private / trainer_state.json
Pixedar's picture
Upload 11 files
a9df6e3 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.916396629941672,
"eval_steps": 500,
"global_step": 18000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 3.3044,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 2.9614,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 2.8054,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 2.5516,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 2.8391,
"step": 125
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 2.4576,
"step": 150
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 2.6535,
"step": 175
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 2.4572,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 2.6712,
"step": 225
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 2.3453,
"step": 250
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 2.5588,
"step": 275
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 2.441,
"step": 300
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 2.6621,
"step": 325
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 2.5328,
"step": 350
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 2.6159,
"step": 375
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 2.4162,
"step": 400
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 2.4839,
"step": 425
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 2.3451,
"step": 450
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 2.5323,
"step": 475
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 2.433,
"step": 500
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 2.5864,
"step": 525
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 2.3248,
"step": 550
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 2.6075,
"step": 575
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 2.3894,
"step": 600
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 2.6016,
"step": 625
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 2.3992,
"step": 650
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 2.4086,
"step": 675
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 2.303,
"step": 700
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 2.4355,
"step": 725
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 2.3693,
"step": 750
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 2.4441,
"step": 775
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 2.3237,
"step": 800
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 2.4797,
"step": 825
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 2.3291,
"step": 850
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 2.3765,
"step": 875
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 2.3485,
"step": 900
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 2.3761,
"step": 925
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 2.1423,
"step": 950
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 2.5231,
"step": 975
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 2.397,
"step": 1000
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 2.384,
"step": 1025
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 2.2133,
"step": 1050
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 2.4448,
"step": 1075
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 2.2569,
"step": 1100
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 2.4679,
"step": 1125
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 2.2591,
"step": 1150
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 2.4579,
"step": 1175
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 2.1727,
"step": 1200
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 2.4947,
"step": 1225
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 2.2964,
"step": 1250
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 2.4288,
"step": 1275
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 2.2271,
"step": 1300
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 2.3102,
"step": 1325
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 2.2645,
"step": 1350
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 2.4434,
"step": 1375
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 2.3105,
"step": 1400
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 2.3998,
"step": 1425
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 2.2101,
"step": 1450
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 2.384,
"step": 1475
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 2.3274,
"step": 1500
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 2.5178,
"step": 1525
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 2.2085,
"step": 1550
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 2.3946,
"step": 1575
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 2.233,
"step": 1600
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 2.4607,
"step": 1625
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 2.3017,
"step": 1650
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 2.5035,
"step": 1675
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 2.2364,
"step": 1700
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 2.3528,
"step": 1725
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 2.2548,
"step": 1750
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 2.48,
"step": 1775
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 2.2583,
"step": 1800
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 2.3577,
"step": 1825
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 2.3183,
"step": 1850
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 2.4696,
"step": 1875
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 2.2888,
"step": 1900
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 2.4176,
"step": 1925
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 2.2553,
"step": 1950
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 2.4154,
"step": 1975
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 2.1653,
"step": 2000
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 2.5119,
"step": 2025
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 2.244,
"step": 2050
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 2.4014,
"step": 2075
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 2.2161,
"step": 2100
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 2.3874,
"step": 2125
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 2.1541,
"step": 2150
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 2.3145,
"step": 2175
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 2.0975,
"step": 2200
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 2.3086,
"step": 2225
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 2.2297,
"step": 2250
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 2.3859,
"step": 2275
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.9693,
"step": 2300
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 2.35,
"step": 2325
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 2.0767,
"step": 2350
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 2.4507,
"step": 2375
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 2.1896,
"step": 2400
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 2.4073,
"step": 2425
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 2.2483,
"step": 2450
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 2.4123,
"step": 2475
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 2.0906,
"step": 2500
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 2.3711,
"step": 2525
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 2.2335,
"step": 2550
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 2.3186,
"step": 2575
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 2.1362,
"step": 2600
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 2.355,
"step": 2625
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 2.1693,
"step": 2650
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 2.4107,
"step": 2675
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 2.2098,
"step": 2700
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 2.2907,
"step": 2725
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 2.2335,
"step": 2750
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 2.3959,
"step": 2775
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 2.1394,
"step": 2800
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 2.4236,
"step": 2825
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 2.1658,
"step": 2850
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 2.2197,
"step": 2875
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 2.101,
"step": 2900
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 2.3416,
"step": 2925
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 2.0611,
"step": 2950
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 2.3655,
"step": 2975
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 2.1746,
"step": 3000
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 2.3561,
"step": 3025
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 2.0445,
"step": 3050
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 2.4062,
"step": 3075
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 2.2256,
"step": 3100
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 2.3861,
"step": 3125
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 2.1233,
"step": 3150
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 2.3603,
"step": 3175
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 2.1118,
"step": 3200
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 2.3485,
"step": 3225
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 2.0191,
"step": 3250
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 2.2822,
"step": 3275
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 2.2297,
"step": 3300
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 2.375,
"step": 3325
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 2.1859,
"step": 3350
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 2.3411,
"step": 3375
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 2.0084,
"step": 3400
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 2.2536,
"step": 3425
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 2.1425,
"step": 3450
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 2.327,
"step": 3475
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 2.1356,
"step": 3500
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 2.2898,
"step": 3525
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 2.2311,
"step": 3550
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 2.3137,
"step": 3575
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 2.1663,
"step": 3600
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 2.3658,
"step": 3625
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 2.2129,
"step": 3650
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 2.4101,
"step": 3675
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.9538,
"step": 3700
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 2.4423,
"step": 3725
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 2.1344,
"step": 3750
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 2.3684,
"step": 3775
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 2.0626,
"step": 3800
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 2.3094,
"step": 3825
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.9973,
"step": 3850
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 2.3486,
"step": 3875
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 2.1224,
"step": 3900
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 2.2489,
"step": 3925
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 2.0245,
"step": 3950
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 2.3817,
"step": 3975
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 2.1786,
"step": 4000
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 2.3312,
"step": 4025
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 2.0933,
"step": 4050
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 2.4028,
"step": 4075
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 2.0841,
"step": 4100
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.2924,
"step": 4125
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 2.0088,
"step": 4150
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 2.3636,
"step": 4175
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 2.1101,
"step": 4200
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 2.3225,
"step": 4225
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 2.1775,
"step": 4250
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 2.3995,
"step": 4275
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 2.0986,
"step": 4300
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 2.3905,
"step": 4325
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 2.0652,
"step": 4350
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.421,
"step": 4375
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 2.0402,
"step": 4400
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 2.3154,
"step": 4425
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 2.1875,
"step": 4450
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 2.3481,
"step": 4475
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 1.9964,
"step": 4500
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 2.2526,
"step": 4525
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 2.1539,
"step": 4550
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 2.2843,
"step": 4575
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.1451,
"step": 4600
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 2.2955,
"step": 4625
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 1.9855,
"step": 4650
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 2.4539,
"step": 4675
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 2.081,
"step": 4700
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 2.3724,
"step": 4725
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 1.9166,
"step": 4750
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 2.2773,
"step": 4775
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 2.0122,
"step": 4800
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 2.2272,
"step": 4825
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.079,
"step": 4850
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.4039,
"step": 4875
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 2.0559,
"step": 4900
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.3187,
"step": 4925
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 2.1956,
"step": 4950
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.2611,
"step": 4975
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.1821,
"step": 5000
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 2.2348,
"step": 5025
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 1.9541,
"step": 5050
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 2.381,
"step": 5075
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.0978,
"step": 5100
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.3287,
"step": 5125
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 2.1468,
"step": 5150
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.3461,
"step": 5175
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 2.1531,
"step": 5200
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.3668,
"step": 5225
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.0828,
"step": 5250
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 2.3509,
"step": 5275
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.0817,
"step": 5300
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 2.3609,
"step": 5325
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 2.1038,
"step": 5350
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 2.1818,
"step": 5375
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 2.0505,
"step": 5400
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 2.224,
"step": 5425
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 2.1014,
"step": 5450
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 2.2959,
"step": 5475
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 2.0824,
"step": 5500
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 2.2384,
"step": 5525
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 2.0479,
"step": 5550
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 2.4386,
"step": 5575
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 2.0333,
"step": 5600
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 2.2645,
"step": 5625
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 2.0613,
"step": 5650
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 2.2632,
"step": 5675
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 2.046,
"step": 5700
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 2.2734,
"step": 5725
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 2.1293,
"step": 5750
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 2.3176,
"step": 5775
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 2.0572,
"step": 5800
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 2.3778,
"step": 5825
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 2.1428,
"step": 5850
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 2.2676,
"step": 5875
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 2.1128,
"step": 5900
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 2.2816,
"step": 5925
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 2.1031,
"step": 5950
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 2.2525,
"step": 5975
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 2.0176,
"step": 6000
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 2.3356,
"step": 6025
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 2.0888,
"step": 6050
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 2.2696,
"step": 6075
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 2.0839,
"step": 6100
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 2.1857,
"step": 6125
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 1.9069,
"step": 6150
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 2.1828,
"step": 6175
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 2.2461,
"step": 6200
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 2.0132,
"step": 6225
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 2.1801,
"step": 6250
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 1.951,
"step": 6275
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 2.2283,
"step": 6300
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 2.0281,
"step": 6325
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 2.1224,
"step": 6350
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 2.0129,
"step": 6375
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 2.1152,
"step": 6400
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 1.9273,
"step": 6425
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 2.1951,
"step": 6450
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 2.0412,
"step": 6475
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 2.2057,
"step": 6500
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 1.981,
"step": 6525
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 2.1747,
"step": 6550
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 2.1069,
"step": 6575
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 2.0922,
"step": 6600
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 2.0494,
"step": 6625
},
{
"epoch": 1.08,
"learning_rate": 0.0002,
"loss": 2.2018,
"step": 6650
},
{
"epoch": 1.08,
"learning_rate": 0.0002,
"loss": 1.9717,
"step": 6675
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 2.1925,
"step": 6700
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 1.9474,
"step": 6725
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 2.144,
"step": 6750
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 1.9665,
"step": 6775
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 2.1815,
"step": 6800
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 1.9634,
"step": 6825
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 2.1305,
"step": 6850
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 2.0749,
"step": 6875
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 2.1224,
"step": 6900
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 1.9251,
"step": 6925
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 2.129,
"step": 6950
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 2.0075,
"step": 6975
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 2.1379,
"step": 7000
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 1.933,
"step": 7025
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 2.1166,
"step": 7050
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 2.0678,
"step": 7075
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 2.27,
"step": 7100
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 2.0561,
"step": 7125
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 2.1881,
"step": 7150
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 2.09,
"step": 7175
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 2.1593,
"step": 7200
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 2.006,
"step": 7225
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 2.2042,
"step": 7250
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 2.0281,
"step": 7275
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 2.1054,
"step": 7300
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 1.8642,
"step": 7325
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 2.2682,
"step": 7350
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 1.9484,
"step": 7375
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 2.2109,
"step": 7400
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 2.0173,
"step": 7425
},
{
"epoch": 1.21,
"learning_rate": 0.0002,
"loss": 2.2798,
"step": 7450
},
{
"epoch": 1.21,
"learning_rate": 0.0002,
"loss": 1.994,
"step": 7475
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 2.2614,
"step": 7500
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 1.9984,
"step": 7525
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 2.2028,
"step": 7550
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 2.0623,
"step": 7575
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 2.1407,
"step": 7600
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 2.033,
"step": 7625
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 2.1987,
"step": 7650
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 1.9441,
"step": 7675
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 2.2531,
"step": 7700
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 2.0297,
"step": 7725
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 2.2639,
"step": 7750
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 2.0911,
"step": 7775
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 2.2469,
"step": 7800
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 1.9717,
"step": 7825
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 2.1949,
"step": 7850
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 1.9858,
"step": 7875
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 2.1441,
"step": 7900
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 1.9925,
"step": 7925
},
{
"epoch": 1.29,
"learning_rate": 0.0002,
"loss": 2.1327,
"step": 7950
},
{
"epoch": 1.29,
"learning_rate": 0.0002,
"loss": 1.8171,
"step": 7975
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 2.1612,
"step": 8000
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 1.982,
"step": 8025
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 2.1808,
"step": 8050
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 1.936,
"step": 8075
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 2.1597,
"step": 8100
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 1.8461,
"step": 8125
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 2.1309,
"step": 8150
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 2.1594,
"step": 8175
},
{
"epoch": 1.33,
"learning_rate": 0.0002,
"loss": 2.1131,
"step": 8200
},
{
"epoch": 1.33,
"learning_rate": 0.0002,
"loss": 1.8944,
"step": 8225
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 2.0845,
"step": 8250
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 2.0502,
"step": 8275
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 2.1475,
"step": 8300
},
{
"epoch": 1.35,
"learning_rate": 0.0002,
"loss": 1.9887,
"step": 8325
},
{
"epoch": 1.35,
"learning_rate": 0.0002,
"loss": 2.0891,
"step": 8350
},
{
"epoch": 1.36,
"learning_rate": 0.0002,
"loss": 1.9402,
"step": 8375
},
{
"epoch": 1.36,
"learning_rate": 0.0002,
"loss": 2.1112,
"step": 8400
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 2.061,
"step": 8425
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 2.2356,
"step": 8450
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 1.9425,
"step": 8475
},
{
"epoch": 1.38,
"learning_rate": 0.0002,
"loss": 2.2236,
"step": 8500
},
{
"epoch": 1.38,
"learning_rate": 0.0002,
"loss": 1.84,
"step": 8525
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 2.1553,
"step": 8550
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 1.8677,
"step": 8575
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 2.2828,
"step": 8600
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 1.907,
"step": 8625
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 2.1928,
"step": 8650
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 1.9845,
"step": 8675
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 2.1495,
"step": 8700
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 1.9496,
"step": 8725
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 2.2124,
"step": 8750
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 2.0276,
"step": 8775
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 2.1716,
"step": 8800
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 2.0762,
"step": 8825
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 2.1875,
"step": 8850
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 1.9567,
"step": 8875
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 2.1331,
"step": 8900
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 1.9842,
"step": 8925
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 2.1453,
"step": 8950
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 2.0681,
"step": 8975
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 2.1777,
"step": 9000
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 1.9691,
"step": 9025
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 2.195,
"step": 9050
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 1.9626,
"step": 9075
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 2.2002,
"step": 9100
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 1.9548,
"step": 9125
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 2.0627,
"step": 9150
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 1.8949,
"step": 9175
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 2.0806,
"step": 9200
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 1.945,
"step": 9225
},
{
"epoch": 1.5,
"learning_rate": 0.0002,
"loss": 2.2907,
"step": 9250
},
{
"epoch": 1.5,
"learning_rate": 0.0002,
"loss": 1.9685,
"step": 9275
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 2.1251,
"step": 9300
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 1.9426,
"step": 9325
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 2.1402,
"step": 9350
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 1.9268,
"step": 9375
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 2.1373,
"step": 9400
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 1.9606,
"step": 9425
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 2.2019,
"step": 9450
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 1.9014,
"step": 9475
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 2.071,
"step": 9500
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 1.9199,
"step": 9525
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 2.224,
"step": 9550
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 2.0849,
"step": 9575
},
{
"epoch": 1.56,
"learning_rate": 0.0002,
"loss": 2.0944,
"step": 9600
},
{
"epoch": 1.56,
"learning_rate": 0.0002,
"loss": 1.9544,
"step": 9625
},
{
"epoch": 1.56,
"learning_rate": 0.0002,
"loss": 2.24,
"step": 9650
},
{
"epoch": 1.57,
"learning_rate": 0.0002,
"loss": 2.0462,
"step": 9675
},
{
"epoch": 1.57,
"learning_rate": 0.0002,
"loss": 2.09,
"step": 9700
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 2.1015,
"step": 9725
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 2.205,
"step": 9750
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 1.9051,
"step": 9775
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 2.1992,
"step": 9800
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 1.9637,
"step": 9825
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 2.1629,
"step": 9850
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 1.9888,
"step": 9875
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 2.0987,
"step": 9900
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 1.982,
"step": 9925
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 2.0717,
"step": 9950
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 1.9921,
"step": 9975
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 2.1405,
"step": 10000
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 1.8981,
"step": 10025
},
{
"epoch": 1.63,
"learning_rate": 0.0002,
"loss": 2.1338,
"step": 10050
},
{
"epoch": 1.63,
"learning_rate": 0.0002,
"loss": 1.8935,
"step": 10075
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 2.12,
"step": 10100
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 1.9227,
"step": 10125
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 2.2621,
"step": 10150
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 2.0911,
"step": 10175
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 2.1278,
"step": 10200
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 2.0571,
"step": 10225
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 2.1462,
"step": 10250
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 1.9058,
"step": 10275
},
{
"epoch": 1.67,
"learning_rate": 0.0002,
"loss": 2.1154,
"step": 10300
},
{
"epoch": 1.67,
"learning_rate": 0.0002,
"loss": 1.9732,
"step": 10325
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 2.2269,
"step": 10350
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 1.976,
"step": 10375
},
{
"epoch": 1.69,
"learning_rate": 0.0002,
"loss": 2.2662,
"step": 10400
},
{
"epoch": 1.69,
"learning_rate": 0.0002,
"loss": 1.9424,
"step": 10425
},
{
"epoch": 1.69,
"learning_rate": 0.0002,
"loss": 2.1602,
"step": 10450
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 1.9668,
"step": 10475
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 2.1355,
"step": 10500
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 2.0549,
"step": 10525
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 2.1387,
"step": 10550
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 2.0052,
"step": 10575
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 2.142,
"step": 10600
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 1.9063,
"step": 10625
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 2.1204,
"step": 10650
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 1.8533,
"step": 10675
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 2.0855,
"step": 10700
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 2.0132,
"step": 10725
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 2.1836,
"step": 10750
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 2.0041,
"step": 10775
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 2.1063,
"step": 10800
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 1.9192,
"step": 10825
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 2.2027,
"step": 10850
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 1.8827,
"step": 10875
},
{
"epoch": 1.77,
"learning_rate": 0.0002,
"loss": 2.1606,
"step": 10900
},
{
"epoch": 1.77,
"learning_rate": 0.0002,
"loss": 2.0008,
"step": 10925
},
{
"epoch": 1.77,
"learning_rate": 0.0002,
"loss": 2.1251,
"step": 10950
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 1.9729,
"step": 10975
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 2.2528,
"step": 11000
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 1.9996,
"step": 11025
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 2.2393,
"step": 11050
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 1.9691,
"step": 11075
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 2.0502,
"step": 11100
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 1.9065,
"step": 11125
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 2.1418,
"step": 11150
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 1.964,
"step": 11175
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 2.1955,
"step": 11200
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 1.9258,
"step": 11225
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 2.1197,
"step": 11250
},
{
"epoch": 1.83,
"learning_rate": 0.0002,
"loss": 1.9796,
"step": 11275
},
{
"epoch": 1.83,
"learning_rate": 0.0002,
"loss": 2.1744,
"step": 11300
},
{
"epoch": 1.83,
"learning_rate": 0.0002,
"loss": 2.0322,
"step": 11325
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 2.1835,
"step": 11350
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 1.8594,
"step": 11375
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 2.1748,
"step": 11400
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 1.9016,
"step": 11425
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 2.1567,
"step": 11450
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 1.9872,
"step": 11475
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 2.1728,
"step": 11500
},
{
"epoch": 1.87,
"learning_rate": 0.0002,
"loss": 1.9501,
"step": 11525
},
{
"epoch": 1.87,
"learning_rate": 0.0002,
"loss": 2.1301,
"step": 11550
},
{
"epoch": 1.88,
"learning_rate": 0.0002,
"loss": 1.9502,
"step": 11575
},
{
"epoch": 1.88,
"learning_rate": 0.0002,
"loss": 2.2177,
"step": 11600
},
{
"epoch": 1.88,
"learning_rate": 0.0002,
"loss": 1.8511,
"step": 11625
},
{
"epoch": 1.89,
"learning_rate": 0.0002,
"loss": 2.1206,
"step": 11650
},
{
"epoch": 1.89,
"learning_rate": 0.0002,
"loss": 1.9794,
"step": 11675
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 2.1558,
"step": 11700
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 2.0251,
"step": 11725
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 2.0404,
"step": 11750
},
{
"epoch": 1.91,
"learning_rate": 0.0002,
"loss": 1.98,
"step": 11775
},
{
"epoch": 1.91,
"learning_rate": 0.0002,
"loss": 2.1388,
"step": 11800
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 1.9228,
"step": 11825
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 2.101,
"step": 11850
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 2.0433,
"step": 11875
},
{
"epoch": 1.93,
"learning_rate": 0.0002,
"loss": 2.0715,
"step": 11900
},
{
"epoch": 1.93,
"learning_rate": 0.0002,
"loss": 2.0914,
"step": 11925
},
{
"epoch": 1.94,
"learning_rate": 0.0002,
"loss": 2.1653,
"step": 11950
},
{
"epoch": 1.94,
"learning_rate": 0.0002,
"loss": 1.868,
"step": 11975
},
{
"epoch": 1.94,
"learning_rate": 0.0002,
"loss": 2.1383,
"step": 12000
},
{
"epoch": 1.95,
"learning_rate": 0.0002,
"loss": 2.046,
"step": 12025
},
{
"epoch": 1.95,
"learning_rate": 0.0002,
"loss": 2.1616,
"step": 12050
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 1.9724,
"step": 12075
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 2.119,
"step": 12100
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 2.0735,
"step": 12125
},
{
"epoch": 1.97,
"learning_rate": 0.0002,
"loss": 2.0902,
"step": 12150
},
{
"epoch": 1.97,
"learning_rate": 0.0002,
"loss": 1.8635,
"step": 12175
},
{
"epoch": 1.98,
"learning_rate": 0.0002,
"loss": 2.1361,
"step": 12200
},
{
"epoch": 1.98,
"learning_rate": 0.0002,
"loss": 2.0382,
"step": 12225
},
{
"epoch": 1.98,
"learning_rate": 0.0002,
"loss": 2.1705,
"step": 12250
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 1.9679,
"step": 12275
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 2.1322,
"step": 12300
},
{
"epoch": 2.0,
"learning_rate": 0.0002,
"loss": 1.951,
"step": 12325
},
{
"epoch": 2.0,
"learning_rate": 0.0002,
"loss": 2.0835,
"step": 12350
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 1.966,
"step": 12375
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 1.8028,
"step": 12400
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 2.0308,
"step": 12425
},
{
"epoch": 2.02,
"learning_rate": 0.0002,
"loss": 1.8125,
"step": 12450
},
{
"epoch": 2.02,
"learning_rate": 0.0002,
"loss": 2.1039,
"step": 12475
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 1.8705,
"step": 12500
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 1.9316,
"step": 12525
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 1.8444,
"step": 12550
},
{
"epoch": 2.04,
"learning_rate": 0.0002,
"loss": 2.036,
"step": 12575
},
{
"epoch": 2.04,
"learning_rate": 0.0002,
"loss": 1.7983,
"step": 12600
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 1.9918,
"step": 12625
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 1.8069,
"step": 12650
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 2.0905,
"step": 12675
},
{
"epoch": 2.06,
"learning_rate": 0.0002,
"loss": 1.7724,
"step": 12700
},
{
"epoch": 2.06,
"learning_rate": 0.0002,
"loss": 1.9975,
"step": 12725
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 1.8409,
"step": 12750
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 1.9692,
"step": 12775
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 1.7994,
"step": 12800
},
{
"epoch": 2.08,
"learning_rate": 0.0002,
"loss": 1.9763,
"step": 12825
},
{
"epoch": 2.08,
"learning_rate": 0.0002,
"loss": 1.8558,
"step": 12850
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 1.9636,
"step": 12875
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 1.8986,
"step": 12900
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 1.9416,
"step": 12925
},
{
"epoch": 2.1,
"learning_rate": 0.0002,
"loss": 1.8119,
"step": 12950
},
{
"epoch": 2.1,
"learning_rate": 0.0002,
"loss": 2.0878,
"step": 12975
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 1.8937,
"step": 13000
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 2.0047,
"step": 13025
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 1.8724,
"step": 13050
},
{
"epoch": 2.12,
"learning_rate": 0.0002,
"loss": 2.0153,
"step": 13075
},
{
"epoch": 2.12,
"learning_rate": 0.0002,
"loss": 1.8569,
"step": 13100
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 1.9678,
"step": 13125
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 1.8821,
"step": 13150
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 2.0174,
"step": 13175
},
{
"epoch": 2.14,
"learning_rate": 0.0002,
"loss": 1.9131,
"step": 13200
},
{
"epoch": 2.14,
"learning_rate": 0.0002,
"loss": 2.007,
"step": 13225
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 1.8461,
"step": 13250
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 1.9892,
"step": 13275
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 1.9311,
"step": 13300
},
{
"epoch": 2.16,
"learning_rate": 0.0002,
"loss": 1.9571,
"step": 13325
},
{
"epoch": 2.16,
"learning_rate": 0.0002,
"loss": 1.8195,
"step": 13350
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 1.9952,
"step": 13375
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 1.9007,
"step": 13400
},
{
"epoch": 2.18,
"learning_rate": 0.0002,
"loss": 1.9624,
"step": 13425
},
{
"epoch": 2.18,
"learning_rate": 0.0002,
"loss": 1.9803,
"step": 13450
},
{
"epoch": 2.18,
"learning_rate": 0.0002,
"loss": 1.976,
"step": 13475
},
{
"epoch": 2.19,
"learning_rate": 0.0002,
"loss": 1.9023,
"step": 13500
},
{
"epoch": 2.19,
"learning_rate": 0.0002,
"loss": 1.9851,
"step": 13525
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 1.8301,
"step": 13550
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 2.0179,
"step": 13575
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 1.9377,
"step": 13600
},
{
"epoch": 2.21,
"learning_rate": 0.0002,
"loss": 2.0117,
"step": 13625
},
{
"epoch": 2.21,
"learning_rate": 0.0002,
"loss": 1.8109,
"step": 13650
},
{
"epoch": 2.22,
"learning_rate": 0.0002,
"loss": 2.0088,
"step": 13675
},
{
"epoch": 2.22,
"learning_rate": 0.0002,
"loss": 1.8731,
"step": 13700
},
{
"epoch": 2.22,
"learning_rate": 0.0002,
"loss": 1.9932,
"step": 13725
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 1.9334,
"step": 13750
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 2.0354,
"step": 13775
},
{
"epoch": 2.24,
"learning_rate": 0.0002,
"loss": 1.8924,
"step": 13800
},
{
"epoch": 2.24,
"learning_rate": 0.0002,
"loss": 2.0722,
"step": 13825
},
{
"epoch": 2.24,
"learning_rate": 0.0002,
"loss": 1.9154,
"step": 13850
},
{
"epoch": 2.25,
"learning_rate": 0.0002,
"loss": 2.0131,
"step": 13875
},
{
"epoch": 2.25,
"learning_rate": 0.0002,
"loss": 1.874,
"step": 13900
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 2.0444,
"step": 13925
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 1.8059,
"step": 13950
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 1.9809,
"step": 13975
},
{
"epoch": 2.27,
"learning_rate": 0.0002,
"loss": 2.0035,
"step": 14000
},
{
"epoch": 2.27,
"learning_rate": 0.0002,
"loss": 1.9896,
"step": 14025
},
{
"epoch": 2.28,
"learning_rate": 0.0002,
"loss": 1.796,
"step": 14050
},
{
"epoch": 2.28,
"learning_rate": 0.0002,
"loss": 2.0247,
"step": 14075
},
{
"epoch": 2.28,
"learning_rate": 0.0002,
"loss": 1.8983,
"step": 14100
},
{
"epoch": 2.29,
"learning_rate": 0.0002,
"loss": 1.9623,
"step": 14125
},
{
"epoch": 2.29,
"learning_rate": 0.0002,
"loss": 1.8984,
"step": 14150
},
{
"epoch": 2.3,
"learning_rate": 0.0002,
"loss": 2.0151,
"step": 14175
},
{
"epoch": 2.3,
"learning_rate": 0.0002,
"loss": 1.8974,
"step": 14200
},
{
"epoch": 2.3,
"learning_rate": 0.0002,
"loss": 2.0638,
"step": 14225
},
{
"epoch": 2.31,
"learning_rate": 0.0002,
"loss": 1.8475,
"step": 14250
},
{
"epoch": 2.31,
"learning_rate": 0.0002,
"loss": 2.0365,
"step": 14275
},
{
"epoch": 2.32,
"learning_rate": 0.0002,
"loss": 1.8248,
"step": 14300
},
{
"epoch": 2.32,
"learning_rate": 0.0002,
"loss": 2.0305,
"step": 14325
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 1.8457,
"step": 14350
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 2.007,
"step": 14375
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 1.9849,
"step": 14400
},
{
"epoch": 2.34,
"learning_rate": 0.0002,
"loss": 2.0349,
"step": 14425
},
{
"epoch": 2.34,
"learning_rate": 0.0002,
"loss": 1.798,
"step": 14450
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 1.9238,
"step": 14475
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 1.8017,
"step": 14500
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 2.154,
"step": 14525
},
{
"epoch": 2.36,
"learning_rate": 0.0002,
"loss": 1.8748,
"step": 14550
},
{
"epoch": 2.36,
"learning_rate": 0.0002,
"loss": 1.9396,
"step": 14575
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 1.8473,
"step": 14600
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 2.0343,
"step": 14625
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 1.8666,
"step": 14650
},
{
"epoch": 2.38,
"learning_rate": 0.0002,
"loss": 2.147,
"step": 14675
},
{
"epoch": 2.38,
"learning_rate": 0.0002,
"loss": 1.8163,
"step": 14700
},
{
"epoch": 2.39,
"learning_rate": 0.0002,
"loss": 1.9378,
"step": 14725
},
{
"epoch": 2.39,
"learning_rate": 0.0002,
"loss": 1.9602,
"step": 14750
},
{
"epoch": 2.39,
"learning_rate": 0.0002,
"loss": 1.9138,
"step": 14775
},
{
"epoch": 2.4,
"learning_rate": 0.0002,
"loss": 1.9567,
"step": 14800
},
{
"epoch": 2.4,
"learning_rate": 0.0002,
"loss": 1.9788,
"step": 14825
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 1.8203,
"step": 14850
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 1.9832,
"step": 14875
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 1.8244,
"step": 14900
},
{
"epoch": 2.42,
"learning_rate": 0.0002,
"loss": 1.9275,
"step": 14925
},
{
"epoch": 2.42,
"learning_rate": 0.0002,
"loss": 1.8399,
"step": 14950
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 2.0371,
"step": 14975
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 1.8246,
"step": 15000
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 2.0811,
"step": 15025
},
{
"epoch": 2.44,
"learning_rate": 0.0002,
"loss": 1.8986,
"step": 15050
},
{
"epoch": 2.44,
"learning_rate": 0.0002,
"loss": 1.9898,
"step": 15075
},
{
"epoch": 2.45,
"learning_rate": 0.0002,
"loss": 1.888,
"step": 15100
},
{
"epoch": 2.45,
"learning_rate": 0.0002,
"loss": 2.0787,
"step": 15125
},
{
"epoch": 2.45,
"learning_rate": 0.0002,
"loss": 1.8367,
"step": 15150
},
{
"epoch": 2.46,
"learning_rate": 0.0002,
"loss": 1.8602,
"step": 15175
},
{
"epoch": 2.46,
"learning_rate": 0.0002,
"loss": 1.8423,
"step": 15200
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 1.9676,
"step": 15225
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 1.9784,
"step": 15250
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 1.9939,
"step": 15275
},
{
"epoch": 2.48,
"learning_rate": 0.0002,
"loss": 1.8257,
"step": 15300
},
{
"epoch": 2.48,
"learning_rate": 0.0002,
"loss": 1.9068,
"step": 15325
},
{
"epoch": 2.49,
"learning_rate": 0.0002,
"loss": 1.7919,
"step": 15350
},
{
"epoch": 2.49,
"learning_rate": 0.0002,
"loss": 1.9515,
"step": 15375
},
{
"epoch": 2.5,
"learning_rate": 0.0002,
"loss": 1.8452,
"step": 15400
},
{
"epoch": 2.5,
"learning_rate": 0.0002,
"loss": 1.9245,
"step": 15425
},
{
"epoch": 2.5,
"learning_rate": 0.0002,
"loss": 1.9447,
"step": 15450
},
{
"epoch": 2.51,
"learning_rate": 0.0002,
"loss": 1.9471,
"step": 15475
},
{
"epoch": 2.51,
"learning_rate": 0.0002,
"loss": 1.7937,
"step": 15500
},
{
"epoch": 2.52,
"learning_rate": 0.0002,
"loss": 2.03,
"step": 15525
},
{
"epoch": 2.52,
"learning_rate": 0.0002,
"loss": 1.92,
"step": 15550
},
{
"epoch": 2.52,
"learning_rate": 0.0002,
"loss": 2.0764,
"step": 15575
},
{
"epoch": 2.53,
"learning_rate": 0.0002,
"loss": 1.9053,
"step": 15600
},
{
"epoch": 2.53,
"learning_rate": 0.0002,
"loss": 1.9965,
"step": 15625
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 1.8693,
"step": 15650
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 1.9852,
"step": 15675
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 2.0139,
"step": 15700
},
{
"epoch": 2.55,
"learning_rate": 0.0002,
"loss": 2.058,
"step": 15725
},
{
"epoch": 2.55,
"learning_rate": 0.0002,
"loss": 1.8563,
"step": 15750
},
{
"epoch": 2.56,
"learning_rate": 0.0002,
"loss": 2.0112,
"step": 15775
},
{
"epoch": 2.56,
"learning_rate": 0.0002,
"loss": 1.7621,
"step": 15800
},
{
"epoch": 2.56,
"learning_rate": 0.0002,
"loss": 1.9343,
"step": 15825
},
{
"epoch": 2.57,
"learning_rate": 0.0002,
"loss": 1.8864,
"step": 15850
},
{
"epoch": 2.57,
"learning_rate": 0.0002,
"loss": 2.0502,
"step": 15875
},
{
"epoch": 2.58,
"learning_rate": 0.0002,
"loss": 1.7788,
"step": 15900
},
{
"epoch": 2.58,
"learning_rate": 0.0002,
"loss": 1.992,
"step": 15925
},
{
"epoch": 2.58,
"learning_rate": 0.0002,
"loss": 1.8523,
"step": 15950
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 2.0512,
"step": 15975
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 1.8112,
"step": 16000
},
{
"epoch": 2.6,
"learning_rate": 0.0002,
"loss": 2.02,
"step": 16025
},
{
"epoch": 2.6,
"learning_rate": 0.0002,
"loss": 1.971,
"step": 16050
},
{
"epoch": 2.6,
"learning_rate": 0.0002,
"loss": 1.9678,
"step": 16075
},
{
"epoch": 2.61,
"learning_rate": 0.0002,
"loss": 1.8012,
"step": 16100
},
{
"epoch": 2.61,
"learning_rate": 0.0002,
"loss": 2.0761,
"step": 16125
},
{
"epoch": 2.62,
"learning_rate": 0.0002,
"loss": 1.8307,
"step": 16150
},
{
"epoch": 2.62,
"learning_rate": 0.0002,
"loss": 2.0867,
"step": 16175
},
{
"epoch": 2.62,
"learning_rate": 0.0002,
"loss": 1.7947,
"step": 16200
},
{
"epoch": 2.63,
"learning_rate": 0.0002,
"loss": 2.011,
"step": 16225
},
{
"epoch": 2.63,
"learning_rate": 0.0002,
"loss": 1.8579,
"step": 16250
},
{
"epoch": 2.64,
"learning_rate": 0.0002,
"loss": 2.0436,
"step": 16275
},
{
"epoch": 2.64,
"learning_rate": 0.0002,
"loss": 1.8294,
"step": 16300
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 2.0266,
"step": 16325
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 1.7949,
"step": 16350
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 2.0295,
"step": 16375
},
{
"epoch": 2.66,
"learning_rate": 0.0002,
"loss": 1.7991,
"step": 16400
},
{
"epoch": 2.66,
"learning_rate": 0.0002,
"loss": 2.0991,
"step": 16425
},
{
"epoch": 2.67,
"learning_rate": 0.0002,
"loss": 1.868,
"step": 16450
},
{
"epoch": 2.67,
"learning_rate": 0.0002,
"loss": 1.9726,
"step": 16475
},
{
"epoch": 2.67,
"learning_rate": 0.0002,
"loss": 1.8131,
"step": 16500
},
{
"epoch": 2.68,
"learning_rate": 0.0002,
"loss": 2.0037,
"step": 16525
},
{
"epoch": 2.68,
"learning_rate": 0.0002,
"loss": 1.8865,
"step": 16550
},
{
"epoch": 2.69,
"learning_rate": 0.0002,
"loss": 1.9698,
"step": 16575
},
{
"epoch": 2.69,
"learning_rate": 0.0002,
"loss": 2.0733,
"step": 16600
},
{
"epoch": 2.69,
"learning_rate": 0.0002,
"loss": 1.9668,
"step": 16625
},
{
"epoch": 2.7,
"learning_rate": 0.0002,
"loss": 1.9262,
"step": 16650
},
{
"epoch": 2.7,
"learning_rate": 0.0002,
"loss": 1.9017,
"step": 16675
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 1.9897,
"step": 16700
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 2.0554,
"step": 16725
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 1.7832,
"step": 16750
},
{
"epoch": 2.72,
"learning_rate": 0.0002,
"loss": 1.9988,
"step": 16775
},
{
"epoch": 2.72,
"learning_rate": 0.0002,
"loss": 1.9868,
"step": 16800
},
{
"epoch": 2.73,
"learning_rate": 0.0002,
"loss": 1.9771,
"step": 16825
},
{
"epoch": 2.73,
"learning_rate": 0.0002,
"loss": 1.7744,
"step": 16850
},
{
"epoch": 2.73,
"learning_rate": 0.0002,
"loss": 2.0474,
"step": 16875
},
{
"epoch": 2.74,
"learning_rate": 0.0002,
"loss": 1.8683,
"step": 16900
},
{
"epoch": 2.74,
"learning_rate": 0.0002,
"loss": 1.9437,
"step": 16925
},
{
"epoch": 2.75,
"learning_rate": 0.0002,
"loss": 1.7555,
"step": 16950
},
{
"epoch": 2.75,
"learning_rate": 0.0002,
"loss": 1.9213,
"step": 16975
},
{
"epoch": 2.75,
"learning_rate": 0.0002,
"loss": 1.9488,
"step": 17000
},
{
"epoch": 2.76,
"learning_rate": 0.0002,
"loss": 2.0102,
"step": 17025
},
{
"epoch": 2.76,
"learning_rate": 0.0002,
"loss": 1.8886,
"step": 17050
},
{
"epoch": 2.77,
"learning_rate": 0.0002,
"loss": 2.0411,
"step": 17075
},
{
"epoch": 2.77,
"learning_rate": 0.0002,
"loss": 1.848,
"step": 17100
},
{
"epoch": 2.77,
"learning_rate": 0.0002,
"loss": 1.998,
"step": 17125
},
{
"epoch": 2.78,
"learning_rate": 0.0002,
"loss": 1.8947,
"step": 17150
},
{
"epoch": 2.78,
"learning_rate": 0.0002,
"loss": 1.9295,
"step": 17175
},
{
"epoch": 2.79,
"learning_rate": 0.0002,
"loss": 1.9264,
"step": 17200
},
{
"epoch": 2.79,
"learning_rate": 0.0002,
"loss": 1.9114,
"step": 17225
},
{
"epoch": 2.79,
"learning_rate": 0.0002,
"loss": 1.8539,
"step": 17250
},
{
"epoch": 2.8,
"learning_rate": 0.0002,
"loss": 1.9194,
"step": 17275
},
{
"epoch": 2.8,
"learning_rate": 0.0002,
"loss": 1.8498,
"step": 17300
},
{
"epoch": 2.81,
"learning_rate": 0.0002,
"loss": 1.9933,
"step": 17325
},
{
"epoch": 2.81,
"learning_rate": 0.0002,
"loss": 1.8788,
"step": 17350
},
{
"epoch": 2.82,
"learning_rate": 0.0002,
"loss": 2.0578,
"step": 17375
},
{
"epoch": 2.82,
"learning_rate": 0.0002,
"loss": 1.8991,
"step": 17400
},
{
"epoch": 2.82,
"learning_rate": 0.0002,
"loss": 1.9389,
"step": 17425
},
{
"epoch": 2.83,
"learning_rate": 0.0002,
"loss": 1.8064,
"step": 17450
},
{
"epoch": 2.83,
"learning_rate": 0.0002,
"loss": 2.0353,
"step": 17475
},
{
"epoch": 2.84,
"learning_rate": 0.0002,
"loss": 2.0251,
"step": 17500
},
{
"epoch": 2.84,
"learning_rate": 0.0002,
"loss": 2.0272,
"step": 17525
},
{
"epoch": 2.84,
"learning_rate": 0.0002,
"loss": 1.8236,
"step": 17550
},
{
"epoch": 2.85,
"learning_rate": 0.0002,
"loss": 1.9476,
"step": 17575
},
{
"epoch": 2.85,
"learning_rate": 0.0002,
"loss": 1.8953,
"step": 17600
},
{
"epoch": 2.86,
"learning_rate": 0.0002,
"loss": 2.0235,
"step": 17625
},
{
"epoch": 2.86,
"learning_rate": 0.0002,
"loss": 1.8151,
"step": 17650
},
{
"epoch": 2.86,
"learning_rate": 0.0002,
"loss": 1.9714,
"step": 17675
},
{
"epoch": 2.87,
"learning_rate": 0.0002,
"loss": 1.8929,
"step": 17700
},
{
"epoch": 2.87,
"learning_rate": 0.0002,
"loss": 2.091,
"step": 17725
},
{
"epoch": 2.88,
"learning_rate": 0.0002,
"loss": 1.914,
"step": 17750
},
{
"epoch": 2.88,
"learning_rate": 0.0002,
"loss": 2.0598,
"step": 17775
},
{
"epoch": 2.88,
"learning_rate": 0.0002,
"loss": 1.893,
"step": 17800
},
{
"epoch": 2.89,
"learning_rate": 0.0002,
"loss": 2.0322,
"step": 17825
},
{
"epoch": 2.89,
"learning_rate": 0.0002,
"loss": 1.8023,
"step": 17850
},
{
"epoch": 2.9,
"learning_rate": 0.0002,
"loss": 2.0638,
"step": 17875
},
{
"epoch": 2.9,
"learning_rate": 0.0002,
"loss": 1.8813,
"step": 17900
},
{
"epoch": 2.9,
"learning_rate": 0.0002,
"loss": 1.9056,
"step": 17925
},
{
"epoch": 2.91,
"learning_rate": 0.0002,
"loss": 1.7811,
"step": 17950
},
{
"epoch": 2.91,
"learning_rate": 0.0002,
"loss": 1.9817,
"step": 17975
},
{
"epoch": 2.92,
"learning_rate": 0.0002,
"loss": 1.7945,
"step": 18000
}
],
"logging_steps": 25,
"max_steps": 18516,
"num_train_epochs": 3,
"save_steps": 2000,
"total_flos": 3.638276029315154e+17,
"trial_name": null,
"trial_params": null
}