{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0219132811959306,
  "eval_steps": 100000000,
  "global_step": 126000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 10.7586,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 5e-05,
      "loss": 6.5292,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99998214600225e-05,
      "loss": 4.7262,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999285842640105e-05,
      "loss": 4.1997,
      "step": 1500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999839315550315e-05,
      "loss": 3.8892,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9997143411362066e-05,
      "loss": 3.6618,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9995536628067196e-05,
      "loss": 3.5132,
      "step": 3000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9993572828568536e-05,
      "loss": 3.4092,
      "step": 3500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999125204091544e-05,
      "loss": 3.3295,
      "step": 4000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998857429825615e-05,
      "loss": 3.2596,
      "step": 4500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9985539638837424e-05,
      "loss": 3.2068,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998214810600389e-05,
      "loss": 3.163,
      "step": 5500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997839974819749e-05,
      "loss": 3.1256,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.997429461895675e-05,
      "loss": 3.0844,
      "step": 6500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996983277691606e-05,
      "loss": 3.0541,
      "step": 7000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996501428580478e-05,
      "loss": 3.0293,
      "step": 7500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.995983921444638e-05,
      "loss": 2.999,
      "step": 8000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9954307636757434e-05,
      "loss": 2.9744,
      "step": 8500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.994841963174656e-05,
      "loss": 2.9525,
      "step": 9000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9942175283513294e-05,
      "loss": 2.9321,
      "step": 9500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993557468124691e-05,
      "loss": 2.9237,
      "step": 10000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9928617919225106e-05,
      "loss": 2.9021,
      "step": 10500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.99213050968127e-05,
      "loss": 2.8766,
      "step": 11000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991363631846019e-05,
      "loss": 2.8691,
      "step": 11500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.990561169370224e-05,
      "loss": 2.8531,
      "step": 12000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.989723133715618e-05,
      "loss": 2.8366,
      "step": 12500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.988849536852028e-05,
      "loss": 2.8265,
      "step": 13000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.987940391257212e-05,
      "loss": 2.8099,
      "step": 13500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.986995709916677e-05,
      "loss": 2.8009,
      "step": 14000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9860155063234944e-05,
      "loss": 2.7881,
      "step": 14500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9849997944781055e-05,
      "loss": 2.7794,
      "step": 15000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9839485888881235e-05,
      "loss": 2.7674,
      "step": 15500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.982861904568127e-05,
      "loss": 2.7559,
      "step": 16000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.981739757039443e-05,
      "loss": 2.7455,
      "step": 16500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9805821623299285e-05,
      "loss": 2.7351,
      "step": 17000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.979389136973737e-05,
      "loss": 2.7317,
      "step": 17500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.978160698011085e-05,
      "loss": 2.7192,
      "step": 18000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.976896862988012e-05,
      "loss": 2.709,
      "step": 18500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.975597649956122e-05,
      "loss": 2.7043,
      "step": 19000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.974263077472334e-05,
      "loss": 2.6954,
      "step": 19500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.97289316459861e-05,
      "loss": 2.6905,
      "step": 20000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9714879309016874e-05,
      "loss": 2.6782,
      "step": 20500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.970047396452798e-05,
      "loss": 2.6751,
      "step": 21000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.968571581827381e-05,
      "loss": 2.6659,
      "step": 21500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9670605081047886e-05,
      "loss": 2.6594,
      "step": 22000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9655141968679866e-05,
      "loss": 2.6553,
      "step": 22500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.963932670203245e-05,
      "loss": 2.6499,
      "step": 23000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.962315950699822e-05,
      "loss": 2.6391,
      "step": 23500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.960664061449643e-05,
      "loss": 2.6313,
      "step": 24000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.958977026046971e-05,
      "loss": 2.6318,
      "step": 24500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.957254868588065e-05,
      "loss": 2.6275,
      "step": 25000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.955497613670842e-05,
      "loss": 2.6192,
      "step": 25500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.953705286394522e-05,
      "loss": 2.6153,
      "step": 26000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9518779123592705e-05,
      "loss": 2.6102,
      "step": 26500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9500155176658345e-05,
      "loss": 2.601,
      "step": 27000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9481181289151655e-05,
      "loss": 2.5991,
      "step": 27500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9461857732080436e-05,
      "loss": 2.5946,
      "step": 28000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.9442184781446876e-05,
      "loss": 2.5911,
      "step": 28500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.9422162718243626e-05,
      "loss": 2.5827,
      "step": 29000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.94017918284498e-05,
      "loss": 2.5785,
      "step": 29500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.938107240302683e-05,
      "loss": 2.5757,
      "step": 30000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.93600047379144e-05,
      "loss": 2.5705,
      "step": 30500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9338589134026124e-05,
      "loss": 2.5703,
      "step": 31000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.931682589724534e-05,
      "loss": 2.5593,
      "step": 31500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.929471533842065e-05,
      "loss": 2.5647,
      "step": 32000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9272257773361574e-05,
      "loss": 2.5564,
      "step": 32500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.924945352283394e-05,
      "loss": 2.5492,
      "step": 33000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.922630291255539e-05,
      "loss": 2.5509,
      "step": 33500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.920280627319067e-05,
      "loss": 2.5466,
      "step": 34000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.917896394034696e-05,
      "loss": 2.5412,
      "step": 34500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.915477625456899e-05,
      "loss": 2.5383,
      "step": 35000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.91302435613343e-05,
      "loss": 2.5364,
      "step": 35500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.91053662110482e-05,
      "loss": 2.5292,
      "step": 36000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.908014455903881e-05,
      "loss": 2.5304,
      "step": 36500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.905457896555199e-05,
      "loss": 2.5257,
      "step": 37000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.902866979574618e-05,
      "loss": 2.5189,
      "step": 37500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9002417419687183e-05,
      "loss": 2.522,
      "step": 38000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.897582221234289e-05,
      "loss": 2.5141,
      "step": 38500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8948884553577926e-05,
      "loss": 2.517,
      "step": 39000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8921604828148206e-05,
      "loss": 2.5077,
      "step": 39500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8893983425695445e-05,
      "loss": 2.5068,
      "step": 40000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8866020740741626e-05,
      "loss": 2.5048,
      "step": 40500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.883771717268331e-05,
      "loss": 2.4972,
      "step": 41000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8809073125785964e-05,
      "loss": 2.4999,
      "step": 41500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8780089009178206e-05,
      "loss": 2.4941,
      "step": 42000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8750765236845894e-05,
      "loss": 2.4891,
      "step": 42500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8721102227626304e-05,
      "loss": 2.4876,
      "step": 43000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.869110040520206e-05,
      "loss": 2.4859,
      "step": 43500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8660760198095145e-05,
      "loss": 2.4833,
      "step": 44000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8630082039660744e-05,
      "loss": 2.4825,
      "step": 44500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.859906636808108e-05,
      "loss": 2.4801,
      "step": 45000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8567713626359134e-05,
      "loss": 2.4763,
      "step": 45500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.853602426231234e-05,
      "loss": 2.4742,
      "step": 46000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.850399872856615e-05,
      "loss": 2.4702,
      "step": 46500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.847163748254762e-05,
      "loss": 2.4679,
      "step": 47000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.843894098647884e-05,
      "loss": 2.4676,
      "step": 47500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.840590970737035e-05,
      "loss": 2.4679,
      "step": 48000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8372544117014436e-05,
      "loss": 2.4644,
      "step": 48500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8338844691978456e-05,
      "loss": 2.4622,
      "step": 49000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.830481191359797e-05,
      "loss": 2.4605,
      "step": 49500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.82704462679699e-05,
      "loss": 2.4552,
      "step": 50000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.823574824594558e-05,
      "loss": 2.4501,
      "step": 50500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.820071834312372e-05,
      "loss": 2.4563,
      "step": 51000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.816535705984337e-05,
      "loss": 2.4467,
      "step": 51500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.812966490117676e-05,
      "loss": 2.4434,
      "step": 52000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8093642376922057e-05,
      "loss": 2.4422,
      "step": 52500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8057290001596106e-05,
      "loss": 2.4462,
      "step": 53000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.802060829442711e-05,
      "loss": 2.4424,
      "step": 53500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.798359777934714e-05,
      "loss": 2.4373,
      "step": 54000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.7946258984984735e-05,
      "loss": 2.4381,
      "step": 54500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7908592444657287e-05,
      "loss": 2.4334,
      "step": 55000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.787059869636346e-05,
      "loss": 2.4328,
      "step": 55500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.783227828277549e-05,
      "loss": 2.4284,
      "step": 56000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.779363175123144e-05,
      "loss": 2.4278,
      "step": 56500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.7754659653727384e-05,
      "loss": 2.4231,
      "step": 57000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.771536254690951e-05,
      "loss": 2.4212,
      "step": 57500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.767574099206619e-05,
      "loss": 2.4247,
      "step": 58000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.763579555511993e-05,
      "loss": 2.4209,
      "step": 58500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.759552680661933e-05,
      "loss": 2.4224,
      "step": 59000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.755493532173092e-05,
      "loss": 2.4204,
      "step": 59500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.75140216802309e-05,
      "loss": 2.4201,
      "step": 60000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7472786466496944e-05,
      "loss": 2.4127,
      "step": 60500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7431230269499764e-05,
      "loss": 2.4146,
      "step": 61000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7389353682794765e-05,
      "loss": 2.4074,
      "step": 61500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7347157304513544e-05,
      "loss": 2.411,
      "step": 62000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7304641737355326e-05,
      "loss": 2.411,
      "step": 62500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7261807588578375e-05,
      "loss": 2.4068,
      "step": 63000
    },
    {
      "epoch": 0.45778831291669553,
      "grad_norm": 0.4898933470249176,
      "learning_rate": 4.721865546999135e-05,
      "loss": 2.3865,
      "step": 63500
    },
    {
      "epoch": 0.4613929453018663,
      "grad_norm": 0.5031083226203918,
      "learning_rate": 4.7175185997944485e-05,
      "loss": 2.3853,
      "step": 64000
    },
    {
      "epoch": 0.46499757768703714,
      "grad_norm": 0.5099691152572632,
      "learning_rate": 4.713139979332089e-05,
      "loss": 2.3843,
      "step": 64500
    },
    {
      "epoch": 0.468602210072208,
      "grad_norm": 0.528971791267395,
      "learning_rate": 4.708729748152759e-05,
      "loss": 2.3867,
      "step": 65000
    },
    {
      "epoch": 0.4722068424573788,
      "grad_norm": 0.5447410941123962,
      "learning_rate": 4.7042879692486655e-05,
      "loss": 2.3871,
      "step": 65500
    },
    {
      "epoch": 0.47581147484254965,
      "grad_norm": 0.5187677145004272,
      "learning_rate": 4.699814706062616e-05,
      "loss": 2.3821,
      "step": 66000
    },
    {
      "epoch": 0.4794161072277205,
      "grad_norm": 0.5079197883605957,
      "learning_rate": 4.6953100224871163e-05,
      "loss": 2.3846,
      "step": 66500
    },
    {
      "epoch": 0.4830207396128913,
      "grad_norm": 0.5204797387123108,
      "learning_rate": 4.6907739828634536e-05,
      "loss": 2.3867,
      "step": 67000
    },
    {
      "epoch": 0.48662537199806216,
      "grad_norm": 0.51589435338974,
      "learning_rate": 4.6862066519807816e-05,
      "loss": 2.3792,
      "step": 67500
    },
    {
      "epoch": 0.490230004383233,
      "grad_norm": 0.5342502593994141,
      "learning_rate": 4.681608095075192e-05,
      "loss": 2.3767,
      "step": 68000
    },
    {
      "epoch": 0.49383463676840383,
      "grad_norm": 0.5265562534332275,
      "learning_rate": 4.676978377828785e-05,
      "loss": 2.3839,
      "step": 68500
    },
    {
      "epoch": 0.4974392691535746,
      "grad_norm": 0.5328825116157532,
      "learning_rate": 4.67231756636873e-05,
      "loss": 2.3813,
      "step": 69000
    },
    {
      "epoch": 0.5010439015387455,
      "grad_norm": 0.4992935061454773,
      "learning_rate": 4.6676257272663194e-05,
      "loss": 2.377,
      "step": 69500
    },
    {
      "epoch": 0.5046485339239163,
      "grad_norm": 0.5137242674827576,
      "learning_rate": 4.662902927536022e-05,
      "loss": 2.3798,
      "step": 70000
    },
    {
      "epoch": 0.5082531663090871,
      "grad_norm": 0.5188072323799133,
      "learning_rate": 4.658149234634523e-05,
      "loss": 2.3783,
      "step": 70500
    },
    {
      "epoch": 0.511857798694258,
      "grad_norm": 0.5286343693733215,
      "learning_rate": 4.65336471645976e-05,
      "loss": 2.3719,
      "step": 71000
    },
    {
      "epoch": 0.5154624310794288,
      "grad_norm": 0.5098246335983276,
      "learning_rate": 4.648549441349953e-05,
      "loss": 2.3712,
      "step": 71500
    },
    {
      "epoch": 0.5190670634645996,
      "grad_norm": 0.528311014175415,
      "learning_rate": 4.643703478082632e-05,
      "loss": 2.37,
      "step": 72000
    },
    {
      "epoch": 0.5226716958497705,
      "grad_norm": 0.5106325745582581,
      "learning_rate": 4.638826895873651e-05,
      "loss": 2.3717,
      "step": 72500
    },
    {
      "epoch": 0.5262763282349413,
      "grad_norm": 0.5130262970924377,
      "learning_rate": 4.6339197643762e-05,
      "loss": 2.3669,
      "step": 73000
    },
    {
      "epoch": 0.5298809606201121,
      "grad_norm": 0.5350182056427002,
      "learning_rate": 4.628982153679811e-05,
      "loss": 2.3675,
      "step": 73500
    },
    {
      "epoch": 0.533485593005283,
      "grad_norm": 0.5258482098579407,
      "learning_rate": 4.6240141343093545e-05,
      "loss": 2.3673,
      "step": 74000
    },
    {
      "epoch": 0.5370902253904538,
      "grad_norm": 0.5219452977180481,
      "learning_rate": 4.6190157772240385e-05,
      "loss": 2.3644,
      "step": 74500
    },
    {
      "epoch": 0.5406948577756246,
      "grad_norm": 0.531566858291626,
      "learning_rate": 4.6139871538163856e-05,
      "loss": 2.366,
      "step": 75000
    },
    {
      "epoch": 0.5442994901607955,
      "grad_norm": 0.5164352655410767,
      "learning_rate": 4.608928335911222e-05,
      "loss": 2.3651,
      "step": 75500
    },
    {
      "epoch": 0.5479041225459663,
      "grad_norm": 0.558618426322937,
      "learning_rate": 4.603839395764646e-05,
      "loss": 2.3618,
      "step": 76000
    },
    {
      "epoch": 0.5515087549311372,
      "grad_norm": 0.5367295145988464,
      "learning_rate": 4.598720406062999e-05,
      "loss": 2.3581,
      "step": 76500
    },
    {
      "epoch": 0.5551133873163079,
      "grad_norm": 0.5240716934204102,
      "learning_rate": 4.593571439921825e-05,
      "loss": 2.3609,
      "step": 77000
    },
    {
      "epoch": 0.5587180197014787,
      "grad_norm": 0.5218440890312195,
      "learning_rate": 4.5883925708848275e-05,
      "loss": 2.3583,
      "step": 77500
    },
    {
      "epoch": 0.5623226520866496,
      "grad_norm": 0.5101985335350037,
      "learning_rate": 4.58318387292282e-05,
      "loss": 2.3587,
      "step": 78000
    },
    {
      "epoch": 0.636668195030798,
      "grad_norm": 0.527410626411438,
      "learning_rate": 4.469806659761896e-05,
      "loss": 2.3498,
      "step": 78500
    },
    {
      "epoch": 0.6407234064641152,
      "grad_norm": 0.4921613335609436,
      "learning_rate": 4.4632426356679503e-05,
      "loss": 2.3469,
      "step": 79000
    },
    {
      "epoch": 0.6447786178974323,
      "grad_norm": 0.4988476037979126,
      "learning_rate": 4.456643111161236e-05,
      "loss": 2.348,
      "step": 79500
    },
    {
      "epoch": 0.6488338293307495,
      "grad_norm": 0.48752740025520325,
      "learning_rate": 4.450008205577918e-05,
      "loss": 2.3469,
      "step": 80000
    },
    {
      "epoch": 0.6528890407640667,
      "grad_norm": 0.5155342817306519,
      "learning_rate": 4.443338038893938e-05,
      "loss": 2.346,
      "step": 80500
    },
    {
      "epoch": 0.656944252197384,
      "grad_norm": 0.4966714680194855,
      "learning_rate": 4.4366327317228506e-05,
      "loss": 2.346,
      "step": 81000
    },
    {
      "epoch": 0.6609994636307011,
      "grad_norm": 0.49693840742111206,
      "learning_rate": 4.429892405313638e-05,
      "loss": 2.3406,
      "step": 81500
    },
    {
      "epoch": 0.6650546750640183,
      "grad_norm": 0.4809018671512604,
      "learning_rate": 4.42311718154852e-05,
      "loss": 2.3429,
      "step": 82000
    },
    {
      "epoch": 0.6691098864973355,
      "grad_norm": 0.5525850653648376,
      "learning_rate": 4.416307182940748e-05,
      "loss": 2.3424,
      "step": 82500
    },
    {
      "epoch": 0.6731650979306526,
      "grad_norm": 0.5057451128959656,
      "learning_rate": 4.409462532632391e-05,
      "loss": 2.3388,
      "step": 83000
    },
    {
      "epoch": 0.6772203093639698,
      "grad_norm": 0.5083210468292236,
      "learning_rate": 4.402583354392112e-05,
      "loss": 2.3395,
      "step": 83500
    },
    {
      "epoch": 0.681275520797287,
      "grad_norm": 0.49632710218429565,
      "learning_rate": 4.3956697726129223e-05,
      "loss": 2.3335,
      "step": 84000
    },
    {
      "epoch": 0.6853307322306041,
      "grad_norm": 0.5196436047554016,
      "learning_rate": 4.388721912309939e-05,
      "loss": 2.3361,
      "step": 84500
    },
    {
      "epoch": 0.6893859436639214,
      "grad_norm": 0.49743419885635376,
      "learning_rate": 4.381739899118121e-05,
      "loss": 2.3352,
      "step": 85000
    },
    {
      "epoch": 0.6934411550972386,
      "grad_norm": 0.4990304112434387,
      "learning_rate": 4.3747238592899994e-05,
      "loss": 2.3314,
      "step": 85500
    },
    {
      "epoch": 0.6974963665305558,
      "grad_norm": 0.4926765263080597,
      "learning_rate": 4.367673919693392e-05,
      "loss": 2.3346,
      "step": 86000
    },
    {
      "epoch": 0.7015515779638729,
      "grad_norm": 0.520667314529419,
      "learning_rate": 4.3605902078091106e-05,
      "loss": 2.3334,
      "step": 86500
    },
    {
      "epoch": 0.7056067893971901,
      "grad_norm": 0.5267304182052612,
      "learning_rate": 4.353472851728656e-05,
      "loss": 2.3295,
      "step": 87000
    },
    {
      "epoch": 0.7096620008305073,
      "grad_norm": 0.5232897996902466,
      "learning_rate": 4.346321980151903e-05,
      "loss": 2.3295,
      "step": 87500
    },
    {
      "epoch": 0.7137172122638245,
      "grad_norm": 0.5169503092765808,
      "learning_rate": 4.3391377223847684e-05,
      "loss": 2.325,
      "step": 88000
    },
    {
      "epoch": 0.7177724236971417,
      "grad_norm": 0.5122842192649841,
      "learning_rate": 4.331920208336879e-05,
      "loss": 2.3294,
      "step": 88500
    },
    {
      "epoch": 0.7218276351304589,
      "grad_norm": 0.5060274600982666,
      "learning_rate": 4.3246695685192185e-05,
      "loss": 2.3226,
      "step": 89000
    },
    {
      "epoch": 0.725882846563776,
      "grad_norm": 0.5004491806030273,
      "learning_rate": 4.317385934041769e-05,
      "loss": 2.3262,
      "step": 89500
    },
    {
      "epoch": 0.7299380579970932,
      "grad_norm": 0.5048273801803589,
      "learning_rate": 4.310069436611139e-05,
      "loss": 2.3279,
      "step": 90000
    },
    {
      "epoch": 0.7339932694304104,
      "grad_norm": 0.5081968307495117,
      "learning_rate": 4.3027202085281844e-05,
      "loss": 2.3263,
      "step": 90500
    },
    {
      "epoch": 0.7380484808637275,
      "grad_norm": 0.5030719041824341,
      "learning_rate": 4.295338382685612e-05,
      "loss": 2.3238,
      "step": 91000
    },
    {
      "epoch": 0.7421036922970448,
      "grad_norm": 0.5075569748878479,
      "learning_rate": 4.28792409256558e-05,
      "loss": 2.324,
      "step": 91500
    },
    {
      "epoch": 0.746158903730362,
      "grad_norm": 0.5366798639297485,
      "learning_rate": 4.2804774722372855e-05,
      "loss": 2.3163,
      "step": 92000
    },
    {
      "epoch": 0.7502141151636792,
      "grad_norm": 0.49396347999572754,
      "learning_rate": 4.272998656354533e-05,
      "loss": 2.3187,
      "step": 92500
    },
    {
      "epoch": 0.7542693265969963,
      "grad_norm": 0.5059054493904114,
      "learning_rate": 4.265487780153309e-05,
      "loss": 2.3147,
      "step": 93000
    },
    {
      "epoch": 0.7583245380303135,
      "grad_norm": 0.5356733202934265,
      "learning_rate": 4.257944979449329e-05,
      "loss": 2.3163,
      "step": 93500
    },
    {
      "epoch": 0.7623797494636307,
      "grad_norm": 0.5293506979942322,
      "learning_rate": 4.250370390635586e-05,
      "loss": 2.3121,
      "step": 94000
    },
    {
      "epoch": 0.7664349608969478,
      "grad_norm": 0.5404239296913147,
      "learning_rate": 4.242764150679883e-05,
      "loss": 2.3174,
      "step": 94500
    },
    {
      "epoch": 0.7704901723302651,
      "grad_norm": 0.5234089493751526,
      "learning_rate": 4.235126397122356e-05,
      "loss": 2.3179,
      "step": 95000
    },
    {
      "epoch": 0.7745453837635823,
      "grad_norm": 0.5130496025085449,
      "learning_rate": 4.227457268072984e-05,
      "loss": 2.3156,
      "step": 95500
    },
    {
      "epoch": 0.7786005951968994,
      "grad_norm": 0.50419020652771,
      "learning_rate": 4.2197569022090984e-05,
      "loss": 2.3128,
      "step": 96000
    },
    {
      "epoch": 0.7826558066302166,
      "grad_norm": 0.515084981918335,
      "learning_rate": 4.21202543877287e-05,
      "loss": 2.3113,
      "step": 96500
    },
    {
      "epoch": 0.7867110180635338,
      "grad_norm": 0.516667366027832,
      "learning_rate": 4.204263017568793e-05,
      "loss": 2.3111,
      "step": 97000
    },
    {
      "epoch": 0.790766229496851,
      "grad_norm": 0.5135312676429749,
      "learning_rate": 4.196469778961157e-05,
      "loss": 2.3105,
      "step": 97500
    },
    {
      "epoch": 0.7948214409301682,
      "grad_norm": 0.5299580097198486,
      "learning_rate": 4.1886458638715065e-05,
      "loss": 2.3097,
      "step": 98000
    },
    {
      "epoch": 0.7988766523634854,
      "grad_norm": 0.5144892334938049,
      "learning_rate": 4.180791413776097e-05,
      "loss": 2.3048,
      "step": 98500
    },
    {
      "epoch": 0.8029318637968026,
      "grad_norm": 0.49354636669158936,
      "learning_rate": 4.172906570703334e-05,
      "loss": 2.3086,
      "step": 99000
    },
    {
      "epoch": 0.8069870752301197,
      "grad_norm": 0.5165512561798096,
      "learning_rate": 4.164991477231205e-05,
      "loss": 2.3016,
      "step": 99500
    },
    {
      "epoch": 0.8110422866634369,
      "grad_norm": 0.5210604071617126,
      "learning_rate": 4.1570462764847015e-05,
      "loss": 2.3061,
      "step": 100000
    },
    {
      "epoch": 0.8150974980967541,
      "grad_norm": 0.5409877896308899,
      "learning_rate": 4.1490711121332295e-05,
      "loss": 2.3047,
      "step": 100500
    },
    {
      "epoch": 0.8191527095300712,
      "grad_norm": 0.5180912613868713,
      "learning_rate": 4.141066128388015e-05,
      "loss": 2.3022,
      "step": 101000
    },
    {
      "epoch": 0.8232079209633885,
      "grad_norm": 0.50931715965271,
      "learning_rate": 4.1330314699994944e-05,
      "loss": 2.3028,
      "step": 101500
    },
    {
      "epoch": 0.8272631323967057,
      "grad_norm": 0.5214248895645142,
      "learning_rate": 4.1249672822546955e-05,
      "loss": 2.3008,
      "step": 102000
    },
    {
      "epoch": 0.8313183438300228,
      "grad_norm": 0.5405526757240295,
      "learning_rate": 4.116873710974615e-05,
      "loss": 2.3023,
      "step": 102500
    },
    {
      "epoch": 0.83537355526334,
      "grad_norm": 0.5330822467803955,
      "learning_rate": 4.1087509025115734e-05,
      "loss": 2.3051,
      "step": 103000
    },
    {
      "epoch": 0.8394287666966572,
      "grad_norm": 0.5409587621688843,
      "learning_rate": 4.1005990037465785e-05,
      "loss": 2.302,
      "step": 103500
    },
    {
      "epoch": 0.8434839781299744,
      "grad_norm": 0.5315828919410706,
      "learning_rate": 4.0924181620866624e-05,
      "loss": 2.2981,
      "step": 104000
    },
    {
      "epoch": 0.8475391895632916,
      "grad_norm": 0.5308647751808167,
      "learning_rate": 4.084208525462219e-05,
      "loss": 2.2938,
      "step": 104500
    },
    {
      "epoch": 0.8515944009966088,
      "grad_norm": 0.5263886451721191,
      "learning_rate": 4.0759702423243276e-05,
      "loss": 2.295,
      "step": 105000
    },
    {
      "epoch": 0.855649612429926,
      "grad_norm": 0.5318562388420105,
      "learning_rate": 4.067703461642069e-05,
      "loss": 2.2964,
      "step": 105500
    },
    {
      "epoch": 0.8597048238632431,
      "grad_norm": 0.5380021333694458,
      "learning_rate": 4.0594083328998335e-05,
      "loss": 2.2895,
      "step": 106000
    },
    {
      "epoch": 0.8637600352965603,
      "grad_norm": 0.537627637386322,
      "learning_rate": 4.0510850060946143e-05,
      "loss": 2.2922,
      "step": 106500
    },
    {
      "epoch": 0.8678152467298775,
      "grad_norm": 0.5394497513771057,
      "learning_rate": 4.042733631733299e-05,
      "loss": 2.2961,
      "step": 107000
    },
    {
      "epoch": 0.8718704581631946,
      "grad_norm": 0.5409214496612549,
      "learning_rate": 4.034354360829946e-05,
      "loss": 2.294,
      "step": 107500
    },
    {
      "epoch": 0.8759256695965119,
      "grad_norm": 0.5413061380386353,
      "learning_rate": 4.025947344903052e-05,
      "loss": 2.2911,
      "step": 108000
    },
    {
      "epoch": 0.8799808810298291,
      "grad_norm": 0.5403317809104919,
      "learning_rate": 4.017512735972816e-05,
      "loss": 2.2906,
      "step": 108500
    },
    {
      "epoch": 0.8840360924631462,
      "grad_norm": 0.5209652185440063,
      "learning_rate": 4.009050686558389e-05,
      "loss": 2.2883,
      "step": 109000
    },
    {
      "epoch": 0.8880913038964634,
      "grad_norm": 0.5193003416061401,
      "learning_rate": 4.000561349675115e-05,
      "loss": 2.2903,
      "step": 109500
    },
    {
      "epoch": 0.8921465153297806,
      "grad_norm": 0.5441010594367981,
      "learning_rate": 3.992044878831764e-05,
      "loss": 2.2841,
      "step": 110000
    },
    {
      "epoch": 0.8962017267630978,
      "grad_norm": 0.536781907081604,
      "learning_rate": 3.9835014280277584e-05,
      "loss": 2.2843,
      "step": 110500
    },
    {
      "epoch": 0.900256938196415,
      "grad_norm": 0.5598934888839722,
      "learning_rate": 3.974931151750385e-05,
      "loss": 2.293,
      "step": 111000
    },
    {
      "epoch": 0.9043121496297322,
      "grad_norm": 0.5360564589500427,
      "learning_rate": 3.9663342049720046e-05,
      "loss": 2.2845,
      "step": 111500
    },
    {
      "epoch": 0.9083673610630494,
      "grad_norm": 0.5301195383071899,
      "learning_rate": 3.957710743147248e-05,
      "loss": 2.2888,
      "step": 112000
    },
    {
      "epoch": 0.9124225724963665,
      "grad_norm": 0.536584198474884,
      "learning_rate": 3.9490609222102056e-05,
      "loss": 2.2867,
      "step": 112500
    },
    {
      "epoch": 0.9164777839296837,
      "grad_norm": 0.5443806648254395,
      "learning_rate": 3.940384898571608e-05,
      "loss": 2.281,
      "step": 113000
    },
    {
      "epoch": 0.9205329953630009,
      "grad_norm": 0.5341424942016602,
      "learning_rate": 3.9316828291159945e-05,
      "loss": 2.2862,
      "step": 113500
    },
    {
      "epoch": 0.924588206796318,
      "grad_norm": 0.5316222906112671,
      "learning_rate": 3.922954871198883e-05,
      "loss": 2.2854,
      "step": 114000
    },
    {
      "epoch": 0.9286434182296353,
      "grad_norm": 0.5432649850845337,
      "learning_rate": 3.914201182643917e-05,
      "loss": 2.2822,
      "step": 114500
    },
    {
      "epoch": 0.9326986296629525,
      "grad_norm": 0.5397413969039917,
      "learning_rate": 3.9054219217400176e-05,
      "loss": 2.2812,
      "step": 115000
    },
    {
      "epoch": 0.9367538410962697,
      "grad_norm": 0.5237780809402466,
      "learning_rate": 3.8966172472385175e-05,
      "loss": 2.2797,
      "step": 115500
    },
    {
      "epoch": 0.9408090525295868,
      "grad_norm": 0.5564081072807312,
      "learning_rate": 3.8877873183502934e-05,
      "loss": 2.2812,
      "step": 116000
    },
    {
      "epoch": 0.944864263962904,
      "grad_norm": 0.5437479615211487,
      "learning_rate": 3.878932294742883e-05,
      "loss": 2.2806,
      "step": 116500
    },
    {
      "epoch": 0.9489194753962212,
      "grad_norm": 0.5499303340911865,
      "learning_rate": 3.870052336537602e-05,
      "loss": 2.2763,
      "step": 117000
    },
    {
      "epoch": 0.9529746868295383,
      "grad_norm": 0.5608726143836975,
      "learning_rate": 3.861147604306647e-05,
      "loss": 2.2767,
      "step": 117500
    },
    {
      "epoch": 0.9570298982628556,
      "grad_norm": 0.5344070196151733,
      "learning_rate": 3.8522182590701905e-05,
      "loss": 2.2747,
      "step": 118000
    },
    {
      "epoch": 0.9610851096961728,
      "grad_norm": 0.5565013289451599,
      "learning_rate": 3.843264462293473e-05,
      "loss": 2.2789,
      "step": 118500
    },
    {
      "epoch": 0.9651403211294899,
      "grad_norm": 0.5644780993461609,
      "learning_rate": 3.83428637588388e-05,
      "loss": 2.2784,
      "step": 119000
    },
    {
      "epoch": 0.9691955325628071,
      "grad_norm": 0.5427052974700928,
      "learning_rate": 3.825284162188012e-05,
      "loss": 2.2732,
      "step": 119500
    },
    {
      "epoch": 0.9732507439961243,
      "grad_norm": 0.546487033367157,
      "learning_rate": 3.8162579839887576e-05,
      "loss": 2.2758,
      "step": 120000
    },
    {
      "epoch": 0.9773059554294414,
      "grad_norm": 0.5496521592140198,
      "learning_rate": 3.807208004502341e-05,
      "loss": 2.2732,
      "step": 120500
    },
    {
      "epoch": 0.9813611668627586,
      "grad_norm": 0.5730307102203369,
      "learning_rate": 3.798134387375374e-05,
      "loss": 2.2674,
      "step": 121000
    },
    {
      "epoch": 0.9854163782960759,
      "grad_norm": 0.5476057529449463,
      "learning_rate": 3.7890372966819e-05,
      "loss": 2.2699,
      "step": 121500
    },
    {
      "epoch": 0.9894715897293931,
      "grad_norm": 0.5156444907188416,
      "learning_rate": 3.7799168969204204e-05,
      "loss": 2.2692,
      "step": 122000
    },
    {
      "epoch": 0.9935268011627102,
      "grad_norm": 0.5473557710647583,
      "learning_rate": 3.770773353010924e-05,
      "loss": 2.2701,
      "step": 122500
    },
    {
      "epoch": 0.9975820125960274,
      "grad_norm": 0.5534400343894958,
      "learning_rate": 3.761606830291911e-05,
      "loss": 2.2712,
      "step": 123000
    },
    {
      "epoch": 1.0016372240293445,
      "grad_norm": 0.5497949123382568,
      "learning_rate": 3.752417494517386e-05,
      "loss": 2.2592,
      "step": 123500
    },
    {
      "epoch": 1.0056924354626617,
      "grad_norm": 0.5588605403900146,
      "learning_rate": 3.7432055118538806e-05,
      "loss": 2.2494,
      "step": 124000
    },
    {
      "epoch": 1.009747646895979,
      "grad_norm": 0.5374738574028015,
      "learning_rate": 3.733971048877435e-05,
      "loss": 2.2487,
      "step": 124500
    },
    {
      "epoch": 1.0138028583292962,
      "grad_norm": 0.5497466325759888,
      "learning_rate": 3.724714272570593e-05,
      "loss": 2.2461,
      "step": 125000
    },
    {
      "epoch": 1.0178580697626134,
      "grad_norm": 0.5669336915016174,
      "learning_rate": 3.715435350319379e-05,
      "loss": 2.2415,
      "step": 125500
    },
    {
      "epoch": 1.0219132811959306,
      "grad_norm": 0.5785222053527832,
      "learning_rate": 3.706134449910271e-05,
      "loss": 2.2502,
      "step": 126000
    }
  ],
  "logging_steps": 500,
  "max_steps": 369894,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 3000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.834342663980319e+19,
  "train_batch_size": 12,
  "trial_name": null,
  "trial_params": null
}