{
  "best_metric": 0.0002855238853953779,
  "best_model_checkpoint": "output_dir/checkpoint-11478",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 15304,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 9.973863042341873e-06,
      "loss": 0.5813,
      "step": 50
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.947726084683743e-06,
      "loss": 0.4291,
      "step": 100
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.921589127025614e-06,
      "loss": 0.3343,
      "step": 150
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.895452169367486e-06,
      "loss": 0.2588,
      "step": 200
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.869315211709358e-06,
      "loss": 0.1974,
      "step": 250
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.84317825405123e-06,
      "loss": 0.1511,
      "step": 300
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.817041296393101e-06,
      "loss": 0.1152,
      "step": 350
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.790904338734973e-06,
      "loss": 0.087,
      "step": 400
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.764767381076845e-06,
      "loss": 0.0699,
      "step": 450
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.738630423418715e-06,
      "loss": 0.0589,
      "step": 500
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.712493465760587e-06,
      "loss": 0.0489,
      "step": 550
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.686356508102458e-06,
      "loss": 0.0392,
      "step": 600
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.660219550444328e-06,
      "loss": 0.038,
      "step": 650
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.6340825927862e-06,
      "loss": 0.0292,
      "step": 700
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.607945635128072e-06,
      "loss": 0.0292,
      "step": 750
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.581808677469944e-06,
      "loss": 0.0259,
      "step": 800
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.555671719811814e-06,
      "loss": 0.02,
      "step": 850
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.529534762153685e-06,
      "loss": 0.0246,
      "step": 900
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.503397804495557e-06,
      "loss": 0.016,
      "step": 950
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.477260846837429e-06,
      "loss": 0.0129,
      "step": 1000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.4511238891793e-06,
      "loss": 0.0131,
      "step": 1050
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.424986931521172e-06,
      "loss": 0.0133,
      "step": 1100
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.398849973863044e-06,
      "loss": 0.0096,
      "step": 1150
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.372713016204914e-06,
      "loss": 0.0097,
      "step": 1200
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.346576058546786e-06,
      "loss": 0.0075,
      "step": 1250
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.320439100888658e-06,
      "loss": 0.0093,
      "step": 1300
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.29430214323053e-06,
      "loss": 0.0064,
      "step": 1350
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.2681651855724e-06,
      "loss": 0.0075,
      "step": 1400
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.242028227914271e-06,
      "loss": 0.0081,
      "step": 1450
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.215891270256143e-06,
      "loss": 0.0039,
      "step": 1500
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.189754312598015e-06,
      "loss": 0.0086,
      "step": 1550
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.163617354939886e-06,
      "loss": 0.0058,
      "step": 1600
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.137480397281758e-06,
      "loss": 0.0067,
      "step": 1650
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.111343439623628e-06,
      "loss": 0.0032,
      "step": 1700
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.0852064819655e-06,
      "loss": 0.0064,
      "step": 1750
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.059069524307372e-06,
      "loss": 0.0032,
      "step": 1800
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.032932566649243e-06,
      "loss": 0.0028,
      "step": 1850
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.006795608991115e-06,
      "loss": 0.0041,
      "step": 1900
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.9997379454926625,
      "eval_loss": 0.0027031726203858852,
      "eval_runtime": 26.3628,
      "eval_samples_per_second": 1160.801,
      "eval_steps_per_second": 18.17,
      "step": 1913
    },
    {
      "epoch": 1.02,
      "learning_rate": 8.980658651332985e-06,
      "loss": 0.0029,
      "step": 1950
    },
    {
      "epoch": 1.05,
      "learning_rate": 8.954521693674857e-06,
      "loss": 0.0023,
      "step": 2000
    },
    {
      "epoch": 1.07,
      "learning_rate": 8.928384736016729e-06,
      "loss": 0.0038,
      "step": 2050
    },
    {
      "epoch": 1.1,
      "learning_rate": 8.9022477783586e-06,
      "loss": 0.0042,
      "step": 2100
    },
    {
      "epoch": 1.12,
      "learning_rate": 8.87611082070047e-06,
      "loss": 0.0037,
      "step": 2150
    },
    {
      "epoch": 1.15,
      "learning_rate": 8.849973863042342e-06,
      "loss": 0.0019,
      "step": 2200
    },
    {
      "epoch": 1.18,
      "learning_rate": 8.823836905384214e-06,
      "loss": 0.0023,
      "step": 2250
    },
    {
      "epoch": 1.2,
      "learning_rate": 8.797699947726086e-06,
      "loss": 0.0037,
      "step": 2300
    },
    {
      "epoch": 1.23,
      "learning_rate": 8.771562990067957e-06,
      "loss": 0.0022,
      "step": 2350
    },
    {
      "epoch": 1.25,
      "learning_rate": 8.745426032409829e-06,
      "loss": 0.0053,
      "step": 2400
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.7192890747517e-06,
      "loss": 0.002,
      "step": 2450
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.69315211709357e-06,
      "loss": 0.0011,
      "step": 2500
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.667015159435443e-06,
      "loss": 0.0049,
      "step": 2550
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.640878201777314e-06,
      "loss": 0.0049,
      "step": 2600
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.614741244119186e-06,
      "loss": 0.0015,
      "step": 2650
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.588604286461056e-06,
      "loss": 0.003,
      "step": 2700
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.562467328802928e-06,
      "loss": 0.0025,
      "step": 2750
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.5363303711448e-06,
      "loss": 0.0011,
      "step": 2800
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.51019341348667e-06,
      "loss": 0.002,
      "step": 2850
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.484056455828541e-06,
      "loss": 0.0013,
      "step": 2900
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.457919498170413e-06,
      "loss": 0.001,
      "step": 2950
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.431782540512285e-06,
      "loss": 0.0016,
      "step": 3000
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.405645582854157e-06,
      "loss": 0.0007,
      "step": 3050
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.379508625196028e-06,
      "loss": 0.0009,
      "step": 3100
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.3533716675379e-06,
      "loss": 0.0009,
      "step": 3150
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.327234709879772e-06,
      "loss": 0.0007,
      "step": 3200
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.301097752221642e-06,
      "loss": 0.0011,
      "step": 3250
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.274960794563513e-06,
      "loss": 0.003,
      "step": 3300
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.248823836905385e-06,
      "loss": 0.0027,
      "step": 3350
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.222686879247255e-06,
      "loss": 0.0011,
      "step": 3400
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.196549921589127e-06,
      "loss": 0.0007,
      "step": 3450
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.170412963930999e-06,
      "loss": 0.0004,
      "step": 3500
    },
    {
      "epoch": 1.86,
      "learning_rate": 8.14427600627287e-06,
      "loss": 0.0008,
      "step": 3550
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.118139048614742e-06,
      "loss": 0.0009,
      "step": 3600
    },
    {
      "epoch": 1.91,
      "learning_rate": 8.092002090956614e-06,
      "loss": 0.0014,
      "step": 3650
    },
    {
      "epoch": 1.93,
      "learning_rate": 8.065865133298486e-06,
      "loss": 0.0005,
      "step": 3700
    },
    {
      "epoch": 1.96,
      "learning_rate": 8.039728175640356e-06,
      "loss": 0.0024,
      "step": 3750
    },
    {
      "epoch": 1.99,
      "learning_rate": 8.013591217982227e-06,
      "loss": 0.0007,
      "step": 3800
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.9999672517684044,
      "eval_loss": 0.0006587088573724031,
      "eval_runtime": 25.8878,
      "eval_samples_per_second": 1182.101,
      "eval_steps_per_second": 18.503,
      "step": 3826
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.9874542603241e-06,
      "loss": 0.0004,
      "step": 3850
    },
    {
      "epoch": 2.04,
      "learning_rate": 7.961317302665971e-06,
      "loss": 0.0004,
      "step": 3900
    },
    {
      "epoch": 2.06,
      "learning_rate": 7.935180345007841e-06,
      "loss": 0.0005,
      "step": 3950
    },
    {
      "epoch": 2.09,
      "learning_rate": 7.909043387349713e-06,
      "loss": 0.0011,
      "step": 4000
    },
    {
      "epoch": 2.12,
      "learning_rate": 7.882906429691584e-06,
      "loss": 0.0009,
      "step": 4050
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.856769472033456e-06,
      "loss": 0.0007,
      "step": 4100
    },
    {
      "epoch": 2.17,
      "learning_rate": 7.830632514375326e-06,
      "loss": 0.0003,
      "step": 4150
    },
    {
      "epoch": 2.2,
      "learning_rate": 7.804495556717198e-06,
      "loss": 0.0026,
      "step": 4200
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.77835859905907e-06,
      "loss": 0.0006,
      "step": 4250
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.752221641400941e-06,
      "loss": 0.001,
      "step": 4300
    },
    {
      "epoch": 2.27,
      "learning_rate": 7.726084683742813e-06,
      "loss": 0.0012,
      "step": 4350
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.699947726084685e-06,
      "loss": 0.0006,
      "step": 4400
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.673810768426557e-06,
      "loss": 0.0019,
      "step": 4450
    },
    {
      "epoch": 2.35,
      "learning_rate": 7.647673810768427e-06,
      "loss": 0.0004,
      "step": 4500
    },
    {
      "epoch": 2.38,
      "learning_rate": 7.6215368531102985e-06,
      "loss": 0.0004,
      "step": 4550
    },
    {
      "epoch": 2.4,
      "learning_rate": 7.59539989545217e-06,
      "loss": 0.0017,
      "step": 4600
    },
    {
      "epoch": 2.43,
      "learning_rate": 7.569262937794042e-06,
      "loss": 0.0004,
      "step": 4650
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.543125980135913e-06,
      "loss": 0.0003,
      "step": 4700
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.516989022477785e-06,
      "loss": 0.0003,
      "step": 4750
    },
    {
      "epoch": 2.51,
      "learning_rate": 7.4908520648196555e-06,
      "loss": 0.0008,
      "step": 4800
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.464715107161526e-06,
      "loss": 0.0024,
      "step": 4850
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.438578149503398e-06,
      "loss": 0.0003,
      "step": 4900
    },
    {
      "epoch": 2.59,
      "learning_rate": 7.41244119184527e-06,
      "loss": 0.0003,
      "step": 4950
    },
    {
      "epoch": 2.61,
      "learning_rate": 7.3863042341871416e-06,
      "loss": 0.0003,
      "step": 5000
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.3601672765290125e-06,
      "loss": 0.0002,
      "step": 5050
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.334030318870884e-06,
      "loss": 0.0002,
      "step": 5100
    },
    {
      "epoch": 2.69,
      "learning_rate": 7.307893361212756e-06,
      "loss": 0.0002,
      "step": 5150
    },
    {
      "epoch": 2.72,
      "learning_rate": 7.281756403554628e-06,
      "loss": 0.0003,
      "step": 5200
    },
    {
      "epoch": 2.74,
      "learning_rate": 7.255619445896498e-06,
      "loss": 0.0003,
      "step": 5250
    },
    {
      "epoch": 2.77,
      "learning_rate": 7.2294824882383695e-06,
      "loss": 0.0002,
      "step": 5300
    },
    {
      "epoch": 2.8,
      "learning_rate": 7.203345530580241e-06,
      "loss": 0.0033,
      "step": 5350
    },
    {
      "epoch": 2.82,
      "learning_rate": 7.177208572922112e-06,
      "loss": 0.0002,
      "step": 5400
    },
    {
      "epoch": 2.85,
      "learning_rate": 7.151071615263984e-06,
      "loss": 0.0002,
      "step": 5450
    },
    {
      "epoch": 2.88,
      "learning_rate": 7.1249346576058556e-06,
      "loss": 0.0003,
      "step": 5500
    },
    {
      "epoch": 2.9,
      "learning_rate": 7.098797699947727e-06,
      "loss": 0.0007,
      "step": 5550
    },
    {
      "epoch": 2.93,
      "learning_rate": 7.072660742289597e-06,
      "loss": 0.0014,
      "step": 5600
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.046523784631469e-06,
      "loss": 0.0003,
      "step": 5650
    },
    {
      "epoch": 2.98,
      "learning_rate": 7.020386826973341e-06,
      "loss": 0.0002,
      "step": 5700
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.9999181227176707,
      "eval_loss": 0.0006406409665942192,
      "eval_runtime": 25.7306,
      "eval_samples_per_second": 1189.325,
      "eval_steps_per_second": 18.616,
      "step": 5739
    },
    {
      "epoch": 3.01,
      "learning_rate": 6.9942498693152126e-06,
      "loss": 0.0002,
      "step": 5750
    },
    {
      "epoch": 3.03,
      "learning_rate": 6.9681129116570834e-06,
      "loss": 0.0004,
      "step": 5800
    },
    {
      "epoch": 3.06,
      "learning_rate": 6.941975953998955e-06,
      "loss": 0.0002,
      "step": 5850
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.915838996340827e-06,
      "loss": 0.0002,
      "step": 5900
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.889702038682697e-06,
      "loss": 0.0001,
      "step": 5950
    },
    {
      "epoch": 3.14,
      "learning_rate": 6.863565081024569e-06,
      "loss": 0.0002,
      "step": 6000
    },
    {
      "epoch": 3.16,
      "learning_rate": 6.8374281233664404e-06,
      "loss": 0.0007,
      "step": 6050
    },
    {
      "epoch": 3.19,
      "learning_rate": 6.811291165708312e-06,
      "loss": 0.0001,
      "step": 6100
    },
    {
      "epoch": 3.21,
      "learning_rate": 6.785154208050183e-06,
      "loss": 0.0001,
      "step": 6150
    },
    {
      "epoch": 3.24,
      "learning_rate": 6.759017250392055e-06,
      "loss": 0.0001,
      "step": 6200
    },
    {
      "epoch": 3.27,
      "learning_rate": 6.7328802927339265e-06,
      "loss": 0.0034,
      "step": 6250
    },
    {
      "epoch": 3.29,
      "learning_rate": 6.706743335075798e-06,
      "loss": 0.0007,
      "step": 6300
    },
    {
      "epoch": 3.32,
      "learning_rate": 6.680606377417669e-06,
      "loss": 0.0001,
      "step": 6350
    },
    {
      "epoch": 3.35,
      "learning_rate": 6.654469419759541e-06,
      "loss": 0.0003,
      "step": 6400
    },
    {
      "epoch": 3.37,
      "learning_rate": 6.628332462101413e-06,
      "loss": 0.0002,
      "step": 6450
    },
    {
      "epoch": 3.4,
      "learning_rate": 6.602195504443283e-06,
      "loss": 0.0001,
      "step": 6500
    },
    {
      "epoch": 3.42,
      "learning_rate": 6.5760585467851544e-06,
      "loss": 0.0001,
      "step": 6550
    },
    {
      "epoch": 3.45,
      "learning_rate": 6.549921589127026e-06,
      "loss": 0.0002,
      "step": 6600
    },
    {
      "epoch": 3.48,
      "learning_rate": 6.523784631468898e-06,
      "loss": 0.0002,
      "step": 6650
    },
    {
      "epoch": 3.5,
      "learning_rate": 6.497647673810769e-06,
      "loss": 0.0001,
      "step": 6700
    },
    {
      "epoch": 3.53,
      "learning_rate": 6.4715107161526405e-06,
      "loss": 0.0001,
      "step": 6750
    },
    {
      "epoch": 3.55,
      "learning_rate": 6.445373758494512e-06,
      "loss": 0.0001,
      "step": 6800
    },
    {
      "epoch": 3.58,
      "learning_rate": 6.419236800836384e-06,
      "loss": 0.0001,
      "step": 6850
    },
    {
      "epoch": 3.61,
      "learning_rate": 6.393099843178254e-06,
      "loss": 0.0028,
      "step": 6900
    },
    {
      "epoch": 3.63,
      "learning_rate": 6.366962885520126e-06,
      "loss": 0.0002,
      "step": 6950
    },
    {
      "epoch": 3.66,
      "learning_rate": 6.3408259278619975e-06,
      "loss": 0.0003,
      "step": 7000
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.3146889702038684e-06,
      "loss": 0.0002,
      "step": 7050
    },
    {
      "epoch": 3.71,
      "learning_rate": 6.28855201254574e-06,
      "loss": 0.0001,
      "step": 7100
    },
    {
      "epoch": 3.74,
      "learning_rate": 6.262415054887612e-06,
      "loss": 0.0001,
      "step": 7150
    },
    {
      "epoch": 3.76,
      "learning_rate": 6.236278097229484e-06,
      "loss": 0.0001,
      "step": 7200
    },
    {
      "epoch": 3.79,
      "learning_rate": 6.210141139571354e-06,
      "loss": 0.0001,
      "step": 7250
    },
    {
      "epoch": 3.82,
      "learning_rate": 6.184004181913225e-06,
      "loss": 0.0001,
      "step": 7300
    },
    {
      "epoch": 3.84,
      "learning_rate": 6.157867224255097e-06,
      "loss": 0.0001,
      "step": 7350
    },
    {
      "epoch": 3.87,
      "learning_rate": 6.131730266596969e-06,
      "loss": 0.0039,
      "step": 7400
    },
    {
      "epoch": 3.89,
      "learning_rate": 6.10559330893884e-06,
      "loss": 0.0036,
      "step": 7450
    },
    {
      "epoch": 3.92,
      "learning_rate": 6.0794563512807115e-06,
      "loss": 0.0023,
      "step": 7500
    },
    {
      "epoch": 3.95,
      "learning_rate": 6.053319393622583e-06,
      "loss": 0.0003,
      "step": 7550
    },
    {
      "epoch": 3.97,
      "learning_rate": 6.027182435964454e-06,
      "loss": 0.0001,
      "step": 7600
    },
    {
      "epoch": 4.0,
      "learning_rate": 6.001045478306325e-06,
      "loss": 0.0001,
      "step": 7650
    },
    {
      "epoch": 4.0,
      "eval_f1": 0.9999836256160861,
      "eval_loss": 0.00036523217568174005,
      "eval_runtime": 25.8257,
      "eval_samples_per_second": 1184.945,
      "eval_steps_per_second": 18.547,
      "step": 7652
    },
    {
      "epoch": 4.03,
      "learning_rate": 5.974908520648197e-06,
      "loss": 0.0001,
      "step": 7700
    },
    {
      "epoch": 4.05,
      "learning_rate": 5.9487715629900685e-06,
      "loss": 0.0001,
      "step": 7750
    },
    {
      "epoch": 4.08,
      "learning_rate": 5.922634605331939e-06,
      "loss": 0.0001,
      "step": 7800
    },
    {
      "epoch": 4.1,
      "learning_rate": 5.896497647673811e-06,
      "loss": 0.0001,
      "step": 7850
    },
    {
      "epoch": 4.13,
      "learning_rate": 5.870360690015683e-06,
      "loss": 0.0001,
      "step": 7900
    },
    {
      "epoch": 4.16,
      "learning_rate": 5.844223732357555e-06,
      "loss": 0.0001,
      "step": 7950
    },
    {
      "epoch": 4.18,
      "learning_rate": 5.8180867746994255e-06,
      "loss": 0.0005,
      "step": 8000
    },
    {
      "epoch": 4.21,
      "learning_rate": 5.791949817041297e-06,
      "loss": 0.0001,
      "step": 8050
    },
    {
      "epoch": 4.23,
      "learning_rate": 5.765812859383169e-06,
      "loss": 0.0001,
      "step": 8100
    },
    {
      "epoch": 4.26,
      "learning_rate": 5.739675901725039e-06,
      "loss": 0.0001,
      "step": 8150
    },
    {
      "epoch": 4.29,
      "learning_rate": 5.713538944066911e-06,
      "loss": 0.0001,
      "step": 8200
    },
    {
      "epoch": 4.31,
      "learning_rate": 5.6874019864087825e-06,
      "loss": 0.0001,
      "step": 8250
    },
    {
      "epoch": 4.34,
      "learning_rate": 5.661265028750654e-06,
      "loss": 0.0001,
      "step": 8300
    },
    {
      "epoch": 4.36,
      "learning_rate": 5.635128071092525e-06,
      "loss": 0.0001,
      "step": 8350
    },
    {
      "epoch": 4.39,
      "learning_rate": 5.608991113434397e-06,
      "loss": 0.0001,
      "step": 8400
    },
    {
      "epoch": 4.42,
      "learning_rate": 5.582854155776269e-06,
      "loss": 0.0021,
      "step": 8450
    },
    {
      "epoch": 4.44,
      "learning_rate": 5.55671719811814e-06,
      "loss": 0.0006,
      "step": 8500
    },
    {
      "epoch": 4.47,
      "learning_rate": 5.53058024046001e-06,
      "loss": 0.0,
      "step": 8550
    },
    {
      "epoch": 4.5,
      "learning_rate": 5.504443282801882e-06,
      "loss": 0.0001,
      "step": 8600
    },
    {
      "epoch": 4.52,
      "learning_rate": 5.478306325143754e-06,
      "loss": 0.0001,
      "step": 8650
    },
    {
      "epoch": 4.55,
      "learning_rate": 5.452169367485625e-06,
      "loss": 0.0,
      "step": 8700
    },
    {
      "epoch": 4.57,
      "learning_rate": 5.4260324098274965e-06,
      "loss": 0.0015,
      "step": 8750
    },
    {
      "epoch": 4.6,
      "learning_rate": 5.399895452169368e-06,
      "loss": 0.0004,
      "step": 8800
    },
    {
      "epoch": 4.63,
      "learning_rate": 5.37375849451124e-06,
      "loss": 0.0,
      "step": 8850
    },
    {
      "epoch": 4.65,
      "learning_rate": 5.34762153685311e-06,
      "loss": 0.002,
      "step": 8900
    },
    {
      "epoch": 4.68,
      "learning_rate": 5.321484579194982e-06,
      "loss": 0.0001,
      "step": 8950
    },
    {
      "epoch": 4.7,
      "learning_rate": 5.2953476215368535e-06,
      "loss": 0.0001,
      "step": 9000
    },
    {
      "epoch": 4.73,
      "learning_rate": 5.269210663878725e-06,
      "loss": 0.0001,
      "step": 9050
    },
    {
      "epoch": 4.76,
      "learning_rate": 5.243073706220596e-06,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 4.78,
      "learning_rate": 5.216936748562468e-06,
      "loss": 0.0,
      "step": 9150
    },
    {
      "epoch": 4.81,
      "learning_rate": 5.19079979090434e-06,
      "loss": 0.0001,
      "step": 9200
    },
    {
      "epoch": 4.84,
      "learning_rate": 5.1646628332462105e-06,
      "loss": 0.0001,
      "step": 9250
    },
    {
      "epoch": 4.86,
      "learning_rate": 5.138525875588081e-06,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 4.89,
      "learning_rate": 5.112388917929953e-06,
      "loss": 0.0,
      "step": 9350
    },
    {
      "epoch": 4.91,
      "learning_rate": 5.086251960271825e-06,
      "loss": 0.0001,
      "step": 9400
    },
    {
      "epoch": 4.94,
      "learning_rate": 5.060115002613696e-06,
      "loss": 0.0008,
      "step": 9450
    },
    {
      "epoch": 4.97,
      "learning_rate": 5.0339780449555675e-06,
      "loss": 0.0001,
      "step": 9500
    },
    {
      "epoch": 4.99,
      "learning_rate": 5.007841087297439e-06,
      "loss": 0.0001,
      "step": 9550
    },
    {
      "epoch": 5.0,
      "eval_f1": 0.9999836256160861,
      "eval_loss": 0.0003559759061317891,
      "eval_runtime": 25.8178,
      "eval_samples_per_second": 1185.304,
      "eval_steps_per_second": 18.553,
      "step": 9565
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.98170412963931e-06,
      "loss": 0.0001,
      "step": 9600
    },
    {
      "epoch": 5.04,
      "learning_rate": 4.955567171981182e-06,
      "loss": 0.0,
      "step": 9650
    },
    {
      "epoch": 5.07,
      "learning_rate": 4.929430214323054e-06,
      "loss": 0.0,
      "step": 9700
    },
    {
      "epoch": 5.1,
      "learning_rate": 4.9032932566649245e-06,
      "loss": 0.0008,
      "step": 9750
    },
    {
      "epoch": 5.12,
      "learning_rate": 4.877156299006796e-06,
      "loss": 0.0001,
      "step": 9800
    },
    {
      "epoch": 5.15,
      "learning_rate": 4.851019341348667e-06,
      "loss": 0.0,
      "step": 9850
    },
    {
      "epoch": 5.18,
      "learning_rate": 4.824882383690539e-06,
      "loss": 0.0,
      "step": 9900
    },
    {
      "epoch": 5.2,
      "learning_rate": 4.79874542603241e-06,
      "loss": 0.0,
      "step": 9950
    },
    {
      "epoch": 5.23,
      "learning_rate": 4.7726084683742815e-06,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 5.25,
      "learning_rate": 4.746471510716153e-06,
      "loss": 0.0,
      "step": 10050
    },
    {
      "epoch": 5.28,
      "learning_rate": 4.720334553058025e-06,
      "loss": 0.0,
      "step": 10100
    },
    {
      "epoch": 5.31,
      "learning_rate": 4.694197595399896e-06,
      "loss": 0.0,
      "step": 10150
    },
    {
      "epoch": 5.33,
      "learning_rate": 4.668060637741767e-06,
      "loss": 0.0,
      "step": 10200
    },
    {
      "epoch": 5.36,
      "learning_rate": 4.6419236800836385e-06,
      "loss": 0.0,
      "step": 10250
    },
    {
      "epoch": 5.38,
      "learning_rate": 4.61578672242551e-06,
      "loss": 0.0,
      "step": 10300
    },
    {
      "epoch": 5.41,
      "learning_rate": 4.589649764767382e-06,
      "loss": 0.0,
      "step": 10350
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.563512807109253e-06,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.537375849451125e-06,
      "loss": 0.0,
      "step": 10450
    },
    {
      "epoch": 5.49,
      "learning_rate": 4.5112388917929955e-06,
      "loss": 0.0,
      "step": 10500
    },
    {
      "epoch": 5.51,
      "learning_rate": 4.485101934134867e-06,
      "loss": 0.0,
      "step": 10550
    },
    {
      "epoch": 5.54,
      "learning_rate": 4.458964976476738e-06,
      "loss": 0.0,
      "step": 10600
    },
    {
      "epoch": 5.57,
      "learning_rate": 4.43282801881861e-06,
      "loss": 0.0,
      "step": 10650
    },
    {
      "epoch": 5.59,
      "learning_rate": 4.4066910611604816e-06,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 5.62,
      "learning_rate": 4.3805541035023525e-06,
      "loss": 0.0001,
      "step": 10750
    },
    {
      "epoch": 5.65,
      "learning_rate": 4.354417145844224e-06,
      "loss": 0.0,
      "step": 10800
    },
    {
      "epoch": 5.67,
      "learning_rate": 4.328280188186095e-06,
      "loss": 0.0,
      "step": 10850
    },
    {
      "epoch": 5.7,
      "learning_rate": 4.302143230527967e-06,
      "loss": 0.0,
      "step": 10900
    },
    {
      "epoch": 5.72,
      "learning_rate": 4.2760062728698386e-06,
      "loss": 0.0001,
      "step": 10950
    },
    {
      "epoch": 5.75,
      "learning_rate": 4.2498693152117095e-06,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 5.78,
      "learning_rate": 4.223732357553581e-06,
      "loss": 0.0014,
      "step": 11050
    },
    {
      "epoch": 5.8,
      "learning_rate": 4.197595399895453e-06,
      "loss": 0.0,
      "step": 11100
    },
    {
      "epoch": 5.83,
      "learning_rate": 4.171458442237324e-06,
      "loss": 0.0,
      "step": 11150
    },
    {
      "epoch": 5.85,
      "learning_rate": 4.1453214845791956e-06,
      "loss": 0.0001,
      "step": 11200
    },
    {
      "epoch": 5.88,
      "learning_rate": 4.1191845269210665e-06,
      "loss": 0.0,
      "step": 11250
    },
    {
      "epoch": 5.91,
      "learning_rate": 4.093047569262938e-06,
      "loss": 0.0,
      "step": 11300
    },
    {
      "epoch": 5.93,
      "learning_rate": 4.06691061160481e-06,
      "loss": 0.0005,
      "step": 11350
    },
    {
      "epoch": 5.96,
      "learning_rate": 4.040773653946681e-06,
      "loss": 0.0,
      "step": 11400
    },
    {
      "epoch": 5.99,
      "learning_rate": 4.0146366962885526e-06,
      "loss": 0.0,
      "step": 11450
    },
    {
      "epoch": 6.0,
      "eval_f1": 0.9999836256160861,
      "eval_loss": 0.0002855238853953779,
      "eval_runtime": 25.8432,
      "eval_samples_per_second": 1184.141,
      "eval_steps_per_second": 18.535,
      "step": 11478
    },
    {
      "epoch": 6.01,
      "learning_rate": 3.9884997386304235e-06,
      "loss": 0.0,
      "step": 11500
    },
    {
      "epoch": 6.04,
      "learning_rate": 3.962362780972295e-06,
      "loss": 0.0,
      "step": 11550
    },
    {
      "epoch": 6.06,
      "learning_rate": 3.936225823314166e-06,
      "loss": 0.0007,
      "step": 11600
    },
    {
      "epoch": 6.09,
      "learning_rate": 3.910088865656038e-06,
      "loss": 0.0,
      "step": 11650
    },
    {
      "epoch": 6.12,
      "learning_rate": 3.8839519079979096e-06,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 6.14,
      "learning_rate": 3.857814950339781e-06,
      "loss": 0.0012,
      "step": 11750
    },
    {
      "epoch": 6.17,
      "learning_rate": 3.831677992681652e-06,
      "loss": 0.0,
      "step": 11800
    },
    {
      "epoch": 6.19,
      "learning_rate": 3.8055410350235235e-06,
      "loss": 0.0,
      "step": 11850
    },
    {
      "epoch": 6.22,
      "learning_rate": 3.7794040773653952e-06,
      "loss": 0.0,
      "step": 11900
    },
    {
      "epoch": 6.25,
      "learning_rate": 3.753267119707266e-06,
      "loss": 0.0,
      "step": 11950
    },
    {
      "epoch": 6.27,
      "learning_rate": 3.727130162049138e-06,
      "loss": 0.0,
      "step": 12000
    },
    {
      "epoch": 6.3,
      "learning_rate": 3.700993204391009e-06,
      "loss": 0.0,
      "step": 12050
    },
    {
      "epoch": 6.33,
      "learning_rate": 3.674856246732881e-06,
      "loss": 0.0,
      "step": 12100
    },
    {
      "epoch": 6.35,
      "learning_rate": 3.648719289074752e-06,
      "loss": 0.0,
      "step": 12150
    },
    {
      "epoch": 6.38,
      "learning_rate": 3.6225823314166235e-06,
      "loss": 0.0,
      "step": 12200
    },
    {
      "epoch": 6.4,
      "learning_rate": 3.596445373758495e-06,
      "loss": 0.0,
      "step": 12250
    },
    {
      "epoch": 6.43,
      "learning_rate": 3.5703084161003658e-06,
      "loss": 0.0,
      "step": 12300
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.5441714584422375e-06,
      "loss": 0.0,
      "step": 12350
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.518034500784109e-06,
      "loss": 0.0008,
      "step": 12400
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.4918975431259805e-06,
      "loss": 0.0,
      "step": 12450
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.465760585467852e-06,
      "loss": 0.0,
      "step": 12500
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.439623627809723e-06,
      "loss": 0.0,
      "step": 12550
    },
    {
      "epoch": 6.59,
      "learning_rate": 3.4134866701515945e-06,
      "loss": 0.0,
      "step": 12600
    },
    {
      "epoch": 6.61,
      "learning_rate": 3.3873497124934662e-06,
      "loss": 0.0,
      "step": 12650
    },
    {
      "epoch": 6.64,
      "learning_rate": 3.3612127548353375e-06,
      "loss": 0.0,
      "step": 12700
    },
    {
      "epoch": 6.66,
      "learning_rate": 3.3350757971772093e-06,
      "loss": 0.0003,
      "step": 12750
    },
    {
      "epoch": 6.69,
      "learning_rate": 3.30893883951908e-06,
      "loss": 0.0,
      "step": 12800
    },
    {
      "epoch": 6.72,
      "learning_rate": 3.2828018818609515e-06,
      "loss": 0.0,
      "step": 12850
    },
    {
      "epoch": 6.74,
      "learning_rate": 3.2566649242028232e-06,
      "loss": 0.0,
      "step": 12900
    },
    {
      "epoch": 6.77,
      "learning_rate": 3.230527966544694e-06,
      "loss": 0.0,
      "step": 12950
    },
    {
      "epoch": 6.8,
      "learning_rate": 3.204391008886566e-06,
      "loss": 0.0,
      "step": 13000
    },
    {
      "epoch": 6.82,
      "learning_rate": 3.178254051228437e-06,
      "loss": 0.0,
      "step": 13050
    },
    {
      "epoch": 6.85,
      "learning_rate": 3.152117093570309e-06,
      "loss": 0.0,
      "step": 13100
    },
    {
      "epoch": 6.87,
      "learning_rate": 3.12598013591218e-06,
      "loss": 0.0,
      "step": 13150
    },
    {
      "epoch": 6.9,
      "learning_rate": 3.0998431782540515e-06,
      "loss": 0.0,
      "step": 13200
    },
    {
      "epoch": 6.93,
      "learning_rate": 3.073706220595923e-06,
      "loss": 0.0,
      "step": 13250
    },
    {
      "epoch": 6.95,
      "learning_rate": 3.0475692629377946e-06,
      "loss": 0.0,
      "step": 13300
    },
    {
      "epoch": 6.98,
      "learning_rate": 3.021432305279666e-06,
      "loss": 0.0,
      "step": 13350
    },
    {
      "epoch": 7.0,
      "eval_f1": 0.9999508752394832,
      "eval_loss": 0.0004308556381147355,
      "eval_runtime": 25.8464,
      "eval_samples_per_second": 1183.997,
      "eval_steps_per_second": 18.533,
      "step": 13391
    },
    {
      "epoch": 7.0,
      "learning_rate": 2.9952953476215368e-06,
      "loss": 0.0,
      "step": 13400
    },
    {
      "epoch": 7.03,
      "learning_rate": 2.9691583899634085e-06,
      "loss": 0.0,
      "step": 13450
    },
    {
      "epoch": 7.06,
      "learning_rate": 2.94302143230528e-06,
      "loss": 0.0,
      "step": 13500
    },
    {
      "epoch": 7.08,
      "learning_rate": 2.9168844746471516e-06,
      "loss": 0.0,
      "step": 13550
    },
    {
      "epoch": 7.11,
      "learning_rate": 2.8907475169890225e-06,
      "loss": 0.0,
      "step": 13600
    },
    {
      "epoch": 7.14,
      "learning_rate": 2.864610559330894e-06,
      "loss": 0.0,
      "step": 13650
    },
    {
      "epoch": 7.16,
      "learning_rate": 2.8384736016727655e-06,
      "loss": 0.0,
      "step": 13700
    },
    {
      "epoch": 7.19,
      "learning_rate": 2.8123366440146373e-06,
      "loss": 0.0,
      "step": 13750
    },
    {
      "epoch": 7.21,
      "learning_rate": 2.786199686356508e-06,
      "loss": 0.0,
      "step": 13800
    },
    {
      "epoch": 7.24,
      "learning_rate": 2.76006272869838e-06,
      "loss": 0.0,
      "step": 13850
    },
    {
      "epoch": 7.27,
      "learning_rate": 2.733925771040251e-06,
      "loss": 0.0,
      "step": 13900
    },
    {
      "epoch": 7.29,
      "learning_rate": 2.707788813382122e-06,
      "loss": 0.0,
      "step": 13950
    },
    {
      "epoch": 7.32,
      "learning_rate": 2.681651855723994e-06,
      "loss": 0.0,
      "step": 14000
    },
    {
      "epoch": 7.34,
      "learning_rate": 2.655514898065865e-06,
      "loss": 0.0,
      "step": 14050
    },
    {
      "epoch": 7.37,
      "learning_rate": 2.629377940407737e-06,
      "loss": 0.0,
      "step": 14100
    },
    {
      "epoch": 7.4,
      "learning_rate": 2.603240982749608e-06,
      "loss": 0.0,
      "step": 14150
    },
    {
      "epoch": 7.42,
      "learning_rate": 2.57710402509148e-06,
      "loss": 0.0,
      "step": 14200
    },
    {
      "epoch": 7.45,
      "learning_rate": 2.550967067433351e-06,
      "loss": 0.0,
      "step": 14250
    },
    {
      "epoch": 7.48,
      "learning_rate": 2.5248301097752226e-06,
      "loss": 0.0,
      "step": 14300
    },
    {
      "epoch": 7.5,
      "learning_rate": 2.498693152117094e-06,
      "loss": 0.0,
      "step": 14350
    },
    {
      "epoch": 7.53,
      "learning_rate": 2.472556194458965e-06,
      "loss": 0.0,
      "step": 14400
    },
    {
      "epoch": 7.55,
      "learning_rate": 2.4464192368008365e-06,
      "loss": 0.0,
      "step": 14450
    },
    {
      "epoch": 7.58,
      "learning_rate": 2.420282279142708e-06,
      "loss": 0.0002,
      "step": 14500
    },
    {
      "epoch": 7.61,
      "learning_rate": 2.3941453214845796e-06,
      "loss": 0.0,
      "step": 14550
    },
    {
      "epoch": 7.63,
      "learning_rate": 2.368008363826451e-06,
      "loss": 0.0,
      "step": 14600
    },
    {
      "epoch": 7.66,
      "learning_rate": 2.341871406168322e-06,
      "loss": 0.0,
      "step": 14650
    },
    {
      "epoch": 7.68,
      "learning_rate": 2.3157344485101935e-06,
      "loss": 0.0,
      "step": 14700
    },
    {
      "epoch": 7.71,
      "learning_rate": 2.289597490852065e-06,
      "loss": 0.0,
      "step": 14750
    },
    {
      "epoch": 7.74,
      "learning_rate": 2.263460533193936e-06,
      "loss": 0.0011,
      "step": 14800
    },
    {
      "epoch": 7.76,
      "learning_rate": 2.237323575535808e-06,
      "loss": 0.0,
      "step": 14850
    },
    {
      "epoch": 7.79,
      "learning_rate": 2.211186617877679e-06,
      "loss": 0.0005,
      "step": 14900
    },
    {
      "epoch": 7.81,
      "learning_rate": 2.1850496602195505e-06,
      "loss": 0.0,
      "step": 14950
    },
    {
      "epoch": 7.84,
      "learning_rate": 2.1589127025614222e-06,
      "loss": 0.0,
      "step": 15000
    },
    {
      "epoch": 7.87,
      "learning_rate": 2.1327757449032935e-06,
      "loss": 0.0,
      "step": 15050
    },
    {
      "epoch": 7.89,
      "learning_rate": 2.106638787245165e-06,
      "loss": 0.0,
      "step": 15100
    },
    {
      "epoch": 7.92,
      "learning_rate": 2.080501829587036e-06,
      "loss": 0.0,
      "step": 15150
    },
    {
      "epoch": 7.95,
      "learning_rate": 2.054364871928908e-06,
      "loss": 0.0,
      "step": 15200
    },
    {
      "epoch": 7.97,
      "learning_rate": 2.028227914270779e-06,
      "loss": 0.0,
      "step": 15250
    },
    {
      "epoch": 8.0,
      "learning_rate": 2.0020909566126505e-06,
      "loss": 0.0,
      "step": 15300
    },
    {
      "epoch": 8.0,
      "eval_f1": 0.9999836256160861,
      "eval_loss": 0.0003341793781146407,
      "eval_runtime": 25.8188,
      "eval_samples_per_second": 1185.259,
      "eval_steps_per_second": 18.552,
      "step": 15304
    }
  ],
  "logging_steps": 50,
  "max_steps": 19130,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1.4404562222727168e+16,
  "trial_name": null,
  "trial_params": null
}