{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.999989076887786,
  "global_step": 228870,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 2.993664525713287e-06,
      "loss": 1.7501,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.987329051426574e-06,
      "loss": 1.6451,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 2.980993577139861e-06,
      "loss": 1.6171,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.974658102853148e-06,
      "loss": 1.5995,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.968322628566435e-06,
      "loss": 1.5905,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.9619871542797224e-06,
      "loss": 1.5876,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.9556516799930094e-06,
      "loss": 1.579,
      "step": 3500
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.9493162057062964e-06,
      "loss": 1.5714,
      "step": 4000
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.9429807314195834e-06,
      "loss": 1.5668,
      "step": 4500
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.9366452571328704e-06,
      "loss": 1.5624,
      "step": 5000
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.9303097828461574e-06,
      "loss": 1.558,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.9239743085594444e-06,
      "loss": 1.5558,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.9176388342727313e-06,
      "loss": 1.5515,
      "step": 6500
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.9113033599860183e-06,
      "loss": 1.5473,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.9049678856993053e-06,
      "loss": 1.5436,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.8986324114125923e-06,
      "loss": 1.5402,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "learning_rate": 2.8922969371258797e-06,
      "loss": 1.539,
      "step": 8500
    },
    {
      "epoch": 0.39,
      "learning_rate": 2.8859614628391667e-06,
      "loss": 1.5364,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.8796259885524533e-06,
      "loss": 1.5345,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.8732905142657403e-06,
      "loss": 1.5297,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.8669550399790273e-06,
      "loss": 1.528,
      "step": 10500
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.8606195656923142e-06,
      "loss": 1.5248,
      "step": 11000
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8542840914056012e-06,
      "loss": 1.5221,
      "step": 11500
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.8479486171188882e-06,
      "loss": 1.5202,
      "step": 12000
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.8416131428321756e-06,
      "loss": 1.5188,
      "step": 12500
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.8352776685454626e-06,
      "loss": 1.5164,
      "step": 13000
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.8289421942587496e-06,
      "loss": 1.5152,
      "step": 13500
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.8226067199720366e-06,
      "loss": 1.5142,
      "step": 14000
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.8162712456853236e-06,
      "loss": 1.5113,
      "step": 14500
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.8099357713986106e-06,
      "loss": 1.5098,
      "step": 15000
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.8036002971118976e-06,
      "loss": 1.508,
      "step": 15500
    },
    {
      "epoch": 0.7,
      "learning_rate": 2.7972648228251845e-06,
      "loss": 1.5063,
      "step": 16000
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.7909293485384715e-06,
      "loss": 1.5051,
      "step": 16500
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.7845938742517585e-06,
      "loss": 1.5044,
      "step": 17000
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7782583999650455e-06,
      "loss": 1.503,
      "step": 17500
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.771922925678333e-06,
      "loss": 1.5013,
      "step": 18000
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.76558745139162e-06,
      "loss": 1.5011,
      "step": 18500
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.759251977104907e-06,
      "loss": 1.5004,
      "step": 19000
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.752916502818194e-06,
      "loss": 1.4989,
      "step": 19500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.746581028531481e-06,
      "loss": 1.498,
      "step": 20000
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.740245554244768e-06,
      "loss": 1.4968,
      "step": 20500
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.733910079958055e-06,
      "loss": 1.4963,
      "step": 21000
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.727574605671342e-06,
      "loss": 1.4961,
      "step": 21500
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.721239131384629e-06,
      "loss": 1.4944,
      "step": 22000
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.714903657097916e-06,
      "loss": 1.4936,
      "step": 22500
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.4786614179611206,
      "eval_runtime": 11.8277,
      "eval_samples_per_second": 83.364,
      "step": 22887
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.7085681828112032e-06,
      "loss": 1.4936,
      "step": 23000
    },
    {
      "epoch": 1.03,
      "learning_rate": 2.7022327085244902e-06,
      "loss": 1.4907,
      "step": 23500
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.695897234237777e-06,
      "loss": 1.4898,
      "step": 24000
    },
    {
      "epoch": 1.07,
      "learning_rate": 2.689561759951064e-06,
      "loss": 1.4892,
      "step": 24500
    },
    {
      "epoch": 1.09,
      "learning_rate": 2.683226285664351e-06,
      "loss": 1.488,
      "step": 25000
    },
    {
      "epoch": 1.11,
      "learning_rate": 2.6768908113776377e-06,
      "loss": 1.4877,
      "step": 25500
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.6705553370909247e-06,
      "loss": 1.4874,
      "step": 26000
    },
    {
      "epoch": 1.16,
      "learning_rate": 2.6642198628042117e-06,
      "loss": 1.4866,
      "step": 26500
    },
    {
      "epoch": 1.18,
      "learning_rate": 2.6578843885174987e-06,
      "loss": 1.4856,
      "step": 27000
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.651548914230786e-06,
      "loss": 1.4851,
      "step": 27500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.645213439944073e-06,
      "loss": 1.4848,
      "step": 28000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.63887796565736e-06,
      "loss": 1.4843,
      "step": 28500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.632542491370647e-06,
      "loss": 1.4842,
      "step": 29000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.626207017083934e-06,
      "loss": 1.4836,
      "step": 29500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.619871542797221e-06,
      "loss": 1.4829,
      "step": 30000
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.613536068510508e-06,
      "loss": 1.4823,
      "step": 30500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.607200594223795e-06,
      "loss": 1.482,
      "step": 31000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.600865119937082e-06,
      "loss": 1.4813,
      "step": 31500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.594529645650369e-06,
      "loss": 1.4812,
      "step": 32000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.5881941713636564e-06,
      "loss": 1.4808,
      "step": 32500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.5818586970769434e-06,
      "loss": 1.48,
      "step": 33000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5755232227902304e-06,
      "loss": 1.4797,
      "step": 33500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5691877485035174e-06,
      "loss": 1.4788,
      "step": 34000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.5628522742168044e-06,
      "loss": 1.4792,
      "step": 34500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.5565167999300914e-06,
      "loss": 1.4783,
      "step": 35000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.5501813256433784e-06,
      "loss": 1.4772,
      "step": 35500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.5438458513566653e-06,
      "loss": 1.4778,
      "step": 36000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.5375103770699523e-06,
      "loss": 1.4772,
      "step": 36500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5311749027832393e-06,
      "loss": 1.4767,
      "step": 37000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5248394284965263e-06,
      "loss": 1.4764,
      "step": 37500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.5185039542098137e-06,
      "loss": 1.4758,
      "step": 38000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.5121684799231007e-06,
      "loss": 1.4755,
      "step": 38500
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.5058330056363877e-06,
      "loss": 1.4754,
      "step": 39000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.4994975313496747e-06,
      "loss": 1.4745,
      "step": 39500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.4931620570629617e-06,
      "loss": 1.4747,
      "step": 40000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.4868265827762487e-06,
      "loss": 1.4741,
      "step": 40500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.4804911084895356e-06,
      "loss": 1.4737,
      "step": 41000
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.4741556342028226e-06,
      "loss": 1.4734,
      "step": 41500
    },
    {
      "epoch": 1.84,
      "learning_rate": 2.4678201599161096e-06,
      "loss": 1.4731,
      "step": 42000
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.4614846856293966e-06,
      "loss": 1.4732,
      "step": 42500
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.4551492113426836e-06,
      "loss": 1.472,
      "step": 43000
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.448813737055971e-06,
      "loss": 1.4717,
      "step": 43500
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.442478262769258e-06,
      "loss": 1.4715,
      "step": 44000
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.436142788482545e-06,
      "loss": 1.4712,
      "step": 44500
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.429807314195832e-06,
      "loss": 1.471,
      "step": 45000
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.423471839909119e-06,
      "loss": 1.4712,
      "step": 45500
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.4626506567001343,
      "eval_runtime": 11.7267,
      "eval_samples_per_second": 84.081,
      "step": 45774
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.417136365622406e-06,
      "loss": 1.4703,
      "step": 46000
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.410800891335693e-06,
      "loss": 1.4686,
      "step": 46500
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.4044654170489795e-06,
      "loss": 1.4685,
      "step": 47000
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.398129942762267e-06,
      "loss": 1.468,
      "step": 47500
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.391794468475554e-06,
      "loss": 1.4677,
      "step": 48000
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.385458994188841e-06,
      "loss": 1.468,
      "step": 48500
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.379123519902128e-06,
      "loss": 1.467,
      "step": 49000
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.372788045615415e-06,
      "loss": 1.4672,
      "step": 49500
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.366452571328702e-06,
      "loss": 1.4672,
      "step": 50000
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.360117097041989e-06,
      "loss": 1.4668,
      "step": 50500
    },
    {
      "epoch": 2.23,
      "learning_rate": 2.353781622755276e-06,
      "loss": 1.4665,
      "step": 51000
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.347446148468563e-06,
      "loss": 1.4661,
      "step": 51500
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.34111067418185e-06,
      "loss": 1.4658,
      "step": 52000
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.334775199895137e-06,
      "loss": 1.4658,
      "step": 52500
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.328439725608424e-06,
      "loss": 1.4657,
      "step": 53000
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.322104251321711e-06,
      "loss": 1.4659,
      "step": 53500
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.315768777034998e-06,
      "loss": 1.4653,
      "step": 54000
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.309433302748285e-06,
      "loss": 1.4654,
      "step": 54500
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.303097828461572e-06,
      "loss": 1.4649,
      "step": 55000
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.296762354174859e-06,
      "loss": 1.4646,
      "step": 55500
    },
    {
      "epoch": 2.45,
      "learning_rate": 2.290426879888146e-06,
      "loss": 1.4646,
      "step": 56000
    },
    {
      "epoch": 2.47,
      "learning_rate": 2.284091405601433e-06,
      "loss": 1.4639,
      "step": 56500
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.27775593131472e-06,
      "loss": 1.464,
      "step": 57000
    },
    {
      "epoch": 2.51,
      "learning_rate": 2.271420457028007e-06,
      "loss": 1.4638,
      "step": 57500
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.265084982741294e-06,
      "loss": 1.4634,
      "step": 58000
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.258749508454581e-06,
      "loss": 1.4629,
      "step": 58500
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.252414034167868e-06,
      "loss": 1.4629,
      "step": 59000
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.246078559881155e-06,
      "loss": 1.4627,
      "step": 59500
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.239743085594442e-06,
      "loss": 1.463,
      "step": 60000
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.233407611307729e-06,
      "loss": 1.4625,
      "step": 60500
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.227072137021016e-06,
      "loss": 1.4622,
      "step": 61000
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.220736662734303e-06,
      "loss": 1.4621,
      "step": 61500
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.21440118844759e-06,
      "loss": 1.4619,
      "step": 62000
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.2080657141608774e-06,
      "loss": 1.4618,
      "step": 62500
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.2017302398741644e-06,
      "loss": 1.4614,
      "step": 63000
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.1953947655874514e-06,
      "loss": 1.4615,
      "step": 63500
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.1890592913007384e-06,
      "loss": 1.4609,
      "step": 64000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.1827238170140254e-06,
      "loss": 1.4609,
      "step": 64500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.1763883427273123e-06,
      "loss": 1.461,
      "step": 65000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.1700528684405993e-06,
      "loss": 1.4609,
      "step": 65500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.1637173941538863e-06,
      "loss": 1.4608,
      "step": 66000
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.1573819198671733e-06,
      "loss": 1.4603,
      "step": 66500
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.1510464455804603e-06,
      "loss": 1.4604,
      "step": 67000
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.1447109712937477e-06,
      "loss": 1.4602,
      "step": 67500
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.1383754970070347e-06,
      "loss": 1.4604,
      "step": 68000
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.1320400227203217e-06,
      "loss": 1.4597,
      "step": 68500
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.4553998708724976,
      "eval_runtime": 12.0973,
      "eval_samples_per_second": 81.506,
      "step": 68661
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.1257045484336087e-06,
      "loss": 1.4592,
      "step": 69000
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.1193690741468957e-06,
      "loss": 1.458,
      "step": 69500
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.1130335998601827e-06,
      "loss": 1.4579,
      "step": 70000
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.1066981255734696e-06,
      "loss": 1.4581,
      "step": 70500
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.1003626512867566e-06,
      "loss": 1.4579,
      "step": 71000
    },
    {
      "epoch": 3.12,
      "learning_rate": 2.0940271770000436e-06,
      "loss": 1.4576,
      "step": 71500
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.0876917027133306e-06,
      "loss": 1.4573,
      "step": 72000
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.0813562284266176e-06,
      "loss": 1.4574,
      "step": 72500
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.075020754139905e-06,
      "loss": 1.457,
      "step": 73000
    },
    {
      "epoch": 3.21,
      "learning_rate": 2.068685279853192e-06,
      "loss": 1.4571,
      "step": 73500
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.062349805566479e-06,
      "loss": 1.4573,
      "step": 74000
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.056014331279766e-06,
      "loss": 1.4569,
      "step": 74500
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.049678856993053e-06,
      "loss": 1.4568,
      "step": 75000
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.04334338270634e-06,
      "loss": 1.4569,
      "step": 75500
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.037007908419627e-06,
      "loss": 1.4567,
      "step": 76000
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.030672434132914e-06,
      "loss": 1.4568,
      "step": 76500
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.024336959846201e-06,
      "loss": 1.4564,
      "step": 77000
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.018001485559488e-06,
      "loss": 1.4559,
      "step": 77500
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.011666011272775e-06,
      "loss": 1.4558,
      "step": 78000
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.0053305369860623e-06,
      "loss": 1.4561,
      "step": 78500
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.9989950626993493e-06,
      "loss": 1.4559,
      "step": 79000
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.9926595884126363e-06,
      "loss": 1.4559,
      "step": 79500
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.9863241141259233e-06,
      "loss": 1.4563,
      "step": 80000
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.9799886398392103e-06,
      "loss": 1.4555,
      "step": 80500
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.9736531655524972e-06,
      "loss": 1.4555,
      "step": 81000
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.9673176912657842e-06,
      "loss": 1.4553,
      "step": 81500
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.9609822169790712e-06,
      "loss": 1.4553,
      "step": 82000
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.9546467426923586e-06,
      "loss": 1.4548,
      "step": 82500
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.9483112684056456e-06,
      "loss": 1.4551,
      "step": 83000
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.9419757941189326e-06,
      "loss": 1.4551,
      "step": 83500
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.9356403198322196e-06,
      "loss": 1.4546,
      "step": 84000
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.929304845545506e-06,
      "loss": 1.4548,
      "step": 84500
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.922969371258793e-06,
      "loss": 1.4545,
      "step": 85000
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.91663389697208e-06,
      "loss": 1.4545,
      "step": 85500
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.910298422685367e-06,
      "loss": 1.4546,
      "step": 86000
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.9039629483986545e-06,
      "loss": 1.4544,
      "step": 86500
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.8976274741119413e-06,
      "loss": 1.454,
      "step": 87000
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.8912919998252285e-06,
      "loss": 1.4548,
      "step": 87500
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.8849565255385155e-06,
      "loss": 1.4542,
      "step": 88000
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.8786210512518025e-06,
      "loss": 1.4544,
      "step": 88500
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.8722855769650897e-06,
      "loss": 1.4543,
      "step": 89000
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.8659501026783765e-06,
      "loss": 1.4539,
      "step": 89500
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.8596146283916634e-06,
      "loss": 1.4538,
      "step": 90000
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.8532791541049502e-06,
      "loss": 1.4541,
      "step": 90500
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.8469436798182374e-06,
      "loss": 1.4535,
      "step": 91000
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.8406082055315244e-06,
      "loss": 1.4536,
      "step": 91500
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.4519952535629272,
      "eval_runtime": 11.774,
      "eval_samples_per_second": 83.744,
      "step": 91548
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.834272731244812e-06,
      "loss": 1.4534,
      "step": 92000
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.8279372569580988e-06,
      "loss": 1.4526,
      "step": 92500
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.8216017826713858e-06,
      "loss": 1.4524,
      "step": 93000
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.8152663083846724e-06,
      "loss": 1.4523,
      "step": 93500
    },
    {
      "epoch": 4.11,
      "learning_rate": 1.8089308340979591e-06,
      "loss": 1.4523,
      "step": 94000
    },
    {
      "epoch": 4.13,
      "learning_rate": 1.8025953598112466e-06,
      "loss": 1.452,
      "step": 94500
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.7962598855245333e-06,
      "loss": 1.4523,
      "step": 95000
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.7899244112378203e-06,
      "loss": 1.4519,
      "step": 95500
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.7835889369511077e-06,
      "loss": 1.4521,
      "step": 96000
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.7772534626643947e-06,
      "loss": 1.452,
      "step": 96500
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.7709179883776815e-06,
      "loss": 1.4516,
      "step": 97000
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.764582514090969e-06,
      "loss": 1.4517,
      "step": 97500
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.7582470398042557e-06,
      "loss": 1.4515,
      "step": 98000
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.7519115655175427e-06,
      "loss": 1.4513,
      "step": 98500
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.7455760912308294e-06,
      "loss": 1.451,
      "step": 99000
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.7392406169441166e-06,
      "loss": 1.4513,
      "step": 99500
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.7329051426574036e-06,
      "loss": 1.4509,
      "step": 100000
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.7265696683706906e-06,
      "loss": 1.4513,
      "step": 100500
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.7202341940839778e-06,
      "loss": 1.4514,
      "step": 101000
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.713898719797265e-06,
      "loss": 1.4513,
      "step": 101500
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.7075632455105518e-06,
      "loss": 1.4509,
      "step": 102000
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.701227771223839e-06,
      "loss": 1.4509,
      "step": 102500
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.694892296937126e-06,
      "loss": 1.451,
      "step": 103000
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.688556822650413e-06,
      "loss": 1.4505,
      "step": 103500
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.6822213483637002e-06,
      "loss": 1.4506,
      "step": 104000
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.675885874076987e-06,
      "loss": 1.4507,
      "step": 104500
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.669550399790274e-06,
      "loss": 1.4506,
      "step": 105000
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.6632149255035607e-06,
      "loss": 1.4504,
      "step": 105500
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.6568794512168481e-06,
      "loss": 1.4502,
      "step": 106000
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.650543976930135e-06,
      "loss": 1.4504,
      "step": 106500
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.644208502643422e-06,
      "loss": 1.4506,
      "step": 107000
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.6378730283567093e-06,
      "loss": 1.4502,
      "step": 107500
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.6315375540699963e-06,
      "loss": 1.4504,
      "step": 108000
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.625202079783283e-06,
      "loss": 1.4505,
      "step": 108500
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.6188666054965705e-06,
      "loss": 1.4502,
      "step": 109000
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.6125311312098573e-06,
      "loss": 1.4503,
      "step": 109500
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.6061956569231442e-06,
      "loss": 1.4501,
      "step": 110000
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.599860182636431e-06,
      "loss": 1.45,
      "step": 110500
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5935247083497182e-06,
      "loss": 1.4501,
      "step": 111000
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5871892340630052e-06,
      "loss": 1.4495,
      "step": 111500
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5808537597762928e-06,
      "loss": 1.4495,
      "step": 112000
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.5745182854895796e-06,
      "loss": 1.4495,
      "step": 112500
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.5681828112028666e-06,
      "loss": 1.4498,
      "step": 113000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5618473369161534e-06,
      "loss": 1.4498,
      "step": 113500
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.5555118626294406e-06,
      "loss": 1.45,
      "step": 114000
    },
    {
      "epoch": 5.0,
      "eval_loss": 1.449528455734253,
      "eval_runtime": 13.3969,
      "eval_samples_per_second": 73.599,
      "step": 114435
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.5491763883427276e-06,
      "loss": 1.4501,
      "step": 114500
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.5428409140560146e-06,
      "loss": 1.4485,
      "step": 115000
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.5365054397693011e-06,
      "loss": 1.4482,
      "step": 115500
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.5301699654825883e-06,
      "loss": 1.4484,
      "step": 116000
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.5238344911958755e-06,
      "loss": 1.4482,
      "step": 116500
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.5174990169091623e-06,
      "loss": 1.4487,
      "step": 117000
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.5111635426224497e-06,
      "loss": 1.4482,
      "step": 117500
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.5048280683357365e-06,
      "loss": 1.4484,
      "step": 118000
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.4984925940490235e-06,
      "loss": 1.4483,
      "step": 118500
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.4921571197623105e-06,
      "loss": 1.4482,
      "step": 119000
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.4858216454755974e-06,
      "loss": 1.4482,
      "step": 119500
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.4794861711888844e-06,
      "loss": 1.448,
      "step": 120000
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.4731506969021714e-06,
      "loss": 1.4481,
      "step": 120500
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.4668152226154584e-06,
      "loss": 1.448,
      "step": 121000
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.4604797483287458e-06,
      "loss": 1.4482,
      "step": 121500
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.4541442740420328e-06,
      "loss": 1.4481,
      "step": 122000
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.4478087997553198e-06,
      "loss": 1.4479,
      "step": 122500
    },
    {
      "epoch": 5.37,
      "learning_rate": 1.4414733254686068e-06,
      "loss": 1.4478,
      "step": 123000
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.4351378511818938e-06,
      "loss": 1.448,
      "step": 123500
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.4288023768951808e-06,
      "loss": 1.4478,
      "step": 124000
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.4224669026084677e-06,
      "loss": 1.4478,
      "step": 124500
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.4161314283217547e-06,
      "loss": 1.4478,
      "step": 125000
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.4097959540350417e-06,
      "loss": 1.4477,
      "step": 125500
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.4034604797483287e-06,
      "loss": 1.448,
      "step": 126000
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.3971250054616157e-06,
      "loss": 1.4477,
      "step": 126500
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.3907895311749031e-06,
      "loss": 1.4478,
      "step": 127000
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.38445405688819e-06,
      "loss": 1.4476,
      "step": 127500
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.378118582601477e-06,
      "loss": 1.4478,
      "step": 128000
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.371783108314764e-06,
      "loss": 1.4474,
      "step": 128500
    },
    {
      "epoch": 5.64,
      "learning_rate": 1.365447634028051e-06,
      "loss": 1.4477,
      "step": 129000
    },
    {
      "epoch": 5.66,
      "learning_rate": 1.359112159741338e-06,
      "loss": 1.4473,
      "step": 129500
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.352776685454625e-06,
      "loss": 1.4474,
      "step": 130000
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.346441211167912e-06,
      "loss": 1.4478,
      "step": 130500
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.340105736881199e-06,
      "loss": 1.4475,
      "step": 131000
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.333770262594486e-06,
      "loss": 1.4473,
      "step": 131500
    },
    {
      "epoch": 5.77,
      "learning_rate": 1.327434788307773e-06,
      "loss": 1.4473,
      "step": 132000
    },
    {
      "epoch": 5.79,
      "learning_rate": 1.3210993140210604e-06,
      "loss": 1.4471,
      "step": 132500
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.314763839734347e-06,
      "loss": 1.4471,
      "step": 133000
    },
    {
      "epoch": 5.83,
      "learning_rate": 1.308428365447634e-06,
      "loss": 1.447,
      "step": 133500
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.302092891160921e-06,
      "loss": 1.4471,
      "step": 134000
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.295757416874208e-06,
      "loss": 1.4473,
      "step": 134500
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.289421942587495e-06,
      "loss": 1.447,
      "step": 135000
    },
    {
      "epoch": 5.92,
      "learning_rate": 1.283086468300782e-06,
      "loss": 1.447,
      "step": 135500
    },
    {
      "epoch": 5.94,
      "learning_rate": 1.276750994014069e-06,
      "loss": 1.4468,
      "step": 136000
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.2704155197273563e-06,
      "loss": 1.4469,
      "step": 136500
    },
    {
      "epoch": 5.99,
      "learning_rate": 1.2640800454406433e-06,
      "loss": 1.4469,
      "step": 137000
    },
    {
      "epoch": 6.0,
      "eval_loss": 1.4480507373809814,
      "eval_runtime": 13.3322,
      "eval_samples_per_second": 73.956,
      "step": 137322
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.2577445711539303e-06,
      "loss": 1.4472,
      "step": 137500
    },
    {
      "epoch": 6.03,
      "learning_rate": 1.2514090968672173e-06,
      "loss": 1.4465,
      "step": 138000
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.2450736225805043e-06,
      "loss": 1.446,
      "step": 138500
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.2387381482937913e-06,
      "loss": 1.4461,
      "step": 139000
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.2324026740070782e-06,
      "loss": 1.4462,
      "step": 139500
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.2260671997203652e-06,
      "loss": 1.4461,
      "step": 140000
    },
    {
      "epoch": 6.14,
      "learning_rate": 1.2197317254336522e-06,
      "loss": 1.4462,
      "step": 140500
    },
    {
      "epoch": 6.16,
      "learning_rate": 1.2133962511469392e-06,
      "loss": 1.446,
      "step": 141000
    },
    {
      "epoch": 6.18,
      "learning_rate": 1.2070607768602262e-06,
      "loss": 1.446,
      "step": 141500
    },
    {
      "epoch": 6.2,
      "learning_rate": 1.2007253025735136e-06,
      "loss": 1.4461,
      "step": 142000
    },
    {
      "epoch": 6.23,
      "learning_rate": 1.1943898282868006e-06,
      "loss": 1.446,
      "step": 142500
    },
    {
      "epoch": 6.25,
      "learning_rate": 1.1880543540000876e-06,
      "loss": 1.4458,
      "step": 143000
    },
    {
      "epoch": 6.27,
      "learning_rate": 1.1817188797133746e-06,
      "loss": 1.4458,
      "step": 143500
    },
    {
      "epoch": 6.29,
      "learning_rate": 1.1753834054266616e-06,
      "loss": 1.4454,
      "step": 144000
    },
    {
      "epoch": 6.31,
      "learning_rate": 1.1690479311399485e-06,
      "loss": 1.4459,
      "step": 144500
    },
    {
      "epoch": 6.34,
      "learning_rate": 1.1627124568532355e-06,
      "loss": 1.4453,
      "step": 145000
    },
    {
      "epoch": 6.36,
      "learning_rate": 1.1563769825665225e-06,
      "loss": 1.4458,
      "step": 145500
    },
    {
      "epoch": 6.38,
      "learning_rate": 1.1500415082798095e-06,
      "loss": 1.4458,
      "step": 146000
    },
    {
      "epoch": 6.4,
      "learning_rate": 1.1437060339930965e-06,
      "loss": 1.4459,
      "step": 146500
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.1373705597063835e-06,
      "loss": 1.4456,
      "step": 147000
    },
    {
      "epoch": 6.44,
      "learning_rate": 1.131035085419671e-06,
      "loss": 1.4456,
      "step": 147500
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.1246996111329579e-06,
      "loss": 1.4456,
      "step": 148000
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.1183641368462449e-06,
      "loss": 1.4454,
      "step": 148500
    },
    {
      "epoch": 6.51,
      "learning_rate": 1.1120286625595319e-06,
      "loss": 1.4456,
      "step": 149000
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.1056931882728189e-06,
      "loss": 1.4455,
      "step": 149500
    },
    {
      "epoch": 6.55,
      "learning_rate": 1.0993577139861058e-06,
      "loss": 1.4453,
      "step": 150000
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.0930222396993928e-06,
      "loss": 1.4454,
      "step": 150500
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.0866867654126794e-06,
      "loss": 1.4454,
      "step": 151000
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.0803512911259668e-06,
      "loss": 1.4455,
      "step": 151500
    },
    {
      "epoch": 6.64,
      "learning_rate": 1.0740158168392538e-06,
      "loss": 1.4455,
      "step": 152000
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.0676803425525408e-06,
      "loss": 1.4454,
      "step": 152500
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.0613448682658278e-06,
      "loss": 1.4453,
      "step": 153000
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.0550093939791148e-06,
      "loss": 1.4455,
      "step": 153500
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.0486739196924017e-06,
      "loss": 1.4454,
      "step": 154000
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.0423384454056887e-06,
      "loss": 1.4455,
      "step": 154500
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.0360029711189757e-06,
      "loss": 1.4453,
      "step": 155000
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.0296674968322627e-06,
      "loss": 1.4451,
      "step": 155500
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.02333202254555e-06,
      "loss": 1.4453,
      "step": 156000
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.016996548258837e-06,
      "loss": 1.4449,
      "step": 156500
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.0106610739721239e-06,
      "loss": 1.4453,
      "step": 157000
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.004325599685411e-06,
      "loss": 1.4452,
      "step": 157500
    },
    {
      "epoch": 6.9,
      "learning_rate": 9.97990125398698e-07,
      "loss": 1.4452,
      "step": 158000
    },
    {
      "epoch": 6.93,
      "learning_rate": 9.91654651111985e-07,
      "loss": 1.4453,
      "step": 158500
    },
    {
      "epoch": 6.95,
      "learning_rate": 9.85319176825272e-07,
      "loss": 1.4449,
      "step": 159000
    },
    {
      "epoch": 6.97,
      "learning_rate": 9.78983702538559e-07,
      "loss": 1.4448,
      "step": 159500
    },
    {
      "epoch": 6.99,
      "learning_rate": 9.72648228251846e-07,
      "loss": 1.445,
      "step": 160000
    },
    {
      "epoch": 7.0,
      "eval_loss": 1.4469441175460815,
      "eval_runtime": 13.3822,
      "eval_samples_per_second": 73.68,
      "step": 160209
    },
    {
      "epoch": 7.01,
      "learning_rate": 9.66312753965133e-07,
      "loss": 1.4454,
      "step": 160500
    },
    {
      "epoch": 7.03,
      "learning_rate": 9.5997727967842e-07,
      "loss": 1.4446,
      "step": 161000
    },
    {
      "epoch": 7.06,
      "learning_rate": 9.536418053917071e-07,
      "loss": 1.4443,
      "step": 161500
    },
    {
      "epoch": 7.08,
      "learning_rate": 9.473063311049942e-07,
      "loss": 1.4444,
      "step": 162000
    },
    {
      "epoch": 7.1,
      "learning_rate": 9.409708568182811e-07,
      "loss": 1.4443,
      "step": 162500
    },
    {
      "epoch": 7.12,
      "learning_rate": 9.346353825315684e-07,
      "loss": 1.4444,
      "step": 163000
    },
    {
      "epoch": 7.14,
      "learning_rate": 9.282999082448553e-07,
      "loss": 1.4448,
      "step": 163500
    },
    {
      "epoch": 7.17,
      "learning_rate": 9.219644339581422e-07,
      "loss": 1.4443,
      "step": 164000
    },
    {
      "epoch": 7.19,
      "learning_rate": 9.156289596714293e-07,
      "loss": 1.4442,
      "step": 164500
    },
    {
      "epoch": 7.21,
      "learning_rate": 9.092934853847164e-07,
      "loss": 1.4446,
      "step": 165000
    },
    {
      "epoch": 7.23,
      "learning_rate": 9.029580110980034e-07,
      "loss": 1.4442,
      "step": 165500
    },
    {
      "epoch": 7.25,
      "learning_rate": 8.966225368112904e-07,
      "loss": 1.4443,
      "step": 166000
    },
    {
      "epoch": 7.27,
      "learning_rate": 8.902870625245774e-07,
      "loss": 1.4445,
      "step": 166500
    },
    {
      "epoch": 7.3,
      "learning_rate": 8.839515882378645e-07,
      "loss": 1.4447,
      "step": 167000
    },
    {
      "epoch": 7.32,
      "learning_rate": 8.776161139511516e-07,
      "loss": 1.4444,
      "step": 167500
    },
    {
      "epoch": 7.34,
      "learning_rate": 8.712806396644386e-07,
      "loss": 1.4442,
      "step": 168000
    },
    {
      "epoch": 7.36,
      "learning_rate": 8.649451653777251e-07,
      "loss": 1.4441,
      "step": 168500
    },
    {
      "epoch": 7.38,
      "learning_rate": 8.586096910910122e-07,
      "loss": 1.444,
      "step": 169000
    },
    {
      "epoch": 7.41,
      "learning_rate": 8.522742168042993e-07,
      "loss": 1.4441,
      "step": 169500
    },
    {
      "epoch": 7.43,
      "learning_rate": 8.459387425175863e-07,
      "loss": 1.4443,
      "step": 170000
    },
    {
      "epoch": 7.45,
      "learning_rate": 8.396032682308732e-07,
      "loss": 1.444,
      "step": 170500
    },
    {
      "epoch": 7.47,
      "learning_rate": 8.332677939441605e-07,
      "loss": 1.4441,
      "step": 171000
    },
    {
      "epoch": 7.49,
      "learning_rate": 8.269323196574474e-07,
      "loss": 1.4441,
      "step": 171500
    },
    {
      "epoch": 7.52,
      "learning_rate": 8.205968453707344e-07,
      "loss": 1.4442,
      "step": 172000
    },
    {
      "epoch": 7.54,
      "learning_rate": 8.142613710840215e-07,
      "loss": 1.4441,
      "step": 172500
    },
    {
      "epoch": 7.56,
      "learning_rate": 8.079258967973084e-07,
      "loss": 1.444,
      "step": 173000
    },
    {
      "epoch": 7.58,
      "learning_rate": 8.015904225105957e-07,
      "loss": 1.4441,
      "step": 173500
    },
    {
      "epoch": 7.6,
      "learning_rate": 7.952549482238825e-07,
      "loss": 1.4439,
      "step": 174000
    },
    {
      "epoch": 7.62,
      "learning_rate": 7.889194739371695e-07,
      "loss": 1.4443,
      "step": 174500
    },
    {
      "epoch": 7.65,
      "learning_rate": 7.825839996504566e-07,
      "loss": 1.4443,
      "step": 175000
    },
    {
      "epoch": 7.67,
      "learning_rate": 7.762485253637437e-07,
      "loss": 1.444,
      "step": 175500
    },
    {
      "epoch": 7.69,
      "learning_rate": 7.699130510770306e-07,
      "loss": 1.444,
      "step": 176000
    },
    {
      "epoch": 7.71,
      "learning_rate": 7.635775767903177e-07,
      "loss": 1.4435,
      "step": 176500
    },
    {
      "epoch": 7.73,
      "learning_rate": 7.572421025036047e-07,
      "loss": 1.4439,
      "step": 177000
    },
    {
      "epoch": 7.76,
      "learning_rate": 7.509066282168918e-07,
      "loss": 1.4439,
      "step": 177500
    },
    {
      "epoch": 7.78,
      "learning_rate": 7.445711539301788e-07,
      "loss": 1.4437,
      "step": 178000
    },
    {
      "epoch": 7.8,
      "learning_rate": 7.382356796434658e-07,
      "loss": 1.444,
      "step": 178500
    },
    {
      "epoch": 7.82,
      "learning_rate": 7.319002053567528e-07,
      "loss": 1.4438,
      "step": 179000
    },
    {
      "epoch": 7.84,
      "learning_rate": 7.255647310700398e-07,
      "loss": 1.4436,
      "step": 179500
    },
    {
      "epoch": 7.86,
      "learning_rate": 7.192292567833268e-07,
      "loss": 1.4438,
      "step": 180000
    },
    {
      "epoch": 7.89,
      "learning_rate": 7.128937824966139e-07,
      "loss": 1.4438,
      "step": 180500
    },
    {
      "epoch": 7.91,
      "learning_rate": 7.065583082099009e-07,
      "loss": 1.4439,
      "step": 181000
    },
    {
      "epoch": 7.93,
      "learning_rate": 7.00222833923188e-07,
      "loss": 1.4439,
      "step": 181500
    },
    {
      "epoch": 7.95,
      "learning_rate": 6.93887359636475e-07,
      "loss": 1.4437,
      "step": 182000
    },
    {
      "epoch": 7.97,
      "learning_rate": 6.87551885349762e-07,
      "loss": 1.4438,
      "step": 182500
    },
    {
      "epoch": 8.0,
      "learning_rate": 6.812164110630491e-07,
      "loss": 1.4438,
      "step": 183000
    },
    {
      "epoch": 8.0,
      "eval_loss": 1.446402907371521,
      "eval_runtime": 12.0912,
      "eval_samples_per_second": 81.547,
      "step": 183096
    },
    {
      "epoch": 8.02,
      "learning_rate": 6.748809367763361e-07,
      "loss": 1.4439,
      "step": 183500
    },
    {
      "epoch": 8.04,
      "learning_rate": 6.68545462489623e-07,
      "loss": 1.4432,
      "step": 184000
    },
    {
      "epoch": 8.06,
      "learning_rate": 6.622099882029101e-07,
      "loss": 1.4433,
      "step": 184500
    },
    {
      "epoch": 8.08,
      "learning_rate": 6.558745139161971e-07,
      "loss": 1.4431,
      "step": 185000
    },
    {
      "epoch": 8.11,
      "learning_rate": 6.495390396294842e-07,
      "loss": 1.4431,
      "step": 185500
    },
    {
      "epoch": 8.13,
      "learning_rate": 6.432035653427712e-07,
      "loss": 1.4432,
      "step": 186000
    },
    {
      "epoch": 8.15,
      "learning_rate": 6.368680910560579e-07,
      "loss": 1.4433,
      "step": 186500
    },
    {
      "epoch": 8.17,
      "learning_rate": 6.30532616769345e-07,
      "loss": 1.4432,
      "step": 187000
    },
    {
      "epoch": 8.19,
      "learning_rate": 6.24197142482632e-07,
      "loss": 1.4434,
      "step": 187500
    },
    {
      "epoch": 8.21,
      "learning_rate": 6.17861668195919e-07,
      "loss": 1.4435,
      "step": 188000
    },
    {
      "epoch": 8.24,
      "learning_rate": 6.11526193909206e-07,
      "loss": 1.4435,
      "step": 188500
    },
    {
      "epoch": 8.26,
      "learning_rate": 6.05190719622493e-07,
      "loss": 1.4433,
      "step": 189000
    },
    {
      "epoch": 8.28,
      "learning_rate": 5.9885524533578e-07,
      "loss": 1.443,
      "step": 189500
    },
    {
      "epoch": 8.3,
      "learning_rate": 5.925197710490671e-07,
      "loss": 1.4433,
      "step": 190000
    },
    {
      "epoch": 8.32,
      "learning_rate": 5.861842967623541e-07,
      "loss": 1.4431,
      "step": 190500
    },
    {
      "epoch": 8.35,
      "learning_rate": 5.798488224756412e-07,
      "loss": 1.4432,
      "step": 191000
    },
    {
      "epoch": 8.37,
      "learning_rate": 5.735133481889282e-07,
      "loss": 1.443,
      "step": 191500
    },
    {
      "epoch": 8.39,
      "learning_rate": 5.671778739022152e-07,
      "loss": 1.443,
      "step": 192000
    },
    {
      "epoch": 8.41,
      "learning_rate": 5.608423996155022e-07,
      "loss": 1.4434,
      "step": 192500
    },
    {
      "epoch": 8.43,
      "learning_rate": 5.545069253287893e-07,
      "loss": 1.4432,
      "step": 193000
    },
    {
      "epoch": 8.45,
      "learning_rate": 5.481714510420763e-07,
      "loss": 1.4429,
      "step": 193500
    },
    {
      "epoch": 8.48,
      "learning_rate": 5.418359767553633e-07,
      "loss": 1.4432,
      "step": 194000
    },
    {
      "epoch": 8.5,
      "learning_rate": 5.355005024686503e-07,
      "loss": 1.4432,
      "step": 194500
    },
    {
      "epoch": 8.52,
      "learning_rate": 5.291650281819373e-07,
      "loss": 1.4429,
      "step": 195000
    },
    {
      "epoch": 8.54,
      "learning_rate": 5.228295538952244e-07,
      "loss": 1.443,
      "step": 195500
    },
    {
      "epoch": 8.56,
      "learning_rate": 5.164940796085115e-07,
      "loss": 1.4433,
      "step": 196000
    },
    {
      "epoch": 8.59,
      "learning_rate": 5.101586053217985e-07,
      "loss": 1.4428,
      "step": 196500
    },
    {
      "epoch": 8.61,
      "learning_rate": 5.038231310350855e-07,
      "loss": 1.4429,
      "step": 197000
    },
    {
      "epoch": 8.63,
      "learning_rate": 4.974876567483725e-07,
      "loss": 1.4431,
      "step": 197500
    },
    {
      "epoch": 8.65,
      "learning_rate": 4.911521824616596e-07,
      "loss": 1.4431,
      "step": 198000
    },
    {
      "epoch": 8.67,
      "learning_rate": 4.848167081749465e-07,
      "loss": 1.443,
      "step": 198500
    },
    {
      "epoch": 8.69,
      "learning_rate": 4.784812338882336e-07,
      "loss": 1.4433,
      "step": 199000
    },
    {
      "epoch": 8.72,
      "learning_rate": 4.7214575960152063e-07,
      "loss": 1.4431,
      "step": 199500
    },
    {
      "epoch": 8.74,
      "learning_rate": 4.6581028531480767e-07,
      "loss": 1.4431,
      "step": 200000
    },
    {
      "epoch": 8.76,
      "learning_rate": 4.5947481102809466e-07,
      "loss": 1.4433,
      "step": 200500
    },
    {
      "epoch": 8.78,
      "learning_rate": 4.531393367413817e-07,
      "loss": 1.4429,
      "step": 201000
    },
    {
      "epoch": 8.8,
      "learning_rate": 4.468038624546688e-07,
      "loss": 1.4431,
      "step": 201500
    },
    {
      "epoch": 8.83,
      "learning_rate": 4.404683881679558e-07,
      "loss": 1.4431,
      "step": 202000
    },
    {
      "epoch": 8.85,
      "learning_rate": 4.341329138812427e-07,
      "loss": 1.4433,
      "step": 202500
    },
    {
      "epoch": 8.87,
      "learning_rate": 4.277974395945298e-07,
      "loss": 1.4428,
      "step": 203000
    },
    {
      "epoch": 8.89,
      "learning_rate": 4.2146196530781685e-07,
      "loss": 1.4429,
      "step": 203500
    },
    {
      "epoch": 8.91,
      "learning_rate": 4.151264910211035e-07,
      "loss": 1.4428,
      "step": 204000
    },
    {
      "epoch": 8.94,
      "learning_rate": 4.0879101673439056e-07,
      "loss": 1.4427,
      "step": 204500
    },
    {
      "epoch": 8.96,
      "learning_rate": 4.024555424476776e-07,
      "loss": 1.4428,
      "step": 205000
    },
    {
      "epoch": 8.98,
      "learning_rate": 3.961200681609647e-07,
      "loss": 1.4428,
      "step": 205500
    },
    {
      "epoch": 9.0,
      "eval_loss": 1.4459742307662964,
      "eval_runtime": 11.8598,
      "eval_samples_per_second": 83.138,
      "step": 205983
    },
    {
      "epoch": 9.0,
      "learning_rate": 3.897845938742517e-07,
      "loss": 1.4438,
      "step": 206000
    },
    {
      "epoch": 9.02,
      "learning_rate": 3.834491195875386e-07,
      "loss": 1.4426,
      "step": 206500
    },
    {
      "epoch": 9.04,
      "learning_rate": 3.771136453008257e-07,
      "loss": 1.4426,
      "step": 207000
    },
    {
      "epoch": 9.07,
      "learning_rate": 3.7077817101411276e-07,
      "loss": 1.4426,
      "step": 207500
    },
    {
      "epoch": 9.09,
      "learning_rate": 3.6444269672739975e-07,
      "loss": 1.4427,
      "step": 208000
    },
    {
      "epoch": 9.11,
      "learning_rate": 3.581072224406868e-07,
      "loss": 1.4428,
      "step": 208500
    },
    {
      "epoch": 9.13,
      "learning_rate": 3.5177174815397383e-07,
      "loss": 1.4425,
      "step": 209000
    },
    {
      "epoch": 9.15,
      "learning_rate": 3.4543627386726087e-07,
      "loss": 1.4428,
      "step": 209500
    },
    {
      "epoch": 9.18,
      "learning_rate": 3.3910079958054786e-07,
      "loss": 1.4425,
      "step": 210000
    },
    {
      "epoch": 9.2,
      "learning_rate": 3.327653252938349e-07,
      "loss": 1.4427,
      "step": 210500
    },
    {
      "epoch": 9.22,
      "learning_rate": 3.2642985100712194e-07,
      "loss": 1.4423,
      "step": 211000
    },
    {
      "epoch": 9.24,
      "learning_rate": 3.20094376720409e-07,
      "loss": 1.4427,
      "step": 211500
    },
    {
      "epoch": 9.26,
      "learning_rate": 3.1375890243369597e-07,
      "loss": 1.4426,
      "step": 212000
    },
    {
      "epoch": 9.28,
      "learning_rate": 3.07423428146983e-07,
      "loss": 1.4427,
      "step": 212500
    },
    {
      "epoch": 9.31,
      "learning_rate": 3.0108795386027005e-07,
      "loss": 1.4427,
      "step": 213000
    },
    {
      "epoch": 9.33,
      "learning_rate": 2.9475247957355704e-07,
      "loss": 1.4425,
      "step": 213500
    },
    {
      "epoch": 9.35,
      "learning_rate": 2.884170052868441e-07,
      "loss": 1.4423,
      "step": 214000
    },
    {
      "epoch": 9.37,
      "learning_rate": 2.820815310001311e-07,
      "loss": 1.4427,
      "step": 214500
    },
    {
      "epoch": 9.39,
      "learning_rate": 2.757460567134181e-07,
      "loss": 1.4426,
      "step": 215000
    },
    {
      "epoch": 9.42,
      "learning_rate": 2.694105824267052e-07,
      "loss": 1.4425,
      "step": 215500
    },
    {
      "epoch": 9.44,
      "learning_rate": 2.630751081399922e-07,
      "loss": 1.4426,
      "step": 216000
    },
    {
      "epoch": 9.46,
      "learning_rate": 2.5673963385327923e-07,
      "loss": 1.4427,
      "step": 216500
    },
    {
      "epoch": 9.48,
      "learning_rate": 2.504041595665663e-07,
      "loss": 1.4426,
      "step": 217000
    },
    {
      "epoch": 9.5,
      "learning_rate": 2.4406868527985326e-07,
      "loss": 1.4424,
      "step": 217500
    },
    {
      "epoch": 9.53,
      "learning_rate": 2.377332109931403e-07,
      "loss": 1.4426,
      "step": 218000
    },
    {
      "epoch": 9.55,
      "learning_rate": 2.3139773670642737e-07,
      "loss": 1.4425,
      "step": 218500
    },
    {
      "epoch": 9.57,
      "learning_rate": 2.2506226241971439e-07,
      "loss": 1.4424,
      "step": 219000
    },
    {
      "epoch": 9.59,
      "learning_rate": 2.187267881330014e-07,
      "loss": 1.4423,
      "step": 219500
    },
    {
      "epoch": 9.61,
      "learning_rate": 2.1239131384628841e-07,
      "loss": 1.4423,
      "step": 220000
    },
    {
      "epoch": 9.63,
      "learning_rate": 2.0605583955957543e-07,
      "loss": 1.4428,
      "step": 220500
    },
    {
      "epoch": 9.66,
      "learning_rate": 1.997203652728625e-07,
      "loss": 1.4423,
      "step": 221000
    },
    {
      "epoch": 9.68,
      "learning_rate": 1.933848909861492e-07,
      "loss": 1.4425,
      "step": 221500
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.870494166994362e-07,
      "loss": 1.4425,
      "step": 222000
    },
    {
      "epoch": 9.72,
      "learning_rate": 1.8071394241272325e-07,
      "loss": 1.4424,
      "step": 222500
    },
    {
      "epoch": 9.74,
      "learning_rate": 1.7437846812601026e-07,
      "loss": 1.4425,
      "step": 223000
    },
    {
      "epoch": 9.77,
      "learning_rate": 1.6804299383929728e-07,
      "loss": 1.4424,
      "step": 223500
    },
    {
      "epoch": 9.79,
      "learning_rate": 1.6170751955258432e-07,
      "loss": 1.4424,
      "step": 224000
    },
    {
      "epoch": 9.81,
      "learning_rate": 1.5537204526587136e-07,
      "loss": 1.4424,
      "step": 224500
    },
    {
      "epoch": 9.83,
      "learning_rate": 1.4903657097915838e-07,
      "loss": 1.4423,
      "step": 225000
    },
    {
      "epoch": 9.85,
      "learning_rate": 1.427010966924454e-07,
      "loss": 1.4426,
      "step": 225500
    },
    {
      "epoch": 9.87,
      "learning_rate": 1.3636562240573243e-07,
      "loss": 1.4427,
      "step": 226000
    },
    {
      "epoch": 9.9,
      "learning_rate": 1.3003014811901947e-07,
      "loss": 1.4423,
      "step": 226500
    },
    {
      "epoch": 9.92,
      "learning_rate": 1.236946738323065e-07,
      "loss": 1.4423,
      "step": 227000
    },
    {
      "epoch": 9.94,
      "learning_rate": 1.1735919954559353e-07,
      "loss": 1.4421,
      "step": 227500
    },
    {
      "epoch": 9.96,
      "learning_rate": 1.1102372525888054e-07,
      "loss": 1.4423,
      "step": 228000
    },
    {
      "epoch": 9.98,
      "learning_rate": 1.0468825097216756e-07,
      "loss": 1.4427,
      "step": 228500
    },
    {
      "epoch": 10.0,
      "eval_loss": 1.4457148313522339,
      "eval_runtime": 13.0706,
      "eval_samples_per_second": 75.436,
      "step": 228870
    }
  ],
  "max_steps": 228870,
  "num_train_epochs": 10,
  "total_flos": 1.288906420929577e+19,
  "trial_name": null,
  "trial_params": null
}