{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025,
      "grad_norm": 8.230839729309082,
      "learning_rate": 4.9755e-05,
      "loss": 4.8912,
      "step": 50
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.688719749450684,
      "learning_rate": 4.9505e-05,
      "loss": 1.6864,
      "step": 100
    },
    {
      "epoch": 0.075,
      "grad_norm": 5.002687454223633,
      "learning_rate": 4.9255e-05,
      "loss": 1.0988,
      "step": 150
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.780168533325195,
      "learning_rate": 4.9005e-05,
      "loss": 0.8088,
      "step": 200
    },
    {
      "epoch": 0.125,
      "grad_norm": 4.83399772644043,
      "learning_rate": 4.8755e-05,
      "loss": 0.6977,
      "step": 250
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.855978488922119,
      "learning_rate": 4.8505e-05,
      "loss": 0.5413,
      "step": 300
    },
    {
      "epoch": 0.175,
      "grad_norm": 2.9968976974487305,
      "learning_rate": 4.8255e-05,
      "loss": 0.4831,
      "step": 350
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.2543575763702393,
      "learning_rate": 4.8005e-05,
      "loss": 0.464,
      "step": 400
    },
    {
      "epoch": 0.225,
      "grad_norm": 2.6646761894226074,
      "learning_rate": 4.7755e-05,
      "loss": 0.4362,
      "step": 450
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7676384449005127,
      "learning_rate": 4.7505e-05,
      "loss": 0.4083,
      "step": 500
    },
    {
      "epoch": 0.275,
      "grad_norm": 3.2820680141448975,
      "learning_rate": 4.725500000000001e-05,
      "loss": 0.3694,
      "step": 550
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.288299322128296,
      "learning_rate": 4.7005e-05,
      "loss": 0.3735,
      "step": 600
    },
    {
      "epoch": 0.325,
      "grad_norm": 3.515395402908325,
      "learning_rate": 4.6755e-05,
      "loss": 0.3563,
      "step": 650
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.2986159324646,
      "learning_rate": 4.6505e-05,
      "loss": 0.3822,
      "step": 700
    },
    {
      "epoch": 0.375,
      "grad_norm": 2.9694278240203857,
      "learning_rate": 4.6255000000000004e-05,
      "loss": 0.3447,
      "step": 750
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.727344512939453,
      "learning_rate": 4.6005000000000004e-05,
      "loss": 0.3232,
      "step": 800
    },
    {
      "epoch": 0.425,
      "grad_norm": 2.925055503845215,
      "learning_rate": 4.5755000000000005e-05,
      "loss": 0.3168,
      "step": 850
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7304507493972778,
      "learning_rate": 4.5505000000000006e-05,
      "loss": 0.3269,
      "step": 900
    },
    {
      "epoch": 0.475,
      "grad_norm": 2.8221561908721924,
      "learning_rate": 4.5255000000000006e-05,
      "loss": 0.3053,
      "step": 950
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.467146396636963,
      "learning_rate": 4.5005e-05,
      "loss": 0.3055,
      "step": 1000
    },
    {
      "epoch": 0.525,
      "grad_norm": 2.420332908630371,
      "learning_rate": 4.4755e-05,
      "loss": 0.3226,
      "step": 1050
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0012848377227783,
      "learning_rate": 4.4505e-05,
      "loss": 0.318,
      "step": 1100
    },
    {
      "epoch": 0.575,
      "grad_norm": 3.340888500213623,
      "learning_rate": 4.4255e-05,
      "loss": 0.3017,
      "step": 1150
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.88847279548645,
      "learning_rate": 4.4005e-05,
      "loss": 0.3067,
      "step": 1200
    },
    {
      "epoch": 0.625,
      "grad_norm": 3.5503618717193604,
      "learning_rate": 4.3755000000000004e-05,
      "loss": 0.2866,
      "step": 1250
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.97746741771698,
      "learning_rate": 4.3505000000000004e-05,
      "loss": 0.2764,
      "step": 1300
    },
    {
      "epoch": 0.675,
      "grad_norm": 2.513754367828369,
      "learning_rate": 4.3255e-05,
      "loss": 0.2738,
      "step": 1350
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.928359866142273,
      "learning_rate": 4.3005e-05,
      "loss": 0.272,
      "step": 1400
    },
    {
      "epoch": 0.725,
      "grad_norm": 3.0566446781158447,
      "learning_rate": 4.2755e-05,
      "loss": 0.2858,
      "step": 1450
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.126354932785034,
      "learning_rate": 4.2505e-05,
      "loss": 0.2687,
      "step": 1500
    },
    {
      "epoch": 0.775,
      "grad_norm": 2.6933629512786865,
      "learning_rate": 4.2255e-05,
      "loss": 0.2625,
      "step": 1550
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8243509531021118,
      "learning_rate": 4.2005e-05,
      "loss": 0.26,
      "step": 1600
    },
    {
      "epoch": 0.825,
      "grad_norm": 1.5713759660720825,
      "learning_rate": 4.1755e-05,
      "loss": 0.2658,
      "step": 1650
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0764338970184326,
      "learning_rate": 4.1504999999999996e-05,
      "loss": 0.2391,
      "step": 1700
    },
    {
      "epoch": 0.875,
      "grad_norm": 1.8904582262039185,
      "learning_rate": 4.1255e-05,
      "loss": 0.2362,
      "step": 1750
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.638389825820923,
      "learning_rate": 4.1005000000000005e-05,
      "loss": 0.2635,
      "step": 1800
    },
    {
      "epoch": 0.925,
      "grad_norm": 3.567256450653076,
      "learning_rate": 4.0755000000000005e-05,
      "loss": 0.2487,
      "step": 1850
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0650864839553833,
      "learning_rate": 4.0505000000000006e-05,
      "loss": 0.2561,
      "step": 1900
    },
    {
      "epoch": 0.975,
      "grad_norm": 1.547837257385254,
      "learning_rate": 4.025500000000001e-05,
      "loss": 0.2529,
      "step": 1950
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.24297297000885,
      "learning_rate": 4.0005e-05,
      "loss": 0.2592,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.17051538825035095,
      "eval_runtime": 64.5766,
      "eval_samples_per_second": 15.485,
      "eval_steps_per_second": 3.871,
      "step": 2000
    },
    {
      "epoch": 1.025,
      "grad_norm": 2.5889508724212646,
      "learning_rate": 3.9755e-05,
      "loss": 0.2366,
      "step": 2050
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.5479001998901367,
      "learning_rate": 3.9505e-05,
      "loss": 0.2362,
      "step": 2100
    },
    {
      "epoch": 1.075,
      "grad_norm": 1.6057442426681519,
      "learning_rate": 3.9255e-05,
      "loss": 0.2418,
      "step": 2150
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.403870582580566,
      "learning_rate": 3.9005000000000003e-05,
      "loss": 0.239,
      "step": 2200
    },
    {
      "epoch": 1.125,
      "grad_norm": 1.7852602005004883,
      "learning_rate": 3.8755000000000004e-05,
      "loss": 0.2317,
      "step": 2250
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.732895851135254,
      "learning_rate": 3.8505000000000005e-05,
      "loss": 0.2362,
      "step": 2300
    },
    {
      "epoch": 1.175,
      "grad_norm": 1.529485821723938,
      "learning_rate": 3.8255e-05,
      "loss": 0.2168,
      "step": 2350
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7006224393844604,
      "learning_rate": 3.8005e-05,
      "loss": 0.2307,
      "step": 2400
    },
    {
      "epoch": 1.225,
      "grad_norm": 1.7951953411102295,
      "learning_rate": 3.7755e-05,
      "loss": 0.2161,
      "step": 2450
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.2083799839019775,
      "learning_rate": 3.7505e-05,
      "loss": 0.2013,
      "step": 2500
    },
    {
      "epoch": 1.275,
      "grad_norm": 3.326378107070923,
      "learning_rate": 3.7255e-05,
      "loss": 0.2323,
      "step": 2550
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.719395160675049,
      "learning_rate": 3.7005e-05,
      "loss": 0.2244,
      "step": 2600
    },
    {
      "epoch": 1.325,
      "grad_norm": 1.8044047355651855,
      "learning_rate": 3.6755e-05,
      "loss": 0.2256,
      "step": 2650
    },
    {
      "epoch": 1.35,
      "grad_norm": 2.9023215770721436,
      "learning_rate": 3.6505e-05,
      "loss": 0.2171,
      "step": 2700
    },
    {
      "epoch": 1.375,
      "grad_norm": 3.302866220474243,
      "learning_rate": 3.6255e-05,
      "loss": 0.2366,
      "step": 2750
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5518901348114014,
      "learning_rate": 3.6005e-05,
      "loss": 0.2179,
      "step": 2800
    },
    {
      "epoch": 1.425,
      "grad_norm": 2.0442991256713867,
      "learning_rate": 3.5755e-05,
      "loss": 0.2054,
      "step": 2850
    },
    {
      "epoch": 1.45,
      "grad_norm": 2.274137020111084,
      "learning_rate": 3.5505e-05,
      "loss": 0.2158,
      "step": 2900
    },
    {
      "epoch": 1.475,
      "grad_norm": 2.449765205383301,
      "learning_rate": 3.5255e-05,
      "loss": 0.2241,
      "step": 2950
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.705886960029602,
      "learning_rate": 3.5005e-05,
      "loss": 0.2107,
      "step": 3000
    },
    {
      "epoch": 1.525,
      "grad_norm": 1.9636527299880981,
      "learning_rate": 3.4755e-05,
      "loss": 0.2127,
      "step": 3050
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.048348903656006,
      "learning_rate": 3.4505e-05,
      "loss": 0.2244,
      "step": 3100
    },
    {
      "epoch": 1.575,
      "grad_norm": 2.6252517700195312,
      "learning_rate": 3.4255e-05,
      "loss": 0.2038,
      "step": 3150
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.407904624938965,
      "learning_rate": 3.4005000000000004e-05,
      "loss": 0.2146,
      "step": 3200
    },
    {
      "epoch": 1.625,
      "grad_norm": 2.5721547603607178,
      "learning_rate": 3.3755000000000005e-05,
      "loss": 0.2236,
      "step": 3250
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.8725879192352295,
      "learning_rate": 3.3505000000000005e-05,
      "loss": 0.2178,
      "step": 3300
    },
    {
      "epoch": 1.675,
      "grad_norm": 3.0591113567352295,
      "learning_rate": 3.3255000000000006e-05,
      "loss": 0.208,
      "step": 3350
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.7534451484680176,
      "learning_rate": 3.3005e-05,
      "loss": 0.2073,
      "step": 3400
    },
    {
      "epoch": 1.725,
      "grad_norm": 2.5346784591674805,
      "learning_rate": 3.2755e-05,
      "loss": 0.1929,
      "step": 3450
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.595905780792236,
      "learning_rate": 3.2505e-05,
      "loss": 0.2034,
      "step": 3500
    },
    {
      "epoch": 1.775,
      "grad_norm": 2.5899410247802734,
      "learning_rate": 3.2255e-05,
      "loss": 0.2061,
      "step": 3550
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.3893258571624756,
      "learning_rate": 3.2005e-05,
      "loss": 0.1996,
      "step": 3600
    },
    {
      "epoch": 1.825,
      "grad_norm": 1.27448570728302,
      "learning_rate": 3.1755000000000003e-05,
      "loss": 0.2008,
      "step": 3650
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.1095352172851562,
      "learning_rate": 3.1505000000000004e-05,
      "loss": 0.194,
      "step": 3700
    },
    {
      "epoch": 1.875,
      "grad_norm": 3.202643632888794,
      "learning_rate": 3.1255e-05,
      "loss": 0.1924,
      "step": 3750
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.5014896392822266,
      "learning_rate": 3.1005e-05,
      "loss": 0.1814,
      "step": 3800
    },
    {
      "epoch": 1.925,
      "grad_norm": 1.8779760599136353,
      "learning_rate": 3.0755e-05,
      "loss": 0.1876,
      "step": 3850
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9704879522323608,
      "learning_rate": 3.0505e-05,
      "loss": 0.2035,
      "step": 3900
    },
    {
      "epoch": 1.975,
      "grad_norm": 1.5915486812591553,
      "learning_rate": 3.0255e-05,
      "loss": 0.1871,
      "step": 3950
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.360573172569275,
      "learning_rate": 3.0004999999999998e-05,
      "loss": 0.1721,
      "step": 4000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.14247964322566986,
      "eval_runtime": 57.8322,
      "eval_samples_per_second": 17.291,
      "eval_steps_per_second": 4.323,
      "step": 4000
    },
    {
      "epoch": 2.025,
      "grad_norm": 1.7904751300811768,
      "learning_rate": 2.9755e-05,
      "loss": 0.1838,
      "step": 4050
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.8318934440612793,
      "learning_rate": 2.9505e-05,
      "loss": 0.1726,
      "step": 4100
    },
    {
      "epoch": 2.075,
      "grad_norm": 2.9856910705566406,
      "learning_rate": 2.9255e-05,
      "loss": 0.1935,
      "step": 4150
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2206937074661255,
      "learning_rate": 2.9004999999999998e-05,
      "loss": 0.173,
      "step": 4200
    },
    {
      "epoch": 2.125,
      "grad_norm": 1.7264775037765503,
      "learning_rate": 2.8754999999999998e-05,
      "loss": 0.2022,
      "step": 4250
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7826628684997559,
      "learning_rate": 2.8505000000000002e-05,
      "loss": 0.1703,
      "step": 4300
    },
    {
      "epoch": 2.175,
      "grad_norm": 1.6034504175186157,
      "learning_rate": 2.8255000000000003e-05,
      "loss": 0.1809,
      "step": 4350
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0662107467651367,
      "learning_rate": 2.8005000000000004e-05,
      "loss": 0.1707,
      "step": 4400
    },
    {
      "epoch": 2.225,
      "grad_norm": 2.3653218746185303,
      "learning_rate": 2.7755000000000004e-05,
      "loss": 0.176,
      "step": 4450
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6109884977340698,
      "learning_rate": 2.7505000000000002e-05,
      "loss": 0.1823,
      "step": 4500
    },
    {
      "epoch": 2.275,
      "grad_norm": 1.0589327812194824,
      "learning_rate": 2.7255000000000002e-05,
      "loss": 0.172,
      "step": 4550
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8202016353607178,
      "learning_rate": 2.7005000000000003e-05,
      "loss": 0.1794,
      "step": 4600
    },
    {
      "epoch": 2.325,
      "grad_norm": 1.9865130186080933,
      "learning_rate": 2.6755000000000004e-05,
      "loss": 0.1734,
      "step": 4650
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5040825605392456,
      "learning_rate": 2.6505e-05,
      "loss": 0.1757,
      "step": 4700
    },
    {
      "epoch": 2.375,
      "grad_norm": 1.8152176141738892,
      "learning_rate": 2.6255000000000002e-05,
      "loss": 0.2076,
      "step": 4750
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0795916318893433,
      "learning_rate": 2.6005000000000003e-05,
      "loss": 0.1967,
      "step": 4800
    },
    {
      "epoch": 2.425,
      "grad_norm": 2.887174129486084,
      "learning_rate": 2.5755e-05,
      "loss": 0.189,
      "step": 4850
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6766159534454346,
      "learning_rate": 2.5505e-05,
      "loss": 0.1768,
      "step": 4900
    },
    {
      "epoch": 2.475,
      "grad_norm": 1.336639404296875,
      "learning_rate": 2.5255e-05,
      "loss": 0.1686,
      "step": 4950
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3105192184448242,
      "learning_rate": 2.5005000000000002e-05,
      "loss": 0.1906,
      "step": 5000
    },
    {
      "epoch": 2.525,
      "grad_norm": 3.529892683029175,
      "learning_rate": 2.4755e-05,
      "loss": 0.1802,
      "step": 5050
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.232103109359741,
      "learning_rate": 2.4505e-05,
      "loss": 0.1843,
      "step": 5100
    },
    {
      "epoch": 2.575,
      "grad_norm": 3.0913138389587402,
      "learning_rate": 2.4255e-05,
      "loss": 0.1825,
      "step": 5150
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.240821361541748,
      "learning_rate": 2.4005e-05,
      "loss": 0.1681,
      "step": 5200
    },
    {
      "epoch": 2.625,
      "grad_norm": 1.1728991270065308,
      "learning_rate": 2.3755000000000002e-05,
      "loss": 0.1711,
      "step": 5250
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0626940727233887,
      "learning_rate": 2.3505000000000003e-05,
      "loss": 0.1685,
      "step": 5300
    },
    {
      "epoch": 2.675,
      "grad_norm": 3.6568186283111572,
      "learning_rate": 2.3255e-05,
      "loss": 0.1835,
      "step": 5350
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.1457467079162598,
      "learning_rate": 2.3005e-05,
      "loss": 0.1844,
      "step": 5400
    },
    {
      "epoch": 2.725,
      "grad_norm": 2.00649094581604,
      "learning_rate": 2.2755e-05,
      "loss": 0.1853,
      "step": 5450
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8169037103652954,
      "learning_rate": 2.2505000000000002e-05,
      "loss": 0.1808,
      "step": 5500
    },
    {
      "epoch": 2.775,
      "grad_norm": 1.9334608316421509,
      "learning_rate": 2.2255e-05,
      "loss": 0.1579,
      "step": 5550
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9589757919311523,
      "learning_rate": 2.2005e-05,
      "loss": 0.149,
      "step": 5600
    },
    {
      "epoch": 2.825,
      "grad_norm": 1.5965051651000977,
      "learning_rate": 2.1755e-05,
      "loss": 0.1826,
      "step": 5650
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2711127996444702,
      "learning_rate": 2.1505e-05,
      "loss": 0.1633,
      "step": 5700
    },
    {
      "epoch": 2.875,
      "grad_norm": 1.6561341285705566,
      "learning_rate": 2.1255e-05,
      "loss": 0.174,
      "step": 5750
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9388906955718994,
      "learning_rate": 2.1005e-05,
      "loss": 0.1692,
      "step": 5800
    },
    {
      "epoch": 2.925,
      "grad_norm": 2.6603758335113525,
      "learning_rate": 2.0755000000000004e-05,
      "loss": 0.1672,
      "step": 5850
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.2559964656829834,
      "learning_rate": 2.0505e-05,
      "loss": 0.1757,
      "step": 5900
    },
    {
      "epoch": 2.975,
      "grad_norm": 1.919048547744751,
      "learning_rate": 2.0255000000000002e-05,
      "loss": 0.1735,
      "step": 5950
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.5572509765625,
      "learning_rate": 2.0005000000000002e-05,
      "loss": 0.1922,
      "step": 6000
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.12903086841106415,
      "eval_runtime": 57.4845,
      "eval_samples_per_second": 17.396,
      "eval_steps_per_second": 4.349,
      "step": 6000
    },
    {
      "epoch": 3.025,
      "grad_norm": 1.5934720039367676,
      "learning_rate": 1.9755e-05,
      "loss": 0.1699,
      "step": 6050
    },
    {
      "epoch": 3.05,
      "grad_norm": 2.1117281913757324,
      "learning_rate": 1.9505e-05,
      "loss": 0.1738,
      "step": 6100
    },
    {
      "epoch": 3.075,
      "grad_norm": 3.2797703742980957,
      "learning_rate": 1.9255e-05,
      "loss": 0.1694,
      "step": 6150
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.5360231399536133,
      "learning_rate": 1.9005000000000002e-05,
      "loss": 0.1512,
      "step": 6200
    },
    {
      "epoch": 3.125,
      "grad_norm": 1.4587829113006592,
      "learning_rate": 1.8755e-05,
      "loss": 0.1697,
      "step": 6250
    },
    {
      "epoch": 3.15,
      "grad_norm": 3.891972064971924,
      "learning_rate": 1.8505e-05,
      "loss": 0.1733,
      "step": 6300
    },
    {
      "epoch": 3.175,
      "grad_norm": 3.0549535751342773,
      "learning_rate": 1.8255e-05,
      "loss": 0.1514,
      "step": 6350
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.4268131256103516,
      "learning_rate": 1.8005e-05,
      "loss": 0.1844,
      "step": 6400
    },
    {
      "epoch": 3.225,
      "grad_norm": 1.3441020250320435,
      "learning_rate": 1.7755000000000002e-05,
      "loss": 0.1581,
      "step": 6450
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.2834066152572632,
      "learning_rate": 1.7505000000000003e-05,
      "loss": 0.1426,
      "step": 6500
    },
    {
      "epoch": 3.275,
      "grad_norm": 2.298339366912842,
      "learning_rate": 1.7255000000000003e-05,
      "loss": 0.1597,
      "step": 6550
    },
    {
      "epoch": 3.3,
      "grad_norm": 2.287705659866333,
      "learning_rate": 1.7005e-05,
      "loss": 0.1716,
      "step": 6600
    },
    {
      "epoch": 3.325,
      "grad_norm": 1.9559084177017212,
      "learning_rate": 1.6755e-05,
      "loss": 0.142,
      "step": 6650
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.5680958032608032,
      "learning_rate": 1.6505000000000002e-05,
      "loss": 0.1573,
      "step": 6700
    },
    {
      "epoch": 3.375,
      "grad_norm": 0.831417441368103,
      "learning_rate": 1.6255e-05,
      "loss": 0.1687,
      "step": 6750
    },
    {
      "epoch": 3.4,
      "grad_norm": 2.815106153488159,
      "learning_rate": 1.6005e-05,
      "loss": 0.1577,
      "step": 6800
    },
    {
      "epoch": 3.425,
      "grad_norm": 3.7050323486328125,
      "learning_rate": 1.5755e-05,
      "loss": 0.1719,
      "step": 6850
    },
    {
      "epoch": 3.45,
      "grad_norm": 2.6608221530914307,
      "learning_rate": 1.5505e-05,
      "loss": 0.1644,
      "step": 6900
    },
    {
      "epoch": 3.475,
      "grad_norm": 2.1603200435638428,
      "learning_rate": 1.5255e-05,
      "loss": 0.1501,
      "step": 6950
    },
    {
      "epoch": 3.5,
      "grad_norm": 5.807039260864258,
      "learning_rate": 1.5005e-05,
      "loss": 0.1557,
      "step": 7000
    },
    {
      "epoch": 3.525,
      "grad_norm": 2.840177297592163,
      "learning_rate": 1.4755e-05,
      "loss": 0.1632,
      "step": 7050
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.3503317832946777,
      "learning_rate": 1.4505000000000003e-05,
      "loss": 0.1546,
      "step": 7100
    },
    {
      "epoch": 3.575,
      "grad_norm": 2.2482845783233643,
      "learning_rate": 1.4255000000000002e-05,
      "loss": 0.1729,
      "step": 7150
    },
    {
      "epoch": 3.6,
      "grad_norm": 2.3148233890533447,
      "learning_rate": 1.4005000000000002e-05,
      "loss": 0.1611,
      "step": 7200
    },
    {
      "epoch": 3.625,
      "grad_norm": 1.579547643661499,
      "learning_rate": 1.3755000000000001e-05,
      "loss": 0.1448,
      "step": 7250
    },
    {
      "epoch": 3.65,
      "grad_norm": 2.187784433364868,
      "learning_rate": 1.3505e-05,
      "loss": 0.1517,
      "step": 7300
    },
    {
      "epoch": 3.675,
      "grad_norm": 2.1139655113220215,
      "learning_rate": 1.3255000000000001e-05,
      "loss": 0.1499,
      "step": 7350
    },
    {
      "epoch": 3.7,
      "grad_norm": 2.0396082401275635,
      "learning_rate": 1.3005e-05,
      "loss": 0.1669,
      "step": 7400
    },
    {
      "epoch": 3.725,
      "grad_norm": 3.240251064300537,
      "learning_rate": 1.2755e-05,
      "loss": 0.154,
      "step": 7450
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.9571031332015991,
      "learning_rate": 1.2505e-05,
      "loss": 0.1563,
      "step": 7500
    },
    {
      "epoch": 3.775,
      "grad_norm": 3.260962963104248,
      "learning_rate": 1.2255e-05,
      "loss": 0.169,
      "step": 7550
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.8954269886016846,
      "learning_rate": 1.2005000000000001e-05,
      "loss": 0.157,
      "step": 7600
    },
    {
      "epoch": 3.825,
      "grad_norm": 2.130872964859009,
      "learning_rate": 1.1755e-05,
      "loss": 0.1573,
      "step": 7650
    },
    {
      "epoch": 3.85,
      "grad_norm": 2.272871494293213,
      "learning_rate": 1.1505e-05,
      "loss": 0.1459,
      "step": 7700
    },
    {
      "epoch": 3.875,
      "grad_norm": 2.3315696716308594,
      "learning_rate": 1.1255e-05,
      "loss": 0.1647,
      "step": 7750
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.6526434421539307,
      "learning_rate": 1.1005e-05,
      "loss": 0.1436,
      "step": 7800
    },
    {
      "epoch": 3.925,
      "grad_norm": 0.8647462129592896,
      "learning_rate": 1.0755000000000001e-05,
      "loss": 0.1579,
      "step": 7850
    },
    {
      "epoch": 3.95,
      "grad_norm": 2.794496774673462,
      "learning_rate": 1.0505e-05,
      "loss": 0.17,
      "step": 7900
    },
    {
      "epoch": 3.975,
      "grad_norm": 1.3235639333724976,
      "learning_rate": 1.0255000000000001e-05,
      "loss": 0.1565,
      "step": 7950
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.952364206314087,
      "learning_rate": 1.0005e-05,
      "loss": 0.1558,
      "step": 8000
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.12239066511392593,
      "eval_runtime": 65.1197,
      "eval_samples_per_second": 15.356,
      "eval_steps_per_second": 3.839,
      "step": 8000
    },
    {
      "epoch": 4.025,
      "grad_norm": 2.830843210220337,
      "learning_rate": 9.755e-06,
      "loss": 0.1528,
      "step": 8050
    },
    {
      "epoch": 4.05,
      "grad_norm": 3.2282028198242188,
      "learning_rate": 9.505e-06,
      "loss": 0.1545,
      "step": 8100
    },
    {
      "epoch": 4.075,
      "grad_norm": 2.520493268966675,
      "learning_rate": 9.255e-06,
      "loss": 0.1643,
      "step": 8150
    },
    {
      "epoch": 4.1,
      "grad_norm": 1.4111956357955933,
      "learning_rate": 9.005000000000001e-06,
      "loss": 0.164,
      "step": 8200
    },
    {
      "epoch": 4.125,
      "grad_norm": 2.735272169113159,
      "learning_rate": 8.755e-06,
      "loss": 0.1411,
      "step": 8250
    },
    {
      "epoch": 4.15,
      "grad_norm": 1.5018057823181152,
      "learning_rate": 8.505e-06,
      "loss": 0.1607,
      "step": 8300
    },
    {
      "epoch": 4.175,
      "grad_norm": 2.3549647331237793,
      "learning_rate": 8.255e-06,
      "loss": 0.1589,
      "step": 8350
    },
    {
      "epoch": 4.2,
      "grad_norm": 1.8919823169708252,
      "learning_rate": 8.005e-06,
      "loss": 0.1354,
      "step": 8400
    },
    {
      "epoch": 4.225,
      "grad_norm": 1.9286110401153564,
      "learning_rate": 7.755e-06,
      "loss": 0.1523,
      "step": 8450
    },
    {
      "epoch": 4.25,
      "grad_norm": 1.6039056777954102,
      "learning_rate": 7.505000000000001e-06,
      "loss": 0.141,
      "step": 8500
    },
    {
      "epoch": 4.275,
      "grad_norm": 1.682285189628601,
      "learning_rate": 7.255000000000001e-06,
      "loss": 0.1542,
      "step": 8550
    },
    {
      "epoch": 4.3,
      "grad_norm": 2.6263935565948486,
      "learning_rate": 7.005000000000001e-06,
      "loss": 0.1622,
      "step": 8600
    },
    {
      "epoch": 4.325,
      "grad_norm": 1.4278345108032227,
      "learning_rate": 6.7550000000000005e-06,
      "loss": 0.1381,
      "step": 8650
    },
    {
      "epoch": 4.35,
      "grad_norm": 2.0380184650421143,
      "learning_rate": 6.505e-06,
      "loss": 0.1579,
      "step": 8700
    },
    {
      "epoch": 4.375,
      "grad_norm": 2.6025896072387695,
      "learning_rate": 6.254999999999999e-06,
      "loss": 0.1742,
      "step": 8750
    },
    {
      "epoch": 4.4,
      "grad_norm": 2.4303677082061768,
      "learning_rate": 6.005e-06,
      "loss": 0.139,
      "step": 8800
    },
    {
      "epoch": 4.425,
      "grad_norm": 1.5707917213439941,
      "learning_rate": 5.755e-06,
      "loss": 0.1442,
      "step": 8850
    },
    {
      "epoch": 4.45,
      "grad_norm": 3.853732109069824,
      "learning_rate": 5.505000000000001e-06,
      "loss": 0.1486,
      "step": 8900
    },
    {
      "epoch": 4.475,
      "grad_norm": 2.794379234313965,
      "learning_rate": 5.2550000000000005e-06,
      "loss": 0.1599,
      "step": 8950
    },
    {
      "epoch": 4.5,
      "grad_norm": 1.6828454732894897,
      "learning_rate": 5.005e-06,
      "loss": 0.1338,
      "step": 9000
    },
    {
      "epoch": 4.525,
      "grad_norm": 1.348374843597412,
      "learning_rate": 4.755e-06,
      "loss": 0.1308,
      "step": 9050
    },
    {
      "epoch": 4.55,
      "grad_norm": 3.534005880355835,
      "learning_rate": 4.505e-06,
      "loss": 0.1622,
      "step": 9100
    },
    {
      "epoch": 4.575,
      "grad_norm": 3.5305874347686768,
      "learning_rate": 4.255e-06,
      "loss": 0.1527,
      "step": 9150
    },
    {
      "epoch": 4.6,
      "grad_norm": 3.4382681846618652,
      "learning_rate": 4.005000000000001e-06,
      "loss": 0.1546,
      "step": 9200
    },
    {
      "epoch": 4.625,
      "grad_norm": 3.7415757179260254,
      "learning_rate": 3.755e-06,
      "loss": 0.1595,
      "step": 9250
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.252262830734253,
      "learning_rate": 3.505e-06,
      "loss": 0.1273,
      "step": 9300
    },
    {
      "epoch": 4.675,
      "grad_norm": 1.3889789581298828,
      "learning_rate": 3.2550000000000006e-06,
      "loss": 0.1519,
      "step": 9350
    },
    {
      "epoch": 4.7,
      "grad_norm": 1.1047695875167847,
      "learning_rate": 3.005e-06,
      "loss": 0.1472,
      "step": 9400
    },
    {
      "epoch": 4.725,
      "grad_norm": 1.56324303150177,
      "learning_rate": 2.7550000000000003e-06,
      "loss": 0.1217,
      "step": 9450
    },
    {
      "epoch": 4.75,
      "grad_norm": 2.3788673877716064,
      "learning_rate": 2.505e-06,
      "loss": 0.1534,
      "step": 9500
    },
    {
      "epoch": 4.775,
      "grad_norm": 1.8210504055023193,
      "learning_rate": 2.255e-06,
      "loss": 0.1424,
      "step": 9550
    },
    {
      "epoch": 4.8,
      "grad_norm": 1.365644097328186,
      "learning_rate": 2.005e-06,
      "loss": 0.1686,
      "step": 9600
    },
    {
      "epoch": 4.825,
      "grad_norm": 1.9589169025421143,
      "learning_rate": 1.7550000000000001e-06,
      "loss": 0.1543,
      "step": 9650
    },
    {
      "epoch": 4.85,
      "grad_norm": 2.8975677490234375,
      "learning_rate": 1.505e-06,
      "loss": 0.1449,
      "step": 9700
    },
    {
      "epoch": 4.875,
      "grad_norm": 0.7453130483627319,
      "learning_rate": 1.255e-06,
      "loss": 0.1548,
      "step": 9750
    },
    {
      "epoch": 4.9,
      "grad_norm": 1.493669867515564,
      "learning_rate": 1.0050000000000001e-06,
      "loss": 0.1537,
      "step": 9800
    },
    {
      "epoch": 4.925,
      "grad_norm": 2.2871060371398926,
      "learning_rate": 7.550000000000001e-07,
      "loss": 0.1611,
      "step": 9850
    },
    {
      "epoch": 4.95,
      "grad_norm": 1.1288138628005981,
      "learning_rate": 5.05e-07,
      "loss": 0.1619,
      "step": 9900
    },
    {
      "epoch": 4.975,
      "grad_norm": 1.5970274209976196,
      "learning_rate": 2.5500000000000005e-07,
      "loss": 0.1349,
      "step": 9950
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.328122615814209,
      "learning_rate": 5e-09,
      "loss": 0.1619,
      "step": 10000
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.12139008939266205,
      "eval_runtime": 58.7123,
      "eval_samples_per_second": 17.032,
      "eval_steps_per_second": 4.258,
      "step": 10000
    }
  ],
  "logging_steps": 50,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 338354503680000.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}