JLB-JLB's picture
VIT_SEIZURE_231126
f31b7d2
{
"best_metric": 0.6336529850959778,
"best_model_checkpoint": "seizure_vit/seizure_vit_jlb_231126_ff_raw_combo_multichannel/checkpoint-1000",
"epoch": 2.0,
"eval_steps": 1000,
"global_step": 11746,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 9.982972926953858e-05,
"loss": 0.6738,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 9.965945853907713e-05,
"loss": 0.632,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 9.94891878086157e-05,
"loss": 0.5383,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 9.931891707815428e-05,
"loss": 0.6127,
"step": 80
},
{
"epoch": 0.02,
"learning_rate": 9.914864634769284e-05,
"loss": 0.5892,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 9.89783756172314e-05,
"loss": 0.5037,
"step": 120
},
{
"epoch": 0.02,
"learning_rate": 9.880810488676997e-05,
"loss": 0.5173,
"step": 140
},
{
"epoch": 0.03,
"learning_rate": 9.863783415630855e-05,
"loss": 0.5828,
"step": 160
},
{
"epoch": 0.03,
"learning_rate": 9.846756342584709e-05,
"loss": 0.5183,
"step": 180
},
{
"epoch": 0.03,
"learning_rate": 9.829729269538567e-05,
"loss": 0.6184,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 9.812702196492424e-05,
"loss": 0.4538,
"step": 220
},
{
"epoch": 0.04,
"learning_rate": 9.79567512344628e-05,
"loss": 0.5137,
"step": 240
},
{
"epoch": 0.04,
"learning_rate": 9.778648050400136e-05,
"loss": 0.6202,
"step": 260
},
{
"epoch": 0.05,
"learning_rate": 9.761620977353994e-05,
"loss": 0.5561,
"step": 280
},
{
"epoch": 0.05,
"learning_rate": 9.74459390430785e-05,
"loss": 0.5051,
"step": 300
},
{
"epoch": 0.05,
"learning_rate": 9.727566831261706e-05,
"loss": 0.488,
"step": 320
},
{
"epoch": 0.06,
"learning_rate": 9.710539758215563e-05,
"loss": 0.4807,
"step": 340
},
{
"epoch": 0.06,
"learning_rate": 9.693512685169421e-05,
"loss": 0.4633,
"step": 360
},
{
"epoch": 0.06,
"learning_rate": 9.676485612123275e-05,
"loss": 0.4685,
"step": 380
},
{
"epoch": 0.07,
"learning_rate": 9.659458539077133e-05,
"loss": 0.5193,
"step": 400
},
{
"epoch": 0.07,
"learning_rate": 9.64243146603099e-05,
"loss": 0.5035,
"step": 420
},
{
"epoch": 0.07,
"learning_rate": 9.625404392984846e-05,
"loss": 0.5037,
"step": 440
},
{
"epoch": 0.08,
"learning_rate": 9.608377319938703e-05,
"loss": 0.423,
"step": 460
},
{
"epoch": 0.08,
"learning_rate": 9.59135024689256e-05,
"loss": 0.4347,
"step": 480
},
{
"epoch": 0.09,
"learning_rate": 9.574323173846417e-05,
"loss": 0.4306,
"step": 500
},
{
"epoch": 0.09,
"learning_rate": 9.557296100800272e-05,
"loss": 0.4378,
"step": 520
},
{
"epoch": 0.09,
"learning_rate": 9.54026902775413e-05,
"loss": 0.466,
"step": 540
},
{
"epoch": 0.1,
"learning_rate": 9.523241954707987e-05,
"loss": 0.458,
"step": 560
},
{
"epoch": 0.1,
"learning_rate": 9.506214881661842e-05,
"loss": 0.4359,
"step": 580
},
{
"epoch": 0.1,
"learning_rate": 9.489187808615699e-05,
"loss": 0.4842,
"step": 600
},
{
"epoch": 0.11,
"learning_rate": 9.472160735569557e-05,
"loss": 0.4788,
"step": 620
},
{
"epoch": 0.11,
"learning_rate": 9.455133662523413e-05,
"loss": 0.4911,
"step": 640
},
{
"epoch": 0.11,
"learning_rate": 9.438106589477269e-05,
"loss": 0.4083,
"step": 660
},
{
"epoch": 0.12,
"learning_rate": 9.421079516431126e-05,
"loss": 0.4315,
"step": 680
},
{
"epoch": 0.12,
"learning_rate": 9.404052443384984e-05,
"loss": 0.3628,
"step": 700
},
{
"epoch": 0.12,
"learning_rate": 9.387025370338838e-05,
"loss": 0.3642,
"step": 720
},
{
"epoch": 0.13,
"learning_rate": 9.369998297292696e-05,
"loss": 0.5454,
"step": 740
},
{
"epoch": 0.13,
"learning_rate": 9.352971224246553e-05,
"loss": 0.416,
"step": 760
},
{
"epoch": 0.13,
"learning_rate": 9.335944151200409e-05,
"loss": 0.412,
"step": 780
},
{
"epoch": 0.14,
"learning_rate": 9.318917078154265e-05,
"loss": 0.4797,
"step": 800
},
{
"epoch": 0.14,
"learning_rate": 9.301890005108123e-05,
"loss": 0.4569,
"step": 820
},
{
"epoch": 0.14,
"learning_rate": 9.284862932061979e-05,
"loss": 0.4158,
"step": 840
},
{
"epoch": 0.15,
"learning_rate": 9.267835859015835e-05,
"loss": 0.3404,
"step": 860
},
{
"epoch": 0.15,
"learning_rate": 9.250808785969692e-05,
"loss": 0.4466,
"step": 880
},
{
"epoch": 0.15,
"learning_rate": 9.23378171292355e-05,
"loss": 0.3901,
"step": 900
},
{
"epoch": 0.16,
"learning_rate": 9.216754639877405e-05,
"loss": 0.3177,
"step": 920
},
{
"epoch": 0.16,
"learning_rate": 9.199727566831262e-05,
"loss": 0.4223,
"step": 940
},
{
"epoch": 0.16,
"learning_rate": 9.18270049378512e-05,
"loss": 0.4389,
"step": 960
},
{
"epoch": 0.17,
"learning_rate": 9.165673420738976e-05,
"loss": 0.4057,
"step": 980
},
{
"epoch": 0.17,
"learning_rate": 9.148646347692832e-05,
"loss": 0.3219,
"step": 1000
},
{
"epoch": 0.17,
"eval_loss": 0.6336529850959778,
"eval_roc_auc": 0.7311492459698388,
"eval_runtime": 61.1981,
"eval_samples_per_second": 62.845,
"eval_steps_per_second": 3.938,
"step": 1000
},
{
"epoch": 0.17,
"learning_rate": 9.131619274646689e-05,
"loss": 0.4128,
"step": 1020
},
{
"epoch": 0.18,
"learning_rate": 9.114592201600546e-05,
"loss": 0.3611,
"step": 1040
},
{
"epoch": 0.18,
"learning_rate": 9.097565128554401e-05,
"loss": 0.3379,
"step": 1060
},
{
"epoch": 0.18,
"learning_rate": 9.080538055508259e-05,
"loss": 0.4438,
"step": 1080
},
{
"epoch": 0.19,
"learning_rate": 9.063510982462116e-05,
"loss": 0.2961,
"step": 1100
},
{
"epoch": 0.19,
"learning_rate": 9.046483909415972e-05,
"loss": 0.4153,
"step": 1120
},
{
"epoch": 0.19,
"learning_rate": 9.029456836369828e-05,
"loss": 0.326,
"step": 1140
},
{
"epoch": 0.2,
"learning_rate": 9.012429763323686e-05,
"loss": 0.3814,
"step": 1160
},
{
"epoch": 0.2,
"learning_rate": 8.995402690277542e-05,
"loss": 0.4118,
"step": 1180
},
{
"epoch": 0.2,
"learning_rate": 8.978375617231398e-05,
"loss": 0.4309,
"step": 1200
},
{
"epoch": 0.21,
"learning_rate": 8.961348544185255e-05,
"loss": 0.3849,
"step": 1220
},
{
"epoch": 0.21,
"learning_rate": 8.944321471139113e-05,
"loss": 0.3703,
"step": 1240
},
{
"epoch": 0.21,
"learning_rate": 8.927294398092967e-05,
"loss": 0.3768,
"step": 1260
},
{
"epoch": 0.22,
"learning_rate": 8.910267325046825e-05,
"loss": 0.3985,
"step": 1280
},
{
"epoch": 0.22,
"learning_rate": 8.893240252000682e-05,
"loss": 0.4281,
"step": 1300
},
{
"epoch": 0.22,
"learning_rate": 8.876213178954538e-05,
"loss": 0.334,
"step": 1320
},
{
"epoch": 0.23,
"learning_rate": 8.859186105908394e-05,
"loss": 0.3255,
"step": 1340
},
{
"epoch": 0.23,
"learning_rate": 8.842159032862252e-05,
"loss": 0.4332,
"step": 1360
},
{
"epoch": 0.23,
"learning_rate": 8.825131959816108e-05,
"loss": 0.4072,
"step": 1380
},
{
"epoch": 0.24,
"learning_rate": 8.808104886769964e-05,
"loss": 0.4141,
"step": 1400
},
{
"epoch": 0.24,
"learning_rate": 8.791077813723821e-05,
"loss": 0.4453,
"step": 1420
},
{
"epoch": 0.25,
"learning_rate": 8.774050740677679e-05,
"loss": 0.3739,
"step": 1440
},
{
"epoch": 0.25,
"learning_rate": 8.757023667631534e-05,
"loss": 0.3046,
"step": 1460
},
{
"epoch": 0.25,
"learning_rate": 8.739996594585391e-05,
"loss": 0.3169,
"step": 1480
},
{
"epoch": 0.26,
"learning_rate": 8.722969521539248e-05,
"loss": 0.2796,
"step": 1500
},
{
"epoch": 0.26,
"learning_rate": 8.705942448493105e-05,
"loss": 0.3502,
"step": 1520
},
{
"epoch": 0.26,
"learning_rate": 8.68891537544696e-05,
"loss": 0.3848,
"step": 1540
},
{
"epoch": 0.27,
"learning_rate": 8.671888302400818e-05,
"loss": 0.3644,
"step": 1560
},
{
"epoch": 0.27,
"learning_rate": 8.654861229354674e-05,
"loss": 0.2799,
"step": 1580
},
{
"epoch": 0.27,
"learning_rate": 8.63783415630853e-05,
"loss": 0.3763,
"step": 1600
},
{
"epoch": 0.28,
"learning_rate": 8.620807083262388e-05,
"loss": 0.3607,
"step": 1620
},
{
"epoch": 0.28,
"learning_rate": 8.603780010216245e-05,
"loss": 0.3296,
"step": 1640
},
{
"epoch": 0.28,
"learning_rate": 8.586752937170101e-05,
"loss": 0.3528,
"step": 1660
},
{
"epoch": 0.29,
"learning_rate": 8.569725864123957e-05,
"loss": 0.3469,
"step": 1680
},
{
"epoch": 0.29,
"learning_rate": 8.552698791077815e-05,
"loss": 0.3347,
"step": 1700
},
{
"epoch": 0.29,
"learning_rate": 8.535671718031671e-05,
"loss": 0.3707,
"step": 1720
},
{
"epoch": 0.3,
"learning_rate": 8.518644644985527e-05,
"loss": 0.3715,
"step": 1740
},
{
"epoch": 0.3,
"learning_rate": 8.501617571939384e-05,
"loss": 0.331,
"step": 1760
},
{
"epoch": 0.3,
"learning_rate": 8.484590498893242e-05,
"loss": 0.3348,
"step": 1780
},
{
"epoch": 0.31,
"learning_rate": 8.467563425847096e-05,
"loss": 0.3262,
"step": 1800
},
{
"epoch": 0.31,
"learning_rate": 8.450536352800954e-05,
"loss": 0.3505,
"step": 1820
},
{
"epoch": 0.31,
"learning_rate": 8.433509279754811e-05,
"loss": 0.2965,
"step": 1840
},
{
"epoch": 0.32,
"learning_rate": 8.416482206708667e-05,
"loss": 0.3128,
"step": 1860
},
{
"epoch": 0.32,
"learning_rate": 8.399455133662523e-05,
"loss": 0.3518,
"step": 1880
},
{
"epoch": 0.32,
"learning_rate": 8.382428060616381e-05,
"loss": 0.3474,
"step": 1900
},
{
"epoch": 0.33,
"learning_rate": 8.365400987570237e-05,
"loss": 0.3195,
"step": 1920
},
{
"epoch": 0.33,
"learning_rate": 8.348373914524093e-05,
"loss": 0.2808,
"step": 1940
},
{
"epoch": 0.33,
"learning_rate": 8.33134684147795e-05,
"loss": 0.3173,
"step": 1960
},
{
"epoch": 0.34,
"learning_rate": 8.314319768431808e-05,
"loss": 0.2855,
"step": 1980
},
{
"epoch": 0.34,
"learning_rate": 8.297292695385663e-05,
"loss": 0.3625,
"step": 2000
},
{
"epoch": 0.34,
"eval_loss": 0.6380915641784668,
"eval_roc_auc": 0.7038481539261572,
"eval_runtime": 61.4872,
"eval_samples_per_second": 62.55,
"eval_steps_per_second": 3.92,
"step": 2000
},
{
"epoch": 0.34,
"learning_rate": 8.28026562233952e-05,
"loss": 0.3596,
"step": 2020
},
{
"epoch": 0.35,
"learning_rate": 8.263238549293378e-05,
"loss": 0.3004,
"step": 2040
},
{
"epoch": 0.35,
"learning_rate": 8.246211476247234e-05,
"loss": 0.314,
"step": 2060
},
{
"epoch": 0.35,
"learning_rate": 8.22918440320109e-05,
"loss": 0.362,
"step": 2080
},
{
"epoch": 0.36,
"learning_rate": 8.212157330154947e-05,
"loss": 0.3542,
"step": 2100
},
{
"epoch": 0.36,
"learning_rate": 8.195130257108803e-05,
"loss": 0.3308,
"step": 2120
},
{
"epoch": 0.36,
"learning_rate": 8.178103184062659e-05,
"loss": 0.3017,
"step": 2140
},
{
"epoch": 0.37,
"learning_rate": 8.161076111016517e-05,
"loss": 0.3408,
"step": 2160
},
{
"epoch": 0.37,
"learning_rate": 8.144049037970374e-05,
"loss": 0.3011,
"step": 2180
},
{
"epoch": 0.37,
"learning_rate": 8.12702196492423e-05,
"loss": 0.2965,
"step": 2200
},
{
"epoch": 0.38,
"learning_rate": 8.109994891878086e-05,
"loss": 0.3184,
"step": 2220
},
{
"epoch": 0.38,
"learning_rate": 8.092967818831944e-05,
"loss": 0.3705,
"step": 2240
},
{
"epoch": 0.38,
"learning_rate": 8.0759407457858e-05,
"loss": 0.2868,
"step": 2260
},
{
"epoch": 0.39,
"learning_rate": 8.058913672739656e-05,
"loss": 0.3639,
"step": 2280
},
{
"epoch": 0.39,
"learning_rate": 8.041886599693513e-05,
"loss": 0.3164,
"step": 2300
},
{
"epoch": 0.4,
"learning_rate": 8.02485952664737e-05,
"loss": 0.3236,
"step": 2320
},
{
"epoch": 0.4,
"learning_rate": 8.007832453601226e-05,
"loss": 0.3542,
"step": 2340
},
{
"epoch": 0.4,
"learning_rate": 7.990805380555083e-05,
"loss": 0.2618,
"step": 2360
},
{
"epoch": 0.41,
"learning_rate": 7.97377830750894e-05,
"loss": 0.297,
"step": 2380
},
{
"epoch": 0.41,
"learning_rate": 7.956751234462796e-05,
"loss": 0.3231,
"step": 2400
},
{
"epoch": 0.41,
"learning_rate": 7.939724161416653e-05,
"loss": 0.2954,
"step": 2420
},
{
"epoch": 0.42,
"learning_rate": 7.92269708837051e-05,
"loss": 0.2564,
"step": 2440
},
{
"epoch": 0.42,
"learning_rate": 7.905670015324366e-05,
"loss": 0.3272,
"step": 2460
},
{
"epoch": 0.42,
"learning_rate": 7.888642942278222e-05,
"loss": 0.2938,
"step": 2480
},
{
"epoch": 0.43,
"learning_rate": 7.87161586923208e-05,
"loss": 0.3463,
"step": 2500
},
{
"epoch": 0.43,
"learning_rate": 7.854588796185936e-05,
"loss": 0.2096,
"step": 2520
},
{
"epoch": 0.43,
"learning_rate": 7.837561723139792e-05,
"loss": 0.2712,
"step": 2540
},
{
"epoch": 0.44,
"learning_rate": 7.820534650093649e-05,
"loss": 0.3231,
"step": 2560
},
{
"epoch": 0.44,
"learning_rate": 7.803507577047507e-05,
"loss": 0.2442,
"step": 2580
},
{
"epoch": 0.44,
"learning_rate": 7.786480504001363e-05,
"loss": 0.2428,
"step": 2600
},
{
"epoch": 0.45,
"learning_rate": 7.769453430955219e-05,
"loss": 0.382,
"step": 2620
},
{
"epoch": 0.45,
"learning_rate": 7.752426357909076e-05,
"loss": 0.325,
"step": 2640
},
{
"epoch": 0.45,
"learning_rate": 7.735399284862932e-05,
"loss": 0.2761,
"step": 2660
},
{
"epoch": 0.46,
"learning_rate": 7.718372211816788e-05,
"loss": 0.3125,
"step": 2680
},
{
"epoch": 0.46,
"learning_rate": 7.701345138770646e-05,
"loss": 0.3256,
"step": 2700
},
{
"epoch": 0.46,
"learning_rate": 7.684318065724503e-05,
"loss": 0.2671,
"step": 2720
},
{
"epoch": 0.47,
"learning_rate": 7.66729099267836e-05,
"loss": 0.3617,
"step": 2740
},
{
"epoch": 0.47,
"learning_rate": 7.650263919632215e-05,
"loss": 0.3185,
"step": 2760
},
{
"epoch": 0.47,
"learning_rate": 7.633236846586073e-05,
"loss": 0.2941,
"step": 2780
},
{
"epoch": 0.48,
"learning_rate": 7.616209773539929e-05,
"loss": 0.3157,
"step": 2800
},
{
"epoch": 0.48,
"learning_rate": 7.599182700493785e-05,
"loss": 0.3067,
"step": 2820
},
{
"epoch": 0.48,
"learning_rate": 7.582155627447642e-05,
"loss": 0.2796,
"step": 2840
},
{
"epoch": 0.49,
"learning_rate": 7.565128554401499e-05,
"loss": 0.2687,
"step": 2860
},
{
"epoch": 0.49,
"learning_rate": 7.548101481355355e-05,
"loss": 0.2798,
"step": 2880
},
{
"epoch": 0.49,
"learning_rate": 7.531074408309212e-05,
"loss": 0.3448,
"step": 2900
},
{
"epoch": 0.5,
"learning_rate": 7.51404733526307e-05,
"loss": 0.2323,
"step": 2920
},
{
"epoch": 0.5,
"learning_rate": 7.497020262216926e-05,
"loss": 0.2239,
"step": 2940
},
{
"epoch": 0.5,
"learning_rate": 7.479993189170782e-05,
"loss": 0.2555,
"step": 2960
},
{
"epoch": 0.51,
"learning_rate": 7.462966116124639e-05,
"loss": 0.28,
"step": 2980
},
{
"epoch": 0.51,
"learning_rate": 7.445939043078495e-05,
"loss": 0.3435,
"step": 3000
},
{
"epoch": 0.51,
"eval_loss": 0.6890058517456055,
"eval_roc_auc": 0.700988039521581,
"eval_runtime": 61.7073,
"eval_samples_per_second": 62.326,
"eval_steps_per_second": 3.906,
"step": 3000
},
{
"epoch": 0.51,
"learning_rate": 7.428911970032351e-05,
"loss": 0.2978,
"step": 3020
},
{
"epoch": 0.52,
"learning_rate": 7.411884896986209e-05,
"loss": 0.2357,
"step": 3040
},
{
"epoch": 0.52,
"learning_rate": 7.394857823940065e-05,
"loss": 0.2486,
"step": 3060
},
{
"epoch": 0.52,
"learning_rate": 7.377830750893922e-05,
"loss": 0.3084,
"step": 3080
},
{
"epoch": 0.53,
"learning_rate": 7.360803677847778e-05,
"loss": 0.3131,
"step": 3100
},
{
"epoch": 0.53,
"learning_rate": 7.343776604801636e-05,
"loss": 0.2714,
"step": 3120
},
{
"epoch": 0.53,
"learning_rate": 7.326749531755492e-05,
"loss": 0.303,
"step": 3140
},
{
"epoch": 0.54,
"learning_rate": 7.309722458709348e-05,
"loss": 0.3084,
"step": 3160
},
{
"epoch": 0.54,
"learning_rate": 7.292695385663205e-05,
"loss": 0.2525,
"step": 3180
},
{
"epoch": 0.54,
"learning_rate": 7.275668312617061e-05,
"loss": 0.2797,
"step": 3200
},
{
"epoch": 0.55,
"learning_rate": 7.258641239570917e-05,
"loss": 0.3152,
"step": 3220
},
{
"epoch": 0.55,
"learning_rate": 7.241614166524775e-05,
"loss": 0.3469,
"step": 3240
},
{
"epoch": 0.56,
"learning_rate": 7.224587093478631e-05,
"loss": 0.215,
"step": 3260
},
{
"epoch": 0.56,
"learning_rate": 7.207560020432488e-05,
"loss": 0.2997,
"step": 3280
},
{
"epoch": 0.56,
"learning_rate": 7.190532947386344e-05,
"loss": 0.2017,
"step": 3300
},
{
"epoch": 0.57,
"learning_rate": 7.173505874340202e-05,
"loss": 0.3189,
"step": 3320
},
{
"epoch": 0.57,
"learning_rate": 7.156478801294058e-05,
"loss": 0.2546,
"step": 3340
},
{
"epoch": 0.57,
"learning_rate": 7.139451728247914e-05,
"loss": 0.3168,
"step": 3360
},
{
"epoch": 0.58,
"learning_rate": 7.122424655201772e-05,
"loss": 0.3053,
"step": 3380
},
{
"epoch": 0.58,
"learning_rate": 7.105397582155628e-05,
"loss": 0.2574,
"step": 3400
},
{
"epoch": 0.58,
"learning_rate": 7.088370509109484e-05,
"loss": 0.2797,
"step": 3420
},
{
"epoch": 0.59,
"learning_rate": 7.071343436063341e-05,
"loss": 0.2862,
"step": 3440
},
{
"epoch": 0.59,
"learning_rate": 7.054316363017199e-05,
"loss": 0.225,
"step": 3460
},
{
"epoch": 0.59,
"learning_rate": 7.037289289971055e-05,
"loss": 0.2689,
"step": 3480
},
{
"epoch": 0.6,
"learning_rate": 7.020262216924911e-05,
"loss": 0.2524,
"step": 3500
},
{
"epoch": 0.6,
"learning_rate": 7.003235143878768e-05,
"loss": 0.3034,
"step": 3520
},
{
"epoch": 0.6,
"learning_rate": 6.986208070832624e-05,
"loss": 0.2497,
"step": 3540
},
{
"epoch": 0.61,
"learning_rate": 6.96918099778648e-05,
"loss": 0.2239,
"step": 3560
},
{
"epoch": 0.61,
"learning_rate": 6.952153924740338e-05,
"loss": 0.2864,
"step": 3580
},
{
"epoch": 0.61,
"learning_rate": 6.935126851694194e-05,
"loss": 0.2678,
"step": 3600
},
{
"epoch": 0.62,
"learning_rate": 6.918099778648051e-05,
"loss": 0.3411,
"step": 3620
},
{
"epoch": 0.62,
"learning_rate": 6.901072705601907e-05,
"loss": 0.2575,
"step": 3640
},
{
"epoch": 0.62,
"learning_rate": 6.884045632555765e-05,
"loss": 0.2435,
"step": 3660
},
{
"epoch": 0.63,
"learning_rate": 6.867018559509621e-05,
"loss": 0.2486,
"step": 3680
},
{
"epoch": 0.63,
"learning_rate": 6.849991486463477e-05,
"loss": 0.3357,
"step": 3700
},
{
"epoch": 0.63,
"learning_rate": 6.832964413417334e-05,
"loss": 0.3054,
"step": 3720
},
{
"epoch": 0.64,
"learning_rate": 6.81593734037119e-05,
"loss": 0.286,
"step": 3740
},
{
"epoch": 0.64,
"learning_rate": 6.798910267325047e-05,
"loss": 0.3039,
"step": 3760
},
{
"epoch": 0.64,
"learning_rate": 6.781883194278904e-05,
"loss": 0.2512,
"step": 3780
},
{
"epoch": 0.65,
"learning_rate": 6.76485612123276e-05,
"loss": 0.2189,
"step": 3800
},
{
"epoch": 0.65,
"learning_rate": 6.747829048186617e-05,
"loss": 0.2627,
"step": 3820
},
{
"epoch": 0.65,
"learning_rate": 6.730801975140474e-05,
"loss": 0.3069,
"step": 3840
},
{
"epoch": 0.66,
"learning_rate": 6.713774902094331e-05,
"loss": 0.2799,
"step": 3860
},
{
"epoch": 0.66,
"learning_rate": 6.696747829048187e-05,
"loss": 0.2688,
"step": 3880
},
{
"epoch": 0.66,
"learning_rate": 6.679720756002043e-05,
"loss": 0.2331,
"step": 3900
},
{
"epoch": 0.67,
"learning_rate": 6.6626936829559e-05,
"loss": 0.3277,
"step": 3920
},
{
"epoch": 0.67,
"learning_rate": 6.645666609909757e-05,
"loss": 0.2201,
"step": 3940
},
{
"epoch": 0.67,
"learning_rate": 6.628639536863613e-05,
"loss": 0.1987,
"step": 3960
},
{
"epoch": 0.68,
"learning_rate": 6.61161246381747e-05,
"loss": 0.2966,
"step": 3980
},
{
"epoch": 0.68,
"learning_rate": 6.594585390771326e-05,
"loss": 0.2538,
"step": 4000
},
{
"epoch": 0.68,
"eval_loss": 0.7448868751525879,
"eval_roc_auc": 0.6926677067082684,
"eval_runtime": 62.2813,
"eval_samples_per_second": 61.752,
"eval_steps_per_second": 3.87,
"step": 4000
},
{
"epoch": 0.68,
"learning_rate": 6.577558317725184e-05,
"loss": 0.2691,
"step": 4020
},
{
"epoch": 0.69,
"learning_rate": 6.56053124467904e-05,
"loss": 0.2737,
"step": 4040
},
{
"epoch": 0.69,
"learning_rate": 6.543504171632897e-05,
"loss": 0.2611,
"step": 4060
},
{
"epoch": 0.69,
"learning_rate": 6.526477098586753e-05,
"loss": 0.232,
"step": 4080
},
{
"epoch": 0.7,
"learning_rate": 6.50945002554061e-05,
"loss": 0.2703,
"step": 4100
},
{
"epoch": 0.7,
"learning_rate": 6.492422952494467e-05,
"loss": 0.3323,
"step": 4120
},
{
"epoch": 0.7,
"learning_rate": 6.475395879448323e-05,
"loss": 0.2773,
"step": 4140
},
{
"epoch": 0.71,
"learning_rate": 6.45836880640218e-05,
"loss": 0.2647,
"step": 4160
},
{
"epoch": 0.71,
"learning_rate": 6.441341733356036e-05,
"loss": 0.3419,
"step": 4180
},
{
"epoch": 0.72,
"learning_rate": 6.424314660309892e-05,
"loss": 0.2698,
"step": 4200
},
{
"epoch": 0.72,
"learning_rate": 6.40728758726375e-05,
"loss": 0.2644,
"step": 4220
},
{
"epoch": 0.72,
"learning_rate": 6.390260514217606e-05,
"loss": 0.2657,
"step": 4240
},
{
"epoch": 0.73,
"learning_rate": 6.373233441171463e-05,
"loss": 0.2753,
"step": 4260
},
{
"epoch": 0.73,
"learning_rate": 6.35620636812532e-05,
"loss": 0.2249,
"step": 4280
},
{
"epoch": 0.73,
"learning_rate": 6.339179295079176e-05,
"loss": 0.2222,
"step": 4300
},
{
"epoch": 0.74,
"learning_rate": 6.322152222033033e-05,
"loss": 0.2065,
"step": 4320
},
{
"epoch": 0.74,
"learning_rate": 6.305125148986889e-05,
"loss": 0.3167,
"step": 4340
},
{
"epoch": 0.74,
"learning_rate": 6.288098075940747e-05,
"loss": 0.2554,
"step": 4360
},
{
"epoch": 0.75,
"learning_rate": 6.271071002894603e-05,
"loss": 0.2563,
"step": 4380
},
{
"epoch": 0.75,
"learning_rate": 6.25404392984846e-05,
"loss": 0.2462,
"step": 4400
},
{
"epoch": 0.75,
"learning_rate": 6.237016856802316e-05,
"loss": 0.2958,
"step": 4420
},
{
"epoch": 0.76,
"learning_rate": 6.219989783756172e-05,
"loss": 0.2109,
"step": 4440
},
{
"epoch": 0.76,
"learning_rate": 6.20296271071003e-05,
"loss": 0.3015,
"step": 4460
},
{
"epoch": 0.76,
"learning_rate": 6.185935637663886e-05,
"loss": 0.2392,
"step": 4480
},
{
"epoch": 0.77,
"learning_rate": 6.168908564617742e-05,
"loss": 0.2581,
"step": 4500
},
{
"epoch": 0.77,
"learning_rate": 6.151881491571599e-05,
"loss": 0.2369,
"step": 4520
},
{
"epoch": 0.77,
"learning_rate": 6.134854418525455e-05,
"loss": 0.1783,
"step": 4540
},
{
"epoch": 0.78,
"learning_rate": 6.117827345479313e-05,
"loss": 0.2681,
"step": 4560
},
{
"epoch": 0.78,
"learning_rate": 6.1008002724331695e-05,
"loss": 0.2185,
"step": 4580
},
{
"epoch": 0.78,
"learning_rate": 6.083773199387026e-05,
"loss": 0.2423,
"step": 4600
},
{
"epoch": 0.79,
"learning_rate": 6.0667461263408823e-05,
"loss": 0.2622,
"step": 4620
},
{
"epoch": 0.79,
"learning_rate": 6.049719053294739e-05,
"loss": 0.3125,
"step": 4640
},
{
"epoch": 0.79,
"learning_rate": 6.032691980248596e-05,
"loss": 0.2123,
"step": 4660
},
{
"epoch": 0.8,
"learning_rate": 6.015664907202452e-05,
"loss": 0.2413,
"step": 4680
},
{
"epoch": 0.8,
"learning_rate": 5.998637834156309e-05,
"loss": 0.2823,
"step": 4700
},
{
"epoch": 0.8,
"learning_rate": 5.9816107611101655e-05,
"loss": 0.3252,
"step": 4720
},
{
"epoch": 0.81,
"learning_rate": 5.9645836880640215e-05,
"loss": 0.2851,
"step": 4740
},
{
"epoch": 0.81,
"learning_rate": 5.947556615017878e-05,
"loss": 0.2642,
"step": 4760
},
{
"epoch": 0.81,
"learning_rate": 5.930529541971736e-05,
"loss": 0.2397,
"step": 4780
},
{
"epoch": 0.82,
"learning_rate": 5.9135024689255925e-05,
"loss": 0.2273,
"step": 4800
},
{
"epoch": 0.82,
"learning_rate": 5.8964753958794486e-05,
"loss": 0.2461,
"step": 4820
},
{
"epoch": 0.82,
"learning_rate": 5.879448322833305e-05,
"loss": 0.194,
"step": 4840
},
{
"epoch": 0.83,
"learning_rate": 5.862421249787162e-05,
"loss": 0.2545,
"step": 4860
},
{
"epoch": 0.83,
"learning_rate": 5.845394176741018e-05,
"loss": 0.2203,
"step": 4880
},
{
"epoch": 0.83,
"learning_rate": 5.828367103694875e-05,
"loss": 0.2455,
"step": 4900
},
{
"epoch": 0.84,
"learning_rate": 5.811340030648732e-05,
"loss": 0.2296,
"step": 4920
},
{
"epoch": 0.84,
"learning_rate": 5.794312957602588e-05,
"loss": 0.197,
"step": 4940
},
{
"epoch": 0.84,
"learning_rate": 5.7772858845564445e-05,
"loss": 0.2931,
"step": 4960
},
{
"epoch": 0.85,
"learning_rate": 5.760258811510302e-05,
"loss": 0.2107,
"step": 4980
},
{
"epoch": 0.85,
"learning_rate": 5.743231738464159e-05,
"loss": 0.2545,
"step": 5000
},
{
"epoch": 0.85,
"eval_loss": 0.7448053956031799,
"eval_roc_auc": 0.7048881955278212,
"eval_runtime": 62.007,
"eval_samples_per_second": 62.025,
"eval_steps_per_second": 3.887,
"step": 5000
},
{
"epoch": 0.85,
"learning_rate": 5.726204665418015e-05,
"loss": 0.2167,
"step": 5020
},
{
"epoch": 0.86,
"learning_rate": 5.7091775923718715e-05,
"loss": 0.2551,
"step": 5040
},
{
"epoch": 0.86,
"learning_rate": 5.692150519325728e-05,
"loss": 0.2133,
"step": 5060
},
{
"epoch": 0.86,
"learning_rate": 5.6751234462795844e-05,
"loss": 0.2438,
"step": 5080
},
{
"epoch": 0.87,
"learning_rate": 5.658096373233441e-05,
"loss": 0.2712,
"step": 5100
},
{
"epoch": 0.87,
"learning_rate": 5.6410693001872986e-05,
"loss": 0.2106,
"step": 5120
},
{
"epoch": 0.88,
"learning_rate": 5.624042227141155e-05,
"loss": 0.2593,
"step": 5140
},
{
"epoch": 0.88,
"learning_rate": 5.6070151540950114e-05,
"loss": 0.2192,
"step": 5160
},
{
"epoch": 0.88,
"learning_rate": 5.589988081048868e-05,
"loss": 0.2185,
"step": 5180
},
{
"epoch": 0.89,
"learning_rate": 5.572961008002725e-05,
"loss": 0.2977,
"step": 5200
},
{
"epoch": 0.89,
"learning_rate": 5.555933934956581e-05,
"loss": 0.1991,
"step": 5220
},
{
"epoch": 0.89,
"learning_rate": 5.538906861910438e-05,
"loss": 0.2229,
"step": 5240
},
{
"epoch": 0.9,
"learning_rate": 5.5218797888642945e-05,
"loss": 0.2102,
"step": 5260
},
{
"epoch": 0.9,
"learning_rate": 5.5048527158181506e-05,
"loss": 0.2456,
"step": 5280
},
{
"epoch": 0.9,
"learning_rate": 5.4878256427720074e-05,
"loss": 0.233,
"step": 5300
},
{
"epoch": 0.91,
"learning_rate": 5.470798569725865e-05,
"loss": 0.2237,
"step": 5320
},
{
"epoch": 0.91,
"learning_rate": 5.4537714966797215e-05,
"loss": 0.2854,
"step": 5340
},
{
"epoch": 0.91,
"learning_rate": 5.4367444236335776e-05,
"loss": 0.2145,
"step": 5360
},
{
"epoch": 0.92,
"learning_rate": 5.4197173505874344e-05,
"loss": 0.2017,
"step": 5380
},
{
"epoch": 0.92,
"learning_rate": 5.402690277541291e-05,
"loss": 0.2388,
"step": 5400
},
{
"epoch": 0.92,
"learning_rate": 5.385663204495147e-05,
"loss": 0.194,
"step": 5420
},
{
"epoch": 0.93,
"learning_rate": 5.368636131449004e-05,
"loss": 0.2278,
"step": 5440
},
{
"epoch": 0.93,
"learning_rate": 5.3516090584028614e-05,
"loss": 0.2491,
"step": 5460
},
{
"epoch": 0.93,
"learning_rate": 5.334581985356717e-05,
"loss": 0.198,
"step": 5480
},
{
"epoch": 0.94,
"learning_rate": 5.3175549123105736e-05,
"loss": 0.2276,
"step": 5500
},
{
"epoch": 0.94,
"learning_rate": 5.300527839264431e-05,
"loss": 0.1705,
"step": 5520
},
{
"epoch": 0.94,
"learning_rate": 5.283500766218288e-05,
"loss": 0.1965,
"step": 5540
},
{
"epoch": 0.95,
"learning_rate": 5.266473693172144e-05,
"loss": 0.2143,
"step": 5560
},
{
"epoch": 0.95,
"learning_rate": 5.2494466201260006e-05,
"loss": 0.2664,
"step": 5580
},
{
"epoch": 0.95,
"learning_rate": 5.2324195470798574e-05,
"loss": 0.2235,
"step": 5600
},
{
"epoch": 0.96,
"learning_rate": 5.2153924740337134e-05,
"loss": 0.2661,
"step": 5620
},
{
"epoch": 0.96,
"learning_rate": 5.19836540098757e-05,
"loss": 0.2513,
"step": 5640
},
{
"epoch": 0.96,
"learning_rate": 5.1813383279414276e-05,
"loss": 0.2753,
"step": 5660
},
{
"epoch": 0.97,
"learning_rate": 5.164311254895283e-05,
"loss": 0.2342,
"step": 5680
},
{
"epoch": 0.97,
"learning_rate": 5.1472841818491405e-05,
"loss": 0.23,
"step": 5700
},
{
"epoch": 0.97,
"learning_rate": 5.130257108802997e-05,
"loss": 0.1684,
"step": 5720
},
{
"epoch": 0.98,
"learning_rate": 5.113230035756854e-05,
"loss": 0.2614,
"step": 5740
},
{
"epoch": 0.98,
"learning_rate": 5.09620296271071e-05,
"loss": 0.1838,
"step": 5760
},
{
"epoch": 0.98,
"learning_rate": 5.079175889664567e-05,
"loss": 0.2122,
"step": 5780
},
{
"epoch": 0.99,
"learning_rate": 5.0621488166184236e-05,
"loss": 0.2172,
"step": 5800
},
{
"epoch": 0.99,
"learning_rate": 5.04512174357228e-05,
"loss": 0.2916,
"step": 5820
},
{
"epoch": 0.99,
"learning_rate": 5.0280946705261364e-05,
"loss": 0.3004,
"step": 5840
},
{
"epoch": 1.0,
"learning_rate": 5.011067597479994e-05,
"loss": 0.235,
"step": 5860
},
{
"epoch": 1.0,
"learning_rate": 4.99404052443385e-05,
"loss": 0.1567,
"step": 5880
},
{
"epoch": 1.0,
"learning_rate": 4.977013451387707e-05,
"loss": 0.1669,
"step": 5900
},
{
"epoch": 1.01,
"learning_rate": 4.9599863783415635e-05,
"loss": 0.1085,
"step": 5920
},
{
"epoch": 1.01,
"learning_rate": 4.9429593052954195e-05,
"loss": 0.2311,
"step": 5940
},
{
"epoch": 1.01,
"learning_rate": 4.925932232249277e-05,
"loss": 0.2435,
"step": 5960
},
{
"epoch": 1.02,
"learning_rate": 4.908905159203133e-05,
"loss": 0.1423,
"step": 5980
},
{
"epoch": 1.02,
"learning_rate": 4.89187808615699e-05,
"loss": 0.1387,
"step": 6000
},
{
"epoch": 1.02,
"eval_loss": 1.0362623929977417,
"eval_roc_auc": 0.7126885075403017,
"eval_runtime": 61.0936,
"eval_samples_per_second": 62.953,
"eval_steps_per_second": 3.945,
"step": 6000
},
{
"epoch": 1.03,
"learning_rate": 4.8748510131108466e-05,
"loss": 0.2067,
"step": 6020
},
{
"epoch": 1.03,
"learning_rate": 4.857823940064703e-05,
"loss": 0.1343,
"step": 6040
},
{
"epoch": 1.03,
"learning_rate": 4.84079686701856e-05,
"loss": 0.1486,
"step": 6060
},
{
"epoch": 1.04,
"learning_rate": 4.823769793972416e-05,
"loss": 0.1633,
"step": 6080
},
{
"epoch": 1.04,
"learning_rate": 4.806742720926273e-05,
"loss": 0.1794,
"step": 6100
},
{
"epoch": 1.04,
"learning_rate": 4.78971564788013e-05,
"loss": 0.1616,
"step": 6120
},
{
"epoch": 1.05,
"learning_rate": 4.7726885748339864e-05,
"loss": 0.2056,
"step": 6140
},
{
"epoch": 1.05,
"learning_rate": 4.755661501787843e-05,
"loss": 0.1553,
"step": 6160
},
{
"epoch": 1.05,
"learning_rate": 4.738634428741699e-05,
"loss": 0.1408,
"step": 6180
},
{
"epoch": 1.06,
"learning_rate": 4.721607355695556e-05,
"loss": 0.1687,
"step": 6200
},
{
"epoch": 1.06,
"learning_rate": 4.704580282649413e-05,
"loss": 0.226,
"step": 6220
},
{
"epoch": 1.06,
"learning_rate": 4.6875532096032695e-05,
"loss": 0.1183,
"step": 6240
},
{
"epoch": 1.07,
"learning_rate": 4.670526136557126e-05,
"loss": 0.1755,
"step": 6260
},
{
"epoch": 1.07,
"learning_rate": 4.6534990635109824e-05,
"loss": 0.1519,
"step": 6280
},
{
"epoch": 1.07,
"learning_rate": 4.636471990464839e-05,
"loss": 0.189,
"step": 6300
},
{
"epoch": 1.08,
"learning_rate": 4.619444917418696e-05,
"loss": 0.117,
"step": 6320
},
{
"epoch": 1.08,
"learning_rate": 4.6024178443725526e-05,
"loss": 0.1572,
"step": 6340
},
{
"epoch": 1.08,
"learning_rate": 4.5853907713264094e-05,
"loss": 0.2093,
"step": 6360
},
{
"epoch": 1.09,
"learning_rate": 4.5683636982802655e-05,
"loss": 0.2142,
"step": 6380
},
{
"epoch": 1.09,
"learning_rate": 4.551336625234123e-05,
"loss": 0.1272,
"step": 6400
},
{
"epoch": 1.09,
"learning_rate": 4.534309552187979e-05,
"loss": 0.1769,
"step": 6420
},
{
"epoch": 1.1,
"learning_rate": 4.517282479141836e-05,
"loss": 0.1769,
"step": 6440
},
{
"epoch": 1.1,
"learning_rate": 4.5002554060956925e-05,
"loss": 0.1614,
"step": 6460
},
{
"epoch": 1.1,
"learning_rate": 4.4832283330495486e-05,
"loss": 0.1334,
"step": 6480
},
{
"epoch": 1.11,
"learning_rate": 4.466201260003406e-05,
"loss": 0.2556,
"step": 6500
},
{
"epoch": 1.11,
"learning_rate": 4.449174186957262e-05,
"loss": 0.1194,
"step": 6520
},
{
"epoch": 1.11,
"learning_rate": 4.432147113911119e-05,
"loss": 0.1542,
"step": 6540
},
{
"epoch": 1.12,
"learning_rate": 4.4151200408649756e-05,
"loss": 0.1457,
"step": 6560
},
{
"epoch": 1.12,
"learning_rate": 4.3980929678188324e-05,
"loss": 0.1292,
"step": 6580
},
{
"epoch": 1.12,
"learning_rate": 4.381065894772689e-05,
"loss": 0.1781,
"step": 6600
},
{
"epoch": 1.13,
"learning_rate": 4.364038821726545e-05,
"loss": 0.1671,
"step": 6620
},
{
"epoch": 1.13,
"learning_rate": 4.347011748680402e-05,
"loss": 0.099,
"step": 6640
},
{
"epoch": 1.13,
"learning_rate": 4.329984675634259e-05,
"loss": 0.1434,
"step": 6660
},
{
"epoch": 1.14,
"learning_rate": 4.3129576025881155e-05,
"loss": 0.1763,
"step": 6680
},
{
"epoch": 1.14,
"learning_rate": 4.295930529541972e-05,
"loss": 0.1572,
"step": 6700
},
{
"epoch": 1.14,
"learning_rate": 4.278903456495828e-05,
"loss": 0.1454,
"step": 6720
},
{
"epoch": 1.15,
"learning_rate": 4.261876383449685e-05,
"loss": 0.1553,
"step": 6740
},
{
"epoch": 1.15,
"learning_rate": 4.244849310403542e-05,
"loss": 0.1623,
"step": 6760
},
{
"epoch": 1.15,
"learning_rate": 4.2278222373573986e-05,
"loss": 0.1495,
"step": 6780
},
{
"epoch": 1.16,
"learning_rate": 4.2107951643112554e-05,
"loss": 0.1567,
"step": 6800
},
{
"epoch": 1.16,
"learning_rate": 4.1937680912651114e-05,
"loss": 0.1625,
"step": 6820
},
{
"epoch": 1.16,
"learning_rate": 4.176741018218968e-05,
"loss": 0.1457,
"step": 6840
},
{
"epoch": 1.17,
"learning_rate": 4.159713945172825e-05,
"loss": 0.1755,
"step": 6860
},
{
"epoch": 1.17,
"learning_rate": 4.142686872126682e-05,
"loss": 0.1821,
"step": 6880
},
{
"epoch": 1.17,
"learning_rate": 4.1256597990805385e-05,
"loss": 0.1225,
"step": 6900
},
{
"epoch": 1.18,
"learning_rate": 4.1086327260343946e-05,
"loss": 0.1536,
"step": 6920
},
{
"epoch": 1.18,
"learning_rate": 4.091605652988251e-05,
"loss": 0.1866,
"step": 6940
},
{
"epoch": 1.19,
"learning_rate": 4.074578579942108e-05,
"loss": 0.2171,
"step": 6960
},
{
"epoch": 1.19,
"learning_rate": 4.057551506895965e-05,
"loss": 0.1534,
"step": 6980
},
{
"epoch": 1.19,
"learning_rate": 4.0405244338498216e-05,
"loss": 0.1765,
"step": 7000
},
{
"epoch": 1.19,
"eval_loss": 0.8541378378868103,
"eval_roc_auc": 0.7223088923556943,
"eval_runtime": 61.2831,
"eval_samples_per_second": 62.758,
"eval_steps_per_second": 3.933,
"step": 7000
},
{
"epoch": 1.2,
"learning_rate": 4.0234973608036783e-05,
"loss": 0.2078,
"step": 7020
},
{
"epoch": 1.2,
"learning_rate": 4.0064702877575344e-05,
"loss": 0.126,
"step": 7040
},
{
"epoch": 1.2,
"learning_rate": 3.989443214711391e-05,
"loss": 0.1366,
"step": 7060
},
{
"epoch": 1.21,
"learning_rate": 3.972416141665248e-05,
"loss": 0.2349,
"step": 7080
},
{
"epoch": 1.21,
"learning_rate": 3.955389068619105e-05,
"loss": 0.1071,
"step": 7100
},
{
"epoch": 1.21,
"learning_rate": 3.9383619955729615e-05,
"loss": 0.1846,
"step": 7120
},
{
"epoch": 1.22,
"learning_rate": 3.9213349225268175e-05,
"loss": 0.0914,
"step": 7140
},
{
"epoch": 1.22,
"learning_rate": 3.904307849480674e-05,
"loss": 0.1938,
"step": 7160
},
{
"epoch": 1.22,
"learning_rate": 3.887280776434531e-05,
"loss": 0.1661,
"step": 7180
},
{
"epoch": 1.23,
"learning_rate": 3.870253703388388e-05,
"loss": 0.1101,
"step": 7200
},
{
"epoch": 1.23,
"learning_rate": 3.8532266303422446e-05,
"loss": 0.1096,
"step": 7220
},
{
"epoch": 1.23,
"learning_rate": 3.836199557296101e-05,
"loss": 0.2188,
"step": 7240
},
{
"epoch": 1.24,
"learning_rate": 3.8191724842499574e-05,
"loss": 0.2026,
"step": 7260
},
{
"epoch": 1.24,
"learning_rate": 3.802145411203814e-05,
"loss": 0.2182,
"step": 7280
},
{
"epoch": 1.24,
"learning_rate": 3.785118338157671e-05,
"loss": 0.1797,
"step": 7300
},
{
"epoch": 1.25,
"learning_rate": 3.768091265111528e-05,
"loss": 0.1607,
"step": 7320
},
{
"epoch": 1.25,
"learning_rate": 3.7510641920653844e-05,
"loss": 0.0883,
"step": 7340
},
{
"epoch": 1.25,
"learning_rate": 3.7340371190192405e-05,
"loss": 0.1716,
"step": 7360
},
{
"epoch": 1.26,
"learning_rate": 3.717010045973097e-05,
"loss": 0.1355,
"step": 7380
},
{
"epoch": 1.26,
"learning_rate": 3.699982972926954e-05,
"loss": 0.1219,
"step": 7400
},
{
"epoch": 1.26,
"learning_rate": 3.682955899880811e-05,
"loss": 0.1925,
"step": 7420
},
{
"epoch": 1.27,
"learning_rate": 3.6659288268346675e-05,
"loss": 0.1672,
"step": 7440
},
{
"epoch": 1.27,
"learning_rate": 3.6489017537885236e-05,
"loss": 0.1202,
"step": 7460
},
{
"epoch": 1.27,
"learning_rate": 3.6318746807423804e-05,
"loss": 0.2098,
"step": 7480
},
{
"epoch": 1.28,
"learning_rate": 3.614847607696237e-05,
"loss": 0.1884,
"step": 7500
},
{
"epoch": 1.28,
"learning_rate": 3.597820534650094e-05,
"loss": 0.1492,
"step": 7520
},
{
"epoch": 1.28,
"learning_rate": 3.5807934616039506e-05,
"loss": 0.1583,
"step": 7540
},
{
"epoch": 1.29,
"learning_rate": 3.5637663885578074e-05,
"loss": 0.1442,
"step": 7560
},
{
"epoch": 1.29,
"learning_rate": 3.5467393155116635e-05,
"loss": 0.1435,
"step": 7580
},
{
"epoch": 1.29,
"learning_rate": 3.52971224246552e-05,
"loss": 0.1081,
"step": 7600
},
{
"epoch": 1.3,
"learning_rate": 3.512685169419377e-05,
"loss": 0.1193,
"step": 7620
},
{
"epoch": 1.3,
"learning_rate": 3.495658096373234e-05,
"loss": 0.176,
"step": 7640
},
{
"epoch": 1.3,
"learning_rate": 3.4786310233270905e-05,
"loss": 0.1538,
"step": 7660
},
{
"epoch": 1.31,
"learning_rate": 3.4616039502809466e-05,
"loss": 0.1509,
"step": 7680
},
{
"epoch": 1.31,
"learning_rate": 3.4445768772348034e-05,
"loss": 0.1852,
"step": 7700
},
{
"epoch": 1.31,
"learning_rate": 3.42754980418866e-05,
"loss": 0.1699,
"step": 7720
},
{
"epoch": 1.32,
"learning_rate": 3.410522731142517e-05,
"loss": 0.1381,
"step": 7740
},
{
"epoch": 1.32,
"learning_rate": 3.3934956580963736e-05,
"loss": 0.1733,
"step": 7760
},
{
"epoch": 1.32,
"learning_rate": 3.37646858505023e-05,
"loss": 0.1489,
"step": 7780
},
{
"epoch": 1.33,
"learning_rate": 3.3594415120040865e-05,
"loss": 0.177,
"step": 7800
},
{
"epoch": 1.33,
"learning_rate": 3.342414438957943e-05,
"loss": 0.1069,
"step": 7820
},
{
"epoch": 1.33,
"learning_rate": 3.3253873659118e-05,
"loss": 0.1411,
"step": 7840
},
{
"epoch": 1.34,
"learning_rate": 3.308360292865657e-05,
"loss": 0.0857,
"step": 7860
},
{
"epoch": 1.34,
"learning_rate": 3.291333219819513e-05,
"loss": 0.1678,
"step": 7880
},
{
"epoch": 1.35,
"learning_rate": 3.2743061467733696e-05,
"loss": 0.163,
"step": 7900
},
{
"epoch": 1.35,
"learning_rate": 3.257279073727226e-05,
"loss": 0.1591,
"step": 7920
},
{
"epoch": 1.35,
"learning_rate": 3.240252000681083e-05,
"loss": 0.1425,
"step": 7940
},
{
"epoch": 1.36,
"learning_rate": 3.22322492763494e-05,
"loss": 0.1205,
"step": 7960
},
{
"epoch": 1.36,
"learning_rate": 3.206197854588796e-05,
"loss": 0.1335,
"step": 7980
},
{
"epoch": 1.36,
"learning_rate": 3.1891707815426534e-05,
"loss": 0.1385,
"step": 8000
},
{
"epoch": 1.36,
"eval_loss": 1.0059247016906738,
"eval_roc_auc": 0.7228289131565262,
"eval_runtime": 61.5799,
"eval_samples_per_second": 62.455,
"eval_steps_per_second": 3.914,
"step": 8000
},
{
"epoch": 1.37,
"learning_rate": 3.1721437084965094e-05,
"loss": 0.0716,
"step": 8020
},
{
"epoch": 1.37,
"learning_rate": 3.155116635450366e-05,
"loss": 0.1291,
"step": 8040
},
{
"epoch": 1.37,
"learning_rate": 3.138089562404223e-05,
"loss": 0.1216,
"step": 8060
},
{
"epoch": 1.38,
"learning_rate": 3.12106248935808e-05,
"loss": 0.1807,
"step": 8080
},
{
"epoch": 1.38,
"learning_rate": 3.1040354163119365e-05,
"loss": 0.1524,
"step": 8100
},
{
"epoch": 1.38,
"learning_rate": 3.0870083432657926e-05,
"loss": 0.1891,
"step": 8120
},
{
"epoch": 1.39,
"learning_rate": 3.069981270219649e-05,
"loss": 0.152,
"step": 8140
},
{
"epoch": 1.39,
"learning_rate": 3.052954197173506e-05,
"loss": 0.1547,
"step": 8160
},
{
"epoch": 1.39,
"learning_rate": 3.0359271241273628e-05,
"loss": 0.1442,
"step": 8180
},
{
"epoch": 1.4,
"learning_rate": 3.0189000510812192e-05,
"loss": 0.1822,
"step": 8200
},
{
"epoch": 1.4,
"learning_rate": 3.0018729780350757e-05,
"loss": 0.1413,
"step": 8220
},
{
"epoch": 1.4,
"learning_rate": 2.9848459049889328e-05,
"loss": 0.1772,
"step": 8240
},
{
"epoch": 1.41,
"learning_rate": 2.9678188319427892e-05,
"loss": 0.1804,
"step": 8260
},
{
"epoch": 1.41,
"learning_rate": 2.950791758896646e-05,
"loss": 0.065,
"step": 8280
},
{
"epoch": 1.41,
"learning_rate": 2.9337646858505024e-05,
"loss": 0.1447,
"step": 8300
},
{
"epoch": 1.42,
"learning_rate": 2.9167376128043588e-05,
"loss": 0.1536,
"step": 8320
},
{
"epoch": 1.42,
"learning_rate": 2.899710539758216e-05,
"loss": 0.0825,
"step": 8340
},
{
"epoch": 1.42,
"learning_rate": 2.8826834667120723e-05,
"loss": 0.1756,
"step": 8360
},
{
"epoch": 1.43,
"learning_rate": 2.865656393665929e-05,
"loss": 0.1154,
"step": 8380
},
{
"epoch": 1.43,
"learning_rate": 2.8486293206197855e-05,
"loss": 0.1146,
"step": 8400
},
{
"epoch": 1.43,
"learning_rate": 2.831602247573642e-05,
"loss": 0.1017,
"step": 8420
},
{
"epoch": 1.44,
"learning_rate": 2.814575174527499e-05,
"loss": 0.1515,
"step": 8440
},
{
"epoch": 1.44,
"learning_rate": 2.7975481014813554e-05,
"loss": 0.1719,
"step": 8460
},
{
"epoch": 1.44,
"learning_rate": 2.7805210284352125e-05,
"loss": 0.1265,
"step": 8480
},
{
"epoch": 1.45,
"learning_rate": 2.763493955389069e-05,
"loss": 0.126,
"step": 8500
},
{
"epoch": 1.45,
"learning_rate": 2.746466882342925e-05,
"loss": 0.1407,
"step": 8520
},
{
"epoch": 1.45,
"learning_rate": 2.729439809296782e-05,
"loss": 0.1655,
"step": 8540
},
{
"epoch": 1.46,
"learning_rate": 2.7124127362506385e-05,
"loss": 0.0895,
"step": 8560
},
{
"epoch": 1.46,
"learning_rate": 2.6953856632044956e-05,
"loss": 0.1781,
"step": 8580
},
{
"epoch": 1.46,
"learning_rate": 2.678358590158352e-05,
"loss": 0.1307,
"step": 8600
},
{
"epoch": 1.47,
"learning_rate": 2.6613315171122084e-05,
"loss": 0.133,
"step": 8620
},
{
"epoch": 1.47,
"learning_rate": 2.6443044440660652e-05,
"loss": 0.1374,
"step": 8640
},
{
"epoch": 1.47,
"learning_rate": 2.6272773710199216e-05,
"loss": 0.196,
"step": 8660
},
{
"epoch": 1.48,
"learning_rate": 2.6102502979737787e-05,
"loss": 0.1383,
"step": 8680
},
{
"epoch": 1.48,
"learning_rate": 2.593223224927635e-05,
"loss": 0.1699,
"step": 8700
},
{
"epoch": 1.48,
"learning_rate": 2.5761961518814916e-05,
"loss": 0.1672,
"step": 8720
},
{
"epoch": 1.49,
"learning_rate": 2.5591690788353483e-05,
"loss": 0.1364,
"step": 8740
},
{
"epoch": 1.49,
"learning_rate": 2.5421420057892047e-05,
"loss": 0.1798,
"step": 8760
},
{
"epoch": 1.49,
"learning_rate": 2.5251149327430618e-05,
"loss": 0.0869,
"step": 8780
},
{
"epoch": 1.5,
"learning_rate": 2.5080878596969182e-05,
"loss": 0.1623,
"step": 8800
},
{
"epoch": 1.5,
"learning_rate": 2.491060786650775e-05,
"loss": 0.114,
"step": 8820
},
{
"epoch": 1.51,
"learning_rate": 2.4740337136046314e-05,
"loss": 0.1396,
"step": 8840
},
{
"epoch": 1.51,
"learning_rate": 2.4570066405584882e-05,
"loss": 0.1033,
"step": 8860
},
{
"epoch": 1.51,
"learning_rate": 2.4399795675123446e-05,
"loss": 0.1672,
"step": 8880
},
{
"epoch": 1.52,
"learning_rate": 2.4229524944662014e-05,
"loss": 0.1259,
"step": 8900
},
{
"epoch": 1.52,
"learning_rate": 2.405925421420058e-05,
"loss": 0.1251,
"step": 8920
},
{
"epoch": 1.52,
"learning_rate": 2.3888983483739145e-05,
"loss": 0.1109,
"step": 8940
},
{
"epoch": 1.53,
"learning_rate": 2.3718712753277713e-05,
"loss": 0.1293,
"step": 8960
},
{
"epoch": 1.53,
"learning_rate": 2.3548442022816277e-05,
"loss": 0.1787,
"step": 8980
},
{
"epoch": 1.53,
"learning_rate": 2.3378171292354845e-05,
"loss": 0.1602,
"step": 9000
},
{
"epoch": 1.53,
"eval_loss": 0.8951427936553955,
"eval_roc_auc": 0.7069682787311493,
"eval_runtime": 61.6174,
"eval_samples_per_second": 62.417,
"eval_steps_per_second": 3.911,
"step": 9000
},
{
"epoch": 1.54,
"learning_rate": 2.3207900561893412e-05,
"loss": 0.1368,
"step": 9020
},
{
"epoch": 1.54,
"learning_rate": 2.303762983143198e-05,
"loss": 0.1516,
"step": 9040
},
{
"epoch": 1.54,
"learning_rate": 2.2867359100970544e-05,
"loss": 0.1231,
"step": 9060
},
{
"epoch": 1.55,
"learning_rate": 2.2697088370509108e-05,
"loss": 0.1646,
"step": 9080
},
{
"epoch": 1.55,
"learning_rate": 2.2526817640047676e-05,
"loss": 0.0975,
"step": 9100
},
{
"epoch": 1.55,
"learning_rate": 2.2356546909586243e-05,
"loss": 0.1324,
"step": 9120
},
{
"epoch": 1.56,
"learning_rate": 2.218627617912481e-05,
"loss": 0.0968,
"step": 9140
},
{
"epoch": 1.56,
"learning_rate": 2.2016005448663375e-05,
"loss": 0.1427,
"step": 9160
},
{
"epoch": 1.56,
"learning_rate": 2.1845734718201943e-05,
"loss": 0.1165,
"step": 9180
},
{
"epoch": 1.57,
"learning_rate": 2.1675463987740507e-05,
"loss": 0.0872,
"step": 9200
},
{
"epoch": 1.57,
"learning_rate": 2.1505193257279074e-05,
"loss": 0.1632,
"step": 9220
},
{
"epoch": 1.57,
"learning_rate": 2.1334922526817642e-05,
"loss": 0.1679,
"step": 9240
},
{
"epoch": 1.58,
"learning_rate": 2.116465179635621e-05,
"loss": 0.1476,
"step": 9260
},
{
"epoch": 1.58,
"learning_rate": 2.0994381065894774e-05,
"loss": 0.1603,
"step": 9280
},
{
"epoch": 1.58,
"learning_rate": 2.0824110335433338e-05,
"loss": 0.1235,
"step": 9300
},
{
"epoch": 1.59,
"learning_rate": 2.0653839604971906e-05,
"loss": 0.1531,
"step": 9320
},
{
"epoch": 1.59,
"learning_rate": 2.0483568874510473e-05,
"loss": 0.11,
"step": 9340
},
{
"epoch": 1.59,
"learning_rate": 2.031329814404904e-05,
"loss": 0.1325,
"step": 9360
},
{
"epoch": 1.6,
"learning_rate": 2.0143027413587605e-05,
"loss": 0.1028,
"step": 9380
},
{
"epoch": 1.6,
"learning_rate": 1.997275668312617e-05,
"loss": 0.1006,
"step": 9400
},
{
"epoch": 1.6,
"learning_rate": 1.9802485952664737e-05,
"loss": 0.1397,
"step": 9420
},
{
"epoch": 1.61,
"learning_rate": 1.9632215222203304e-05,
"loss": 0.0992,
"step": 9440
},
{
"epoch": 1.61,
"learning_rate": 1.9461944491741872e-05,
"loss": 0.1622,
"step": 9460
},
{
"epoch": 1.61,
"learning_rate": 1.929167376128044e-05,
"loss": 0.0952,
"step": 9480
},
{
"epoch": 1.62,
"learning_rate": 1.9121403030819e-05,
"loss": 0.1258,
"step": 9500
},
{
"epoch": 1.62,
"learning_rate": 1.8951132300357568e-05,
"loss": 0.0781,
"step": 9520
},
{
"epoch": 1.62,
"learning_rate": 1.8780861569896135e-05,
"loss": 0.0882,
"step": 9540
},
{
"epoch": 1.63,
"learning_rate": 1.8610590839434703e-05,
"loss": 0.172,
"step": 9560
},
{
"epoch": 1.63,
"learning_rate": 1.844032010897327e-05,
"loss": 0.1168,
"step": 9580
},
{
"epoch": 1.63,
"learning_rate": 1.8270049378511835e-05,
"loss": 0.1813,
"step": 9600
},
{
"epoch": 1.64,
"learning_rate": 1.80997786480504e-05,
"loss": 0.1029,
"step": 9620
},
{
"epoch": 1.64,
"learning_rate": 1.7929507917588966e-05,
"loss": 0.1432,
"step": 9640
},
{
"epoch": 1.64,
"learning_rate": 1.7759237187127534e-05,
"loss": 0.1418,
"step": 9660
},
{
"epoch": 1.65,
"learning_rate": 1.75889664566661e-05,
"loss": 0.0893,
"step": 9680
},
{
"epoch": 1.65,
"learning_rate": 1.7418695726204666e-05,
"loss": 0.0952,
"step": 9700
},
{
"epoch": 1.66,
"learning_rate": 1.724842499574323e-05,
"loss": 0.1023,
"step": 9720
},
{
"epoch": 1.66,
"learning_rate": 1.7078154265281798e-05,
"loss": 0.1191,
"step": 9740
},
{
"epoch": 1.66,
"learning_rate": 1.6907883534820365e-05,
"loss": 0.1692,
"step": 9760
},
{
"epoch": 1.67,
"learning_rate": 1.6737612804358933e-05,
"loss": 0.154,
"step": 9780
},
{
"epoch": 1.67,
"learning_rate": 1.65673420738975e-05,
"loss": 0.0899,
"step": 9800
},
{
"epoch": 1.67,
"learning_rate": 1.6397071343436064e-05,
"loss": 0.1009,
"step": 9820
},
{
"epoch": 1.68,
"learning_rate": 1.622680061297463e-05,
"loss": 0.1591,
"step": 9840
},
{
"epoch": 1.68,
"learning_rate": 1.6056529882513196e-05,
"loss": 0.1025,
"step": 9860
},
{
"epoch": 1.68,
"learning_rate": 1.5886259152051764e-05,
"loss": 0.084,
"step": 9880
},
{
"epoch": 1.69,
"learning_rate": 1.571598842159033e-05,
"loss": 0.1421,
"step": 9900
},
{
"epoch": 1.69,
"learning_rate": 1.5545717691128896e-05,
"loss": 0.0654,
"step": 9920
},
{
"epoch": 1.69,
"learning_rate": 1.537544696066746e-05,
"loss": 0.1395,
"step": 9940
},
{
"epoch": 1.7,
"learning_rate": 1.5205176230206027e-05,
"loss": 0.0952,
"step": 9960
},
{
"epoch": 1.7,
"learning_rate": 1.5034905499744595e-05,
"loss": 0.1298,
"step": 9980
},
{
"epoch": 1.7,
"learning_rate": 1.4864634769283162e-05,
"loss": 0.1158,
"step": 10000
},
{
"epoch": 1.7,
"eval_loss": 1.0356404781341553,
"eval_roc_auc": 0.7220488819552782,
"eval_runtime": 61.5294,
"eval_samples_per_second": 62.507,
"eval_steps_per_second": 3.917,
"step": 10000
},
{
"epoch": 1.71,
"learning_rate": 1.4694364038821728e-05,
"loss": 0.1728,
"step": 10020
},
{
"epoch": 1.71,
"learning_rate": 1.4524093308360293e-05,
"loss": 0.1284,
"step": 10040
},
{
"epoch": 1.71,
"learning_rate": 1.435382257789886e-05,
"loss": 0.17,
"step": 10060
},
{
"epoch": 1.72,
"learning_rate": 1.4183551847437426e-05,
"loss": 0.1086,
"step": 10080
},
{
"epoch": 1.72,
"learning_rate": 1.4013281116975994e-05,
"loss": 0.1215,
"step": 10100
},
{
"epoch": 1.72,
"learning_rate": 1.384301038651456e-05,
"loss": 0.1668,
"step": 10120
},
{
"epoch": 1.73,
"learning_rate": 1.3672739656053124e-05,
"loss": 0.1301,
"step": 10140
},
{
"epoch": 1.73,
"learning_rate": 1.3502468925591691e-05,
"loss": 0.1798,
"step": 10160
},
{
"epoch": 1.73,
"learning_rate": 1.3332198195130257e-05,
"loss": 0.0903,
"step": 10180
},
{
"epoch": 1.74,
"learning_rate": 1.3161927464668825e-05,
"loss": 0.0811,
"step": 10200
},
{
"epoch": 1.74,
"learning_rate": 1.299165673420739e-05,
"loss": 0.0816,
"step": 10220
},
{
"epoch": 1.74,
"learning_rate": 1.2821386003745955e-05,
"loss": 0.1538,
"step": 10240
},
{
"epoch": 1.75,
"learning_rate": 1.2651115273284522e-05,
"loss": 0.1193,
"step": 10260
},
{
"epoch": 1.75,
"learning_rate": 1.248084454282309e-05,
"loss": 0.1594,
"step": 10280
},
{
"epoch": 1.75,
"learning_rate": 1.2310573812361656e-05,
"loss": 0.1308,
"step": 10300
},
{
"epoch": 1.76,
"learning_rate": 1.2140303081900222e-05,
"loss": 0.1467,
"step": 10320
},
{
"epoch": 1.76,
"learning_rate": 1.1970032351438788e-05,
"loss": 0.121,
"step": 10340
},
{
"epoch": 1.76,
"learning_rate": 1.1799761620977355e-05,
"loss": 0.1305,
"step": 10360
},
{
"epoch": 1.77,
"learning_rate": 1.1629490890515921e-05,
"loss": 0.1346,
"step": 10380
},
{
"epoch": 1.77,
"learning_rate": 1.1459220160054487e-05,
"loss": 0.1502,
"step": 10400
},
{
"epoch": 1.77,
"learning_rate": 1.1288949429593053e-05,
"loss": 0.1624,
"step": 10420
},
{
"epoch": 1.78,
"learning_rate": 1.111867869913162e-05,
"loss": 0.1255,
"step": 10440
},
{
"epoch": 1.78,
"learning_rate": 1.0948407968670186e-05,
"loss": 0.1101,
"step": 10460
},
{
"epoch": 1.78,
"learning_rate": 1.0778137238208752e-05,
"loss": 0.1208,
"step": 10480
},
{
"epoch": 1.79,
"learning_rate": 1.060786650774732e-05,
"loss": 0.1024,
"step": 10500
},
{
"epoch": 1.79,
"learning_rate": 1.0437595777285886e-05,
"loss": 0.0722,
"step": 10520
},
{
"epoch": 1.79,
"learning_rate": 1.0267325046824451e-05,
"loss": 0.1018,
"step": 10540
},
{
"epoch": 1.8,
"learning_rate": 1.0097054316363017e-05,
"loss": 0.0841,
"step": 10560
},
{
"epoch": 1.8,
"learning_rate": 9.926783585901583e-06,
"loss": 0.1512,
"step": 10580
},
{
"epoch": 1.8,
"learning_rate": 9.75651285544015e-06,
"loss": 0.1237,
"step": 10600
},
{
"epoch": 1.81,
"learning_rate": 9.586242124978717e-06,
"loss": 0.1827,
"step": 10620
},
{
"epoch": 1.81,
"learning_rate": 9.415971394517283e-06,
"loss": 0.0992,
"step": 10640
},
{
"epoch": 1.82,
"learning_rate": 9.24570066405585e-06,
"loss": 0.1336,
"step": 10660
},
{
"epoch": 1.82,
"learning_rate": 9.075429933594416e-06,
"loss": 0.1758,
"step": 10680
},
{
"epoch": 1.82,
"learning_rate": 8.905159203132982e-06,
"loss": 0.1011,
"step": 10700
},
{
"epoch": 1.83,
"learning_rate": 8.734888472671548e-06,
"loss": 0.1394,
"step": 10720
},
{
"epoch": 1.83,
"learning_rate": 8.564617742210114e-06,
"loss": 0.1043,
"step": 10740
},
{
"epoch": 1.83,
"learning_rate": 8.394347011748681e-06,
"loss": 0.0739,
"step": 10760
},
{
"epoch": 1.84,
"learning_rate": 8.224076281287247e-06,
"loss": 0.1109,
"step": 10780
},
{
"epoch": 1.84,
"learning_rate": 8.053805550825813e-06,
"loss": 0.0993,
"step": 10800
},
{
"epoch": 1.84,
"learning_rate": 7.88353482036438e-06,
"loss": 0.1058,
"step": 10820
},
{
"epoch": 1.85,
"learning_rate": 7.713264089902945e-06,
"loss": 0.0842,
"step": 10840
},
{
"epoch": 1.85,
"learning_rate": 7.542993359441512e-06,
"loss": 0.126,
"step": 10860
},
{
"epoch": 1.85,
"learning_rate": 7.372722628980079e-06,
"loss": 0.1401,
"step": 10880
},
{
"epoch": 1.86,
"learning_rate": 7.202451898518645e-06,
"loss": 0.1465,
"step": 10900
},
{
"epoch": 1.86,
"learning_rate": 7.032181168057212e-06,
"loss": 0.1528,
"step": 10920
},
{
"epoch": 1.86,
"learning_rate": 6.861910437595778e-06,
"loss": 0.1731,
"step": 10940
},
{
"epoch": 1.87,
"learning_rate": 6.691639707134343e-06,
"loss": 0.1032,
"step": 10960
},
{
"epoch": 1.87,
"learning_rate": 6.52136897667291e-06,
"loss": 0.1229,
"step": 10980
},
{
"epoch": 1.87,
"learning_rate": 6.351098246211476e-06,
"loss": 0.0973,
"step": 11000
},
{
"epoch": 1.87,
"eval_loss": 1.0616440773010254,
"eval_roc_auc": 0.7132085283411336,
"eval_runtime": 61.5896,
"eval_samples_per_second": 62.446,
"eval_steps_per_second": 3.913,
"step": 11000
},
{
"epoch": 1.88,
"learning_rate": 6.180827515750043e-06,
"loss": 0.1092,
"step": 11020
},
{
"epoch": 1.88,
"learning_rate": 6.010556785288609e-06,
"loss": 0.067,
"step": 11040
},
{
"epoch": 1.88,
"learning_rate": 5.840286054827176e-06,
"loss": 0.1596,
"step": 11060
},
{
"epoch": 1.89,
"learning_rate": 5.670015324365742e-06,
"loss": 0.1131,
"step": 11080
},
{
"epoch": 1.89,
"learning_rate": 5.499744593904308e-06,
"loss": 0.0583,
"step": 11100
},
{
"epoch": 1.89,
"learning_rate": 5.329473863442874e-06,
"loss": 0.092,
"step": 11120
},
{
"epoch": 1.9,
"learning_rate": 5.1592031329814406e-06,
"loss": 0.1005,
"step": 11140
},
{
"epoch": 1.9,
"learning_rate": 4.988932402520007e-06,
"loss": 0.0978,
"step": 11160
},
{
"epoch": 1.9,
"learning_rate": 4.818661672058573e-06,
"loss": 0.1147,
"step": 11180
},
{
"epoch": 1.91,
"learning_rate": 4.64839094159714e-06,
"loss": 0.0973,
"step": 11200
},
{
"epoch": 1.91,
"learning_rate": 4.478120211135706e-06,
"loss": 0.1623,
"step": 11220
},
{
"epoch": 1.91,
"learning_rate": 4.3078494806742725e-06,
"loss": 0.1298,
"step": 11240
},
{
"epoch": 1.92,
"learning_rate": 4.137578750212838e-06,
"loss": 0.111,
"step": 11260
},
{
"epoch": 1.92,
"learning_rate": 3.967308019751405e-06,
"loss": 0.0789,
"step": 11280
},
{
"epoch": 1.92,
"learning_rate": 3.797037289289971e-06,
"loss": 0.1562,
"step": 11300
},
{
"epoch": 1.93,
"learning_rate": 3.6267665588285377e-06,
"loss": 0.1051,
"step": 11320
},
{
"epoch": 1.93,
"learning_rate": 3.456495828367104e-06,
"loss": 0.1105,
"step": 11340
},
{
"epoch": 1.93,
"learning_rate": 3.28622509790567e-06,
"loss": 0.1209,
"step": 11360
},
{
"epoch": 1.94,
"learning_rate": 3.1159543674442367e-06,
"loss": 0.1163,
"step": 11380
},
{
"epoch": 1.94,
"learning_rate": 2.9456836369828025e-06,
"loss": 0.1602,
"step": 11400
},
{
"epoch": 1.94,
"learning_rate": 2.7754129065213693e-06,
"loss": 0.1459,
"step": 11420
},
{
"epoch": 1.95,
"learning_rate": 2.6051421760599356e-06,
"loss": 0.0989,
"step": 11440
},
{
"epoch": 1.95,
"learning_rate": 2.434871445598502e-06,
"loss": 0.1318,
"step": 11460
},
{
"epoch": 1.95,
"learning_rate": 2.264600715137068e-06,
"loss": 0.1068,
"step": 11480
},
{
"epoch": 1.96,
"learning_rate": 2.0943299846756345e-06,
"loss": 0.1266,
"step": 11500
},
{
"epoch": 1.96,
"learning_rate": 1.924059254214201e-06,
"loss": 0.0798,
"step": 11520
},
{
"epoch": 1.96,
"learning_rate": 1.7537885237527669e-06,
"loss": 0.1307,
"step": 11540
},
{
"epoch": 1.97,
"learning_rate": 1.5835177932913334e-06,
"loss": 0.1236,
"step": 11560
},
{
"epoch": 1.97,
"learning_rate": 1.4132470628298997e-06,
"loss": 0.1716,
"step": 11580
},
{
"epoch": 1.98,
"learning_rate": 1.242976332368466e-06,
"loss": 0.0998,
"step": 11600
},
{
"epoch": 1.98,
"learning_rate": 1.0727056019070323e-06,
"loss": 0.07,
"step": 11620
},
{
"epoch": 1.98,
"learning_rate": 9.024348714455986e-07,
"loss": 0.1267,
"step": 11640
},
{
"epoch": 1.99,
"learning_rate": 7.321641409841648e-07,
"loss": 0.0816,
"step": 11660
},
{
"epoch": 1.99,
"learning_rate": 5.618934105227311e-07,
"loss": 0.1601,
"step": 11680
},
{
"epoch": 1.99,
"learning_rate": 3.916226800612975e-07,
"loss": 0.123,
"step": 11700
},
{
"epoch": 2.0,
"learning_rate": 2.213519495998638e-07,
"loss": 0.0952,
"step": 11720
},
{
"epoch": 2.0,
"learning_rate": 5.1081219138430104e-08,
"loss": 0.1558,
"step": 11740
},
{
"epoch": 2.0,
"step": 11746,
"total_flos": 1.4562604656332587e+19,
"train_loss": 0.22876313900634984,
"train_runtime": 3986.2573,
"train_samples_per_second": 47.143,
"train_steps_per_second": 2.947
}
],
"logging_steps": 20,
"max_steps": 11746,
"num_train_epochs": 2,
"save_steps": 1000,
"total_flos": 1.4562604656332587e+19,
"trial_name": null,
"trial_params": null
}