765,
      "step": 20900
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.3556160704293655e-05,
      "loss": 0.386,
      "step": 20950
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.3462901480956467e-05,
      "loss": 0.3738,
      "step": 21000
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.336964225761928e-05,
      "loss": 0.3785,
      "step": 21050
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.3276383034282092e-05,
      "loss": 0.3901,
      "step": 21100
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.3183123810944902e-05,
      "loss": 0.3876,
      "step": 21150
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.3089864587607714e-05,
      "loss": 0.3769,
      "step": 21200
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.2996605364270528e-05,
      "loss": 0.3808,
      "step": 21250
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.290334614093334e-05,
      "loss": 0.3727,
      "step": 21300
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.2810086917596151e-05,
      "loss": 0.3604,
      "step": 21350
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.2716827694258962e-05,
      "loss": 0.3852,
      "step": 21400
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.2623568470921777e-05,
      "loss": 0.3731,
      "step": 21450
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.2530309247584587e-05,
      "loss": 0.3604,
      "step": 21500
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.2437050024247399e-05,
      "loss": 0.3891,
      "step": 21550
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.234379080091021e-05,
      "loss": 0.3605,
      "step": 21600
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.2250531577573022e-05,
      "loss": 0.3812,
      "step": 21650
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.2157272354235834e-05,
      "loss": 0.3714,
      "step": 21700
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.2064013130898646e-05,
      "loss": 0.3836,
      "step": 21750
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.1970753907561458e-05,
      "loss": 0.3876,
      "step": 21800
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.187749468422427e-05,
      "loss": 0.3899,
      "step": 21850
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.1784235460887082e-05,
      "loss": 0.3924,
      "step": 21900
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.1690976237549895e-05,
      "loss": 0.3722,
      "step": 21950
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.1597717014212705e-05,
      "loss": 0.3681,
      "step": 22000
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.1504457790875519e-05,
      "loss": 0.3682,
      "step": 22050
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.1411198567538329e-05,
      "loss": 0.3911,
      "step": 22100
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.1317939344201142e-05,
      "loss": 0.3843,
      "step": 22150
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.1224680120863954e-05,
      "loss": 0.3684,
      "step": 22200
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.1131420897526766e-05,
      "loss": 0.369,
      "step": 22250
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.1038161674189578e-05,
      "loss": 0.3831,
      "step": 22300
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.094490245085239e-05,
      "loss": 0.3821,
      "step": 22350
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.0851643227515202e-05,
      "loss": 0.386,
      "step": 22400
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.0758384004178013e-05,
      "loss": 0.375,
      "step": 22450
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.0665124780840825e-05,
      "loss": 0.3873,
      "step": 22500
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.0571865557503637e-05,
      "loss": 0.3766,
      "step": 22550
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.0478606334166449e-05,
      "loss": 0.3804,
      "step": 22600
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.0385347110829263e-05,
      "loss": 0.3724,
      "step": 22650
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.0292087887492073e-05,
      "loss": 0.3775,
      "step": 22700
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.0198828664154886e-05,
      "loss": 0.3718,
      "step": 22750
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.0105569440817696e-05,
      "loss": 0.3664,
      "step": 22800
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.001231021748051e-05,
      "loss": 0.3841,
      "step": 22850
    },
    {
      "epoch": 4.87,
      "learning_rate": 9.919050994143322e-06,
      "loss": 0.3722,
      "step": 22900
    },
    {
      "epoch": 4.88,
      "learning_rate": 9.825791770806134e-06,
      "loss": 0.3717,
      "step": 22950
    },
    {
      "epoch": 4.89,
      "learning_rate": 9.732532547468945e-06,
      "loss": 0.3665,
      "step": 23000
    },
    {
      "epoch": 4.9,
      "learning_rate": 9.639273324131757e-06,
      "loss": 0.3687,
      "step": 23050
    },
    {
      "epoch": 4.91,
      "learning_rate": 9.546014100794569e-06,
      "loss": 0.3823,
      "step": 23100
    },
    {
      "epoch": 4.92,
      "learning_rate": 9.452754877457381e-06,
      "loss": 0.3726,
      "step": 23150
    },
    {
      "epoch": 4.93,
      "learning_rate": 9.359495654120193e-06,
      "loss": 0.359,
      "step": 23200
    },
    {
      "epoch": 4.94,
      "learning_rate": 9.266236430783006e-06,
      "loss": 0.371,
      "step": 23250
    },
    {
      "epoch": 4.95,
      "learning_rate": 9.172977207445816e-06,
      "loss": 0.377,
      "step": 23300
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.07971798410863e-06,
      "loss": 0.3793,
      "step": 23350
    },
    {
      "epoch": 4.98,
      "learning_rate": 8.98645876077144e-06,
      "loss": 0.379,
      "step": 23400
    },
    {
      "epoch": 4.99,
      "learning_rate": 8.893199537434254e-06,
      "loss": 0.3718,
      "step": 23450
    },
    {
      "epoch": 5.0,
      "learning_rate": 8.799940314097064e-06,
      "loss": 0.3749,
      "step": 23500
    },
    {
      "epoch": 5.01,
      "learning_rate": 8.706681090759877e-06,
      "loss": 0.3659,
      "step": 23550
    },
    {
      "epoch": 5.02,
      "learning_rate": 8.613421867422689e-06,
      "loss": 0.3753,
      "step": 23600
    },
    {
      "epoch": 5.03,
      "learning_rate": 8.5201626440855e-06,
      "loss": 0.3629,
      "step": 23650
    },
    {
      "epoch": 5.04,
      "learning_rate": 8.426903420748313e-06,
      "loss": 0.3733,
      "step": 23700
    },
    {
      "epoch": 5.05,
      "learning_rate": 8.333644197411123e-06,
      "loss": 0.3886,
      "step": 23750
    },
    {
      "epoch": 5.06,
      "learning_rate": 8.240384974073936e-06,
      "loss": 0.3641,
      "step": 23800
    },
    {
      "epoch": 5.07,
      "learning_rate": 8.147125750736748e-06,
      "loss": 0.3761,
      "step": 23850
    },
    {
      "epoch": 5.08,
      "learning_rate": 8.05386652739956e-06,
      "loss": 0.3617,
      "step": 23900
    },
    {
      "epoch": 5.09,
      "learning_rate": 7.960607304062372e-06,
      "loss": 0.3726,
      "step": 23950
    },
    {
      "epoch": 5.1,
      "learning_rate": 7.867348080725184e-06,
      "loss": 0.3749,
      "step": 24000
    }
  ],
  "max_steps": 28218,
  "num_train_epochs": 6,
  "total_flos": 9.599266478715476e+17,
  "trial_name": null,
  "trial_params": null
}