name / trainer_state.json
CoryMagic's picture
Upload 10 files
c05b5f6
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.653271338425772,
"global_step": 11000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 1e-07,
"loss": 10.8195,
"step": 100
},
{
"epoch": 0.07,
"learning_rate": 2e-07,
"loss": 10.198,
"step": 200
},
{
"epoch": 0.1,
"learning_rate": 3e-07,
"loss": 9.7934,
"step": 300
},
{
"epoch": 0.13,
"learning_rate": 4e-07,
"loss": 9.6558,
"step": 400
},
{
"epoch": 0.17,
"learning_rate": 5e-07,
"loss": 9.5031,
"step": 500
},
{
"epoch": 0.2,
"learning_rate": 6e-07,
"loss": 9.3498,
"step": 600
},
{
"epoch": 0.23,
"learning_rate": 7e-07,
"loss": 9.171,
"step": 700
},
{
"epoch": 0.27,
"learning_rate": 8e-07,
"loss": 8.9623,
"step": 800
},
{
"epoch": 0.3,
"learning_rate": 9e-07,
"loss": 8.885,
"step": 900
},
{
"epoch": 0.33,
"learning_rate": 1e-06,
"loss": 8.6453,
"step": 1000
},
{
"epoch": 0.37,
"learning_rate": 9.96564754379938e-07,
"loss": 8.5159,
"step": 1100
},
{
"epoch": 0.4,
"learning_rate": 9.931295087598764e-07,
"loss": 8.3516,
"step": 1200
},
{
"epoch": 0.43,
"learning_rate": 9.896942631398145e-07,
"loss": 8.2104,
"step": 1300
},
{
"epoch": 0.46,
"learning_rate": 9.862590175197526e-07,
"loss": 8.0953,
"step": 1400
},
{
"epoch": 0.5,
"learning_rate": 9.828237718996907e-07,
"loss": 7.9979,
"step": 1500
},
{
"epoch": 0.53,
"learning_rate": 9.79388526279629e-07,
"loss": 7.8445,
"step": 1600
},
{
"epoch": 0.56,
"learning_rate": 9.759532806595672e-07,
"loss": 7.7416,
"step": 1700
},
{
"epoch": 0.6,
"learning_rate": 9.725180350395053e-07,
"loss": 7.649,
"step": 1800
},
{
"epoch": 0.63,
"learning_rate": 9.690827894194434e-07,
"loss": 7.5877,
"step": 1900
},
{
"epoch": 0.66,
"learning_rate": 9.656475437993815e-07,
"loss": 7.4996,
"step": 2000
},
{
"epoch": 0.7,
"learning_rate": 9.622122981793198e-07,
"loss": 7.412,
"step": 2100
},
{
"epoch": 0.73,
"learning_rate": 9.58777052559258e-07,
"loss": 7.4068,
"step": 2200
},
{
"epoch": 0.76,
"learning_rate": 9.55341806939196e-07,
"loss": 7.3062,
"step": 2300
},
{
"epoch": 0.8,
"learning_rate": 9.519065613191343e-07,
"loss": 7.2333,
"step": 2400
},
{
"epoch": 0.83,
"learning_rate": 9.484713156990724e-07,
"loss": 7.09,
"step": 2500
},
{
"epoch": 0.86,
"learning_rate": 9.450360700790106e-07,
"loss": 7.0942,
"step": 2600
},
{
"epoch": 0.9,
"learning_rate": 9.416008244589487e-07,
"loss": 7.0845,
"step": 2700
},
{
"epoch": 0.93,
"learning_rate": 9.38165578838887e-07,
"loss": 6.8747,
"step": 2800
},
{
"epoch": 0.96,
"learning_rate": 9.347303332188251e-07,
"loss": 7.0027,
"step": 2900
},
{
"epoch": 1.0,
"learning_rate": 9.312950875987633e-07,
"loss": 6.9904,
"step": 3000
},
{
"epoch": 1.03,
"learning_rate": 9.278598419787015e-07,
"loss": 6.9223,
"step": 3100
},
{
"epoch": 1.06,
"learning_rate": 9.244245963586395e-07,
"loss": 6.8921,
"step": 3200
},
{
"epoch": 1.1,
"learning_rate": 9.209893507385778e-07,
"loss": 6.8751,
"step": 3300
},
{
"epoch": 1.13,
"learning_rate": 9.17554105118516e-07,
"loss": 6.7727,
"step": 3400
},
{
"epoch": 1.16,
"learning_rate": 9.141188594984541e-07,
"loss": 6.8281,
"step": 3500
},
{
"epoch": 1.2,
"learning_rate": 9.106836138783922e-07,
"loss": 6.8361,
"step": 3600
},
{
"epoch": 1.23,
"learning_rate": 9.072483682583304e-07,
"loss": 6.7184,
"step": 3700
},
{
"epoch": 1.26,
"learning_rate": 9.038131226382685e-07,
"loss": 6.8271,
"step": 3800
},
{
"epoch": 1.3,
"learning_rate": 9.003778770182068e-07,
"loss": 6.7768,
"step": 3900
},
{
"epoch": 1.33,
"learning_rate": 8.96942631398145e-07,
"loss": 6.7108,
"step": 4000
},
{
"epoch": 1.36,
"learning_rate": 8.935073857780831e-07,
"loss": 6.6765,
"step": 4100
},
{
"epoch": 1.39,
"learning_rate": 8.900721401580212e-07,
"loss": 6.6567,
"step": 4200
},
{
"epoch": 1.43,
"learning_rate": 8.866368945379594e-07,
"loss": 6.7092,
"step": 4300
},
{
"epoch": 1.46,
"learning_rate": 8.832016489178976e-07,
"loss": 6.7071,
"step": 4400
},
{
"epoch": 1.49,
"learning_rate": 8.797664032978358e-07,
"loss": 6.5984,
"step": 4500
},
{
"epoch": 1.53,
"learning_rate": 8.763311576777739e-07,
"loss": 6.5961,
"step": 4600
},
{
"epoch": 1.56,
"learning_rate": 8.728959120577121e-07,
"loss": 6.5279,
"step": 4700
},
{
"epoch": 1.59,
"learning_rate": 8.694606664376502e-07,
"loss": 6.55,
"step": 4800
},
{
"epoch": 1.63,
"learning_rate": 8.660254208175884e-07,
"loss": 6.5216,
"step": 4900
},
{
"epoch": 1.66,
"learning_rate": 8.625901751975267e-07,
"loss": 6.5823,
"step": 5000
},
{
"epoch": 1.69,
"learning_rate": 8.591549295774648e-07,
"loss": 6.4402,
"step": 5100
},
{
"epoch": 1.73,
"learning_rate": 8.557196839574029e-07,
"loss": 6.4921,
"step": 5200
},
{
"epoch": 1.76,
"learning_rate": 8.522844383373411e-07,
"loss": 6.4431,
"step": 5300
},
{
"epoch": 1.79,
"learning_rate": 8.488491927172793e-07,
"loss": 6.4583,
"step": 5400
},
{
"epoch": 1.83,
"learning_rate": 8.454139470972173e-07,
"loss": 6.408,
"step": 5500
},
{
"epoch": 1.86,
"learning_rate": 8.419787014771555e-07,
"loss": 6.3809,
"step": 5600
},
{
"epoch": 1.89,
"learning_rate": 8.385434558570938e-07,
"loss": 6.5188,
"step": 5700
},
{
"epoch": 1.93,
"learning_rate": 8.351082102370319e-07,
"loss": 6.3411,
"step": 5800
},
{
"epoch": 1.96,
"learning_rate": 8.316729646169701e-07,
"loss": 6.3358,
"step": 5900
},
{
"epoch": 1.99,
"learning_rate": 8.282377189969082e-07,
"loss": 6.4091,
"step": 6000
},
{
"epoch": 2.03,
"learning_rate": 8.248024733768464e-07,
"loss": 6.2717,
"step": 6100
},
{
"epoch": 2.06,
"learning_rate": 8.213672277567846e-07,
"loss": 6.3028,
"step": 6200
},
{
"epoch": 2.09,
"learning_rate": 8.179319821367228e-07,
"loss": 6.354,
"step": 6300
},
{
"epoch": 2.13,
"learning_rate": 8.14496736516661e-07,
"loss": 6.2591,
"step": 6400
},
{
"epoch": 2.16,
"learning_rate": 8.11061490896599e-07,
"loss": 6.1371,
"step": 6500
},
{
"epoch": 2.19,
"learning_rate": 8.076262452765372e-07,
"loss": 6.2945,
"step": 6600
},
{
"epoch": 2.23,
"learning_rate": 8.041909996564754e-07,
"loss": 6.3096,
"step": 6700
},
{
"epoch": 2.26,
"learning_rate": 8.007557540364136e-07,
"loss": 6.2438,
"step": 6800
},
{
"epoch": 2.29,
"learning_rate": 7.973205084163517e-07,
"loss": 6.2362,
"step": 6900
},
{
"epoch": 2.32,
"learning_rate": 7.938852627962899e-07,
"loss": 6.3412,
"step": 7000
},
{
"epoch": 2.36,
"learning_rate": 7.90450017176228e-07,
"loss": 6.1895,
"step": 7100
},
{
"epoch": 2.39,
"learning_rate": 7.870147715561662e-07,
"loss": 6.263,
"step": 7200
},
{
"epoch": 2.42,
"learning_rate": 7.835795259361044e-07,
"loss": 6.183,
"step": 7300
},
{
"epoch": 2.46,
"learning_rate": 7.801442803160426e-07,
"loss": 6.1913,
"step": 7400
},
{
"epoch": 2.49,
"learning_rate": 7.767090346959807e-07,
"loss": 6.2258,
"step": 7500
},
{
"epoch": 2.52,
"learning_rate": 7.732737890759189e-07,
"loss": 6.1109,
"step": 7600
},
{
"epoch": 2.56,
"learning_rate": 7.698385434558571e-07,
"loss": 6.1534,
"step": 7700
},
{
"epoch": 2.59,
"learning_rate": 7.664032978357952e-07,
"loss": 6.2051,
"step": 7800
},
{
"epoch": 2.62,
"learning_rate": 7.629680522157333e-07,
"loss": 6.2167,
"step": 7900
},
{
"epoch": 2.66,
"learning_rate": 7.595328065956716e-07,
"loss": 6.1834,
"step": 8000
},
{
"epoch": 2.69,
"learning_rate": 7.560975609756097e-07,
"loss": 6.1025,
"step": 8100
},
{
"epoch": 2.72,
"learning_rate": 7.526623153555479e-07,
"loss": 6.1659,
"step": 8200
},
{
"epoch": 2.76,
"learning_rate": 7.492270697354861e-07,
"loss": 6.153,
"step": 8300
},
{
"epoch": 2.79,
"learning_rate": 7.457918241154242e-07,
"loss": 6.1415,
"step": 8400
},
{
"epoch": 2.82,
"learning_rate": 7.423565784953624e-07,
"loss": 6.2841,
"step": 8500
},
{
"epoch": 2.86,
"learning_rate": 7.389213328753006e-07,
"loss": 6.14,
"step": 8600
},
{
"epoch": 2.89,
"learning_rate": 7.354860872552388e-07,
"loss": 6.1218,
"step": 8700
},
{
"epoch": 2.92,
"learning_rate": 7.320508416351768e-07,
"loss": 6.1991,
"step": 8800
},
{
"epoch": 2.96,
"learning_rate": 7.28615596015115e-07,
"loss": 5.9527,
"step": 8900
},
{
"epoch": 2.99,
"learning_rate": 7.251803503950532e-07,
"loss": 6.1217,
"step": 9000
},
{
"epoch": 3.02,
"learning_rate": 7.217451047749914e-07,
"loss": 6.0276,
"step": 9100
},
{
"epoch": 3.06,
"learning_rate": 7.183098591549296e-07,
"loss": 6.0417,
"step": 9200
},
{
"epoch": 3.09,
"learning_rate": 7.148746135348677e-07,
"loss": 6.1444,
"step": 9300
},
{
"epoch": 3.12,
"learning_rate": 7.114393679148058e-07,
"loss": 5.9651,
"step": 9400
},
{
"epoch": 3.16,
"learning_rate": 7.08004122294744e-07,
"loss": 6.1054,
"step": 9500
},
{
"epoch": 3.19,
"learning_rate": 7.045688766746822e-07,
"loss": 5.9115,
"step": 9600
},
{
"epoch": 3.22,
"learning_rate": 7.011336310546205e-07,
"loss": 6.0053,
"step": 9700
},
{
"epoch": 3.25,
"learning_rate": 6.976983854345585e-07,
"loss": 5.9664,
"step": 9800
},
{
"epoch": 3.29,
"learning_rate": 6.942631398144967e-07,
"loss": 6.0903,
"step": 9900
},
{
"epoch": 3.32,
"learning_rate": 6.908278941944349e-07,
"loss": 6.1041,
"step": 10000
},
{
"epoch": 3.35,
"learning_rate": 6.87392648574373e-07,
"loss": 6.151,
"step": 10100
},
{
"epoch": 3.39,
"learning_rate": 6.839574029543112e-07,
"loss": 5.9338,
"step": 10200
},
{
"epoch": 3.42,
"learning_rate": 6.805221573342494e-07,
"loss": 6.1562,
"step": 10300
},
{
"epoch": 3.45,
"learning_rate": 6.770869117141875e-07,
"loss": 5.9743,
"step": 10400
},
{
"epoch": 3.49,
"learning_rate": 6.736516660941257e-07,
"loss": 6.0068,
"step": 10500
},
{
"epoch": 3.52,
"learning_rate": 6.702164204740639e-07,
"loss": 6.061,
"step": 10600
},
{
"epoch": 3.55,
"learning_rate": 6.66781174854002e-07,
"loss": 6.077,
"step": 10700
},
{
"epoch": 3.59,
"learning_rate": 6.633459292339401e-07,
"loss": 5.9978,
"step": 10800
},
{
"epoch": 3.62,
"learning_rate": 6.599106836138784e-07,
"loss": 6.1817,
"step": 10900
},
{
"epoch": 3.65,
"learning_rate": 6.564754379938166e-07,
"loss": 6.1392,
"step": 11000
}
],
"max_steps": 30110,
"num_train_epochs": 10,
"start_time": 1683297044.048102,
"total_flos": 1.1772773793792e+19,
"trial_name": null,
"trial_params": null
}