text2sql_wizard_13b / trainer_state.json
chuqian.wbc
commit from root
638b60f
raw
history blame contribute delete
No virus
55.9 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"global_step": 924,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 0.0,
"loss": 1.5234,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 0.0,
"loss": 1.5583,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 0.0,
"loss": 1.5219,
"step": 6
},
{
"epoch": 0.03,
"learning_rate": 0.0,
"loss": 1.4819,
"step": 8
},
{
"epoch": 0.03,
"learning_rate": 0.0,
"loss": 1.5571,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 0.0,
"loss": 1.5618,
"step": 12
},
{
"epoch": 0.05,
"learning_rate": 0.0,
"loss": 1.5302,
"step": 14
},
{
"epoch": 0.05,
"learning_rate": 0.0,
"loss": 1.399,
"step": 16
},
{
"epoch": 0.06,
"learning_rate": 2.0000000000000003e-06,
"loss": 1.3733,
"step": 18
},
{
"epoch": 0.06,
"learning_rate": 6e-06,
"loss": 1.6481,
"step": 20
},
{
"epoch": 0.07,
"learning_rate": 1e-05,
"loss": 1.4742,
"step": 22
},
{
"epoch": 0.08,
"learning_rate": 1.4e-05,
"loss": 1.3546,
"step": 24
},
{
"epoch": 0.08,
"learning_rate": 1.6000000000000003e-05,
"loss": 1.3144,
"step": 26
},
{
"epoch": 0.09,
"learning_rate": 2e-05,
"loss": 1.5075,
"step": 28
},
{
"epoch": 0.1,
"learning_rate": 1.9999763715279134e-05,
"loss": 1.4149,
"step": 30
},
{
"epoch": 0.1,
"learning_rate": 1.9999054872282624e-05,
"loss": 1.228,
"step": 32
},
{
"epoch": 0.11,
"learning_rate": 1.9997873504508222e-05,
"loss": 1.3756,
"step": 34
},
{
"epoch": 0.12,
"learning_rate": 1.9996219667783766e-05,
"loss": 1.5109,
"step": 36
},
{
"epoch": 0.12,
"learning_rate": 1.9994093440264524e-05,
"loss": 1.4623,
"step": 38
},
{
"epoch": 0.13,
"learning_rate": 1.9991494922429505e-05,
"loss": 1.4063,
"step": 40
},
{
"epoch": 0.14,
"learning_rate": 1.9988424237076728e-05,
"loss": 1.4229,
"step": 42
},
{
"epoch": 0.14,
"learning_rate": 1.9984881529317394e-05,
"loss": 1.4154,
"step": 44
},
{
"epoch": 0.15,
"learning_rate": 1.9980866966569053e-05,
"loss": 1.3689,
"step": 46
},
{
"epoch": 0.16,
"learning_rate": 1.9976380738547667e-05,
"loss": 1.4078,
"step": 48
},
{
"epoch": 0.16,
"learning_rate": 1.9971423057258664e-05,
"loss": 1.4648,
"step": 50
},
{
"epoch": 0.17,
"learning_rate": 1.9965994156986913e-05,
"loss": 1.3209,
"step": 52
},
{
"epoch": 0.18,
"learning_rate": 1.996310307903586e-05,
"loss": 1.5759,
"step": 54
},
{
"epoch": 0.18,
"learning_rate": 1.9956967838282882e-05,
"loss": 1.3839,
"step": 56
},
{
"epoch": 0.19,
"learning_rate": 1.9950362061656633e-05,
"loss": 1.3,
"step": 58
},
{
"epoch": 0.19,
"learning_rate": 1.9943286061325932e-05,
"loss": 1.4333,
"step": 60
},
{
"epoch": 0.2,
"learning_rate": 1.9935740171680926e-05,
"loss": 1.4935,
"step": 62
},
{
"epoch": 0.21,
"learning_rate": 1.99277247493173e-05,
"loss": 1.3597,
"step": 64
},
{
"epoch": 0.21,
"learning_rate": 1.991924017301943e-05,
"loss": 1.5478,
"step": 66
},
{
"epoch": 0.22,
"learning_rate": 1.9910286843742458e-05,
"loss": 1.4124,
"step": 68
},
{
"epoch": 0.23,
"learning_rate": 1.9900865184593364e-05,
"loss": 1.4587,
"step": 70
},
{
"epoch": 0.23,
"learning_rate": 1.9890975640810973e-05,
"loss": 1.4449,
"step": 72
},
{
"epoch": 0.24,
"learning_rate": 1.98806186797449e-05,
"loss": 1.4994,
"step": 74
},
{
"epoch": 0.25,
"learning_rate": 1.9869794790833478e-05,
"loss": 1.4709,
"step": 76
},
{
"epoch": 0.25,
"learning_rate": 1.9858504485580618e-05,
"loss": 1.336,
"step": 78
},
{
"epoch": 0.26,
"learning_rate": 1.984674829753165e-05,
"loss": 1.4401,
"step": 80
},
{
"epoch": 0.27,
"learning_rate": 1.9834526782248092e-05,
"loss": 1.4636,
"step": 82
},
{
"epoch": 0.27,
"learning_rate": 1.9821840517281415e-05,
"loss": 1.5029,
"step": 84
},
{
"epoch": 0.28,
"learning_rate": 1.9808690102145727e-05,
"loss": 1.5737,
"step": 86
},
{
"epoch": 0.29,
"learning_rate": 1.9795076158289466e-05,
"loss": 1.4333,
"step": 88
},
{
"epoch": 0.29,
"learning_rate": 1.9780999329066018e-05,
"loss": 1.472,
"step": 90
},
{
"epoch": 0.3,
"learning_rate": 1.9766460279703312e-05,
"loss": 1.5217,
"step": 92
},
{
"epoch": 0.31,
"learning_rate": 1.9751459697272396e-05,
"loss": 1.3852,
"step": 94
},
{
"epoch": 0.31,
"learning_rate": 1.973599829065495e-05,
"loss": 1.4694,
"step": 96
},
{
"epoch": 0.32,
"learning_rate": 1.9720076790509814e-05,
"loss": 1.2218,
"step": 98
},
{
"epoch": 0.32,
"learning_rate": 1.970369594923842e-05,
"loss": 1.3959,
"step": 100
},
{
"epoch": 0.33,
"learning_rate": 1.968685654094928e-05,
"loss": 1.2564,
"step": 102
},
{
"epoch": 0.34,
"learning_rate": 1.9669559361421366e-05,
"loss": 1.3412,
"step": 104
},
{
"epoch": 0.34,
"learning_rate": 1.9651805228066522e-05,
"loss": 1.5168,
"step": 106
},
{
"epoch": 0.35,
"learning_rate": 1.9633594979890842e-05,
"loss": 1.4346,
"step": 108
},
{
"epoch": 0.36,
"learning_rate": 1.9614929477455007e-05,
"loss": 1.3223,
"step": 110
},
{
"epoch": 0.36,
"learning_rate": 1.959580960283362e-05,
"loss": 1.4993,
"step": 112
},
{
"epoch": 0.37,
"learning_rate": 1.9576236259573537e-05,
"loss": 1.5611,
"step": 114
},
{
"epoch": 0.38,
"learning_rate": 1.9556210372651134e-05,
"loss": 1.4156,
"step": 116
},
{
"epoch": 0.38,
"learning_rate": 1.9535732888428644e-05,
"loss": 1.3979,
"step": 118
},
{
"epoch": 0.39,
"learning_rate": 1.951480477460939e-05,
"loss": 1.3124,
"step": 120
},
{
"epoch": 0.4,
"learning_rate": 1.949342702019208e-05,
"loss": 1.4111,
"step": 122
},
{
"epoch": 0.4,
"learning_rate": 1.9471600635424058e-05,
"loss": 1.4656,
"step": 124
},
{
"epoch": 0.41,
"learning_rate": 1.9449326651753572e-05,
"loss": 1.2345,
"step": 126
},
{
"epoch": 0.42,
"learning_rate": 1.9426606121781028e-05,
"loss": 1.3033,
"step": 128
},
{
"epoch": 0.42,
"learning_rate": 1.940344011920924e-05,
"loss": 1.3654,
"step": 130
},
{
"epoch": 0.43,
"learning_rate": 1.93798297387927e-05,
"loss": 1.2962,
"step": 132
},
{
"epoch": 0.44,
"learning_rate": 1.9355776096285833e-05,
"loss": 1.5612,
"step": 134
},
{
"epoch": 0.44,
"learning_rate": 1.9331280328390284e-05,
"loss": 1.4371,
"step": 136
},
{
"epoch": 0.45,
"learning_rate": 1.9306343592701187e-05,
"loss": 1.5658,
"step": 138
},
{
"epoch": 0.45,
"learning_rate": 1.928096706765247e-05,
"loss": 1.3471,
"step": 140
},
{
"epoch": 0.46,
"learning_rate": 1.925515195246116e-05,
"loss": 1.3567,
"step": 142
},
{
"epoch": 0.47,
"learning_rate": 1.9228899467070713e-05,
"loss": 1.3916,
"step": 144
},
{
"epoch": 0.47,
"learning_rate": 1.9202210852093367e-05,
"loss": 1.4113,
"step": 146
},
{
"epoch": 0.48,
"learning_rate": 1.917508736875151e-05,
"loss": 1.4357,
"step": 148
},
{
"epoch": 0.49,
"learning_rate": 1.914753029881808e-05,
"loss": 1.4278,
"step": 150
},
{
"epoch": 0.49,
"learning_rate": 1.9119540944555992e-05,
"loss": 1.2472,
"step": 152
},
{
"epoch": 0.5,
"learning_rate": 1.9091120628656597e-05,
"loss": 1.3739,
"step": 154
},
{
"epoch": 0.51,
"learning_rate": 1.906227069417718e-05,
"loss": 1.3371,
"step": 156
},
{
"epoch": 0.51,
"learning_rate": 1.9032992504477475e-05,
"loss": 1.362,
"step": 158
},
{
"epoch": 0.52,
"learning_rate": 1.9003287443155274e-05,
"loss": 1.3747,
"step": 160
},
{
"epoch": 0.53,
"learning_rate": 1.8973156913980988e-05,
"loss": 1.428,
"step": 162
},
{
"epoch": 0.53,
"learning_rate": 1.8942602340831356e-05,
"loss": 1.4416,
"step": 164
},
{
"epoch": 0.54,
"learning_rate": 1.891162516762214e-05,
"loss": 1.3246,
"step": 166
},
{
"epoch": 0.55,
"learning_rate": 1.8880226858239878e-05,
"loss": 1.3395,
"step": 168
},
{
"epoch": 0.55,
"learning_rate": 1.8848408896472728e-05,
"loss": 1.4583,
"step": 170
},
{
"epoch": 0.56,
"learning_rate": 1.8816172785940335e-05,
"loss": 1.36,
"step": 172
},
{
"epoch": 0.56,
"learning_rate": 1.878352005002277e-05,
"loss": 1.4412,
"step": 174
},
{
"epoch": 0.57,
"learning_rate": 1.875045223178855e-05,
"loss": 1.3123,
"step": 176
},
{
"epoch": 0.58,
"learning_rate": 1.871697089392172e-05,
"loss": 1.4006,
"step": 178
},
{
"epoch": 0.58,
"learning_rate": 1.8683077618647997e-05,
"loss": 1.2854,
"step": 180
},
{
"epoch": 0.59,
"learning_rate": 1.8648774007659986e-05,
"loss": 1.5193,
"step": 182
},
{
"epoch": 0.6,
"learning_rate": 1.8614061682041532e-05,
"loss": 1.2914,
"step": 184
},
{
"epoch": 0.6,
"learning_rate": 1.8578942282191057e-05,
"loss": 1.3192,
"step": 186
},
{
"epoch": 0.61,
"learning_rate": 1.8543417467744085e-05,
"loss": 1.428,
"step": 188
},
{
"epoch": 0.62,
"learning_rate": 1.8507488917494788e-05,
"loss": 1.4875,
"step": 190
},
{
"epoch": 0.62,
"learning_rate": 1.847115832931666e-05,
"loss": 1.6362,
"step": 192
},
{
"epoch": 0.63,
"learning_rate": 1.8434427420082278e-05,
"loss": 1.2374,
"step": 194
},
{
"epoch": 0.64,
"learning_rate": 1.8397297925582167e-05,
"loss": 1.5176,
"step": 196
},
{
"epoch": 0.64,
"learning_rate": 1.8359771600442778e-05,
"loss": 1.3912,
"step": 198
},
{
"epoch": 0.65,
"learning_rate": 1.832185021804356e-05,
"loss": 1.5812,
"step": 200
},
{
"epoch": 0.66,
"learning_rate": 1.828353557043317e-05,
"loss": 1.3373,
"step": 202
},
{
"epoch": 0.66,
"learning_rate": 1.8244829468244768e-05,
"loss": 1.3064,
"step": 204
},
{
"epoch": 0.67,
"learning_rate": 1.8205733740610462e-05,
"loss": 1.5292,
"step": 206
},
{
"epoch": 0.68,
"learning_rate": 1.8166250235074874e-05,
"loss": 1.471,
"step": 208
},
{
"epoch": 0.68,
"learning_rate": 1.8126380817507817e-05,
"loss": 1.4468,
"step": 210
},
{
"epoch": 0.69,
"learning_rate": 1.8086127372016132e-05,
"loss": 1.2668,
"step": 212
},
{
"epoch": 0.69,
"learning_rate": 1.804549180085465e-05,
"loss": 1.416,
"step": 214
},
{
"epoch": 0.7,
"learning_rate": 1.8004476024336284e-05,
"loss": 1.5767,
"step": 216
},
{
"epoch": 0.71,
"learning_rate": 1.7963081980741298e-05,
"loss": 1.2952,
"step": 218
},
{
"epoch": 0.71,
"learning_rate": 1.79213116262257e-05,
"loss": 1.5002,
"step": 220
},
{
"epoch": 0.72,
"learning_rate": 1.78791669347288e-05,
"loss": 1.2971,
"step": 222
},
{
"epoch": 0.73,
"learning_rate": 1.7836649897879924e-05,
"loss": 1.2372,
"step": 224
},
{
"epoch": 0.73,
"learning_rate": 1.779376252490432e-05,
"loss": 1.3897,
"step": 226
},
{
"epoch": 0.74,
"learning_rate": 1.7750506842528174e-05,
"loss": 1.4017,
"step": 228
},
{
"epoch": 0.75,
"learning_rate": 1.7706884894882853e-05,
"loss": 1.4475,
"step": 230
},
{
"epoch": 0.75,
"learning_rate": 1.76628987434083e-05,
"loss": 1.4873,
"step": 232
},
{
"epoch": 0.76,
"learning_rate": 1.7618550466755626e-05,
"loss": 1.4008,
"step": 234
},
{
"epoch": 0.77,
"learning_rate": 1.7573842160688858e-05,
"loss": 1.3481,
"step": 236
},
{
"epoch": 0.77,
"learning_rate": 1.752877593798592e-05,
"loss": 1.4984,
"step": 238
},
{
"epoch": 0.78,
"learning_rate": 1.748335392833879e-05,
"loss": 1.5237,
"step": 240
},
{
"epoch": 0.79,
"learning_rate": 1.7437578278252834e-05,
"loss": 1.4043,
"step": 242
},
{
"epoch": 0.79,
"learning_rate": 1.7391451150945395e-05,
"loss": 1.3393,
"step": 244
},
{
"epoch": 0.8,
"learning_rate": 1.7344974726243555e-05,
"loss": 1.3782,
"step": 246
},
{
"epoch": 0.81,
"learning_rate": 1.729815120048112e-05,
"loss": 1.3935,
"step": 248
},
{
"epoch": 0.81,
"learning_rate": 1.7250982786394833e-05,
"loss": 1.4633,
"step": 250
},
{
"epoch": 0.82,
"learning_rate": 1.7203471713019805e-05,
"loss": 1.2838,
"step": 252
},
{
"epoch": 0.82,
"learning_rate": 1.7155620225584176e-05,
"loss": 1.3142,
"step": 254
},
{
"epoch": 0.83,
"learning_rate": 1.7107430585403024e-05,
"loss": 1.2973,
"step": 256
},
{
"epoch": 0.84,
"learning_rate": 1.7058905069771477e-05,
"loss": 1.4129,
"step": 258
},
{
"epoch": 0.84,
"learning_rate": 1.7010045971857117e-05,
"loss": 1.3484,
"step": 260
},
{
"epoch": 0.85,
"learning_rate": 1.696085560059162e-05,
"loss": 1.5596,
"step": 262
},
{
"epoch": 0.86,
"learning_rate": 1.6911336280561595e-05,
"loss": 1.2894,
"step": 264
},
{
"epoch": 0.86,
"learning_rate": 1.6861490351898798e-05,
"loss": 1.2303,
"step": 266
},
{
"epoch": 0.87,
"learning_rate": 1.6811320170169496e-05,
"loss": 1.5192,
"step": 268
},
{
"epoch": 0.88,
"learning_rate": 1.676082810626316e-05,
"loss": 1.3415,
"step": 270
},
{
"epoch": 0.88,
"learning_rate": 1.6710016546280443e-05,
"loss": 1.3893,
"step": 272
},
{
"epoch": 0.89,
"learning_rate": 1.6658887891420393e-05,
"loss": 1.4108,
"step": 274
},
{
"epoch": 0.9,
"learning_rate": 1.6607444557866995e-05,
"loss": 1.401,
"step": 276
},
{
"epoch": 0.9,
"learning_rate": 1.6555688976675e-05,
"loss": 1.5258,
"step": 278
},
{
"epoch": 0.91,
"learning_rate": 1.6503623593655013e-05,
"loss": 1.4698,
"step": 280
},
{
"epoch": 0.92,
"learning_rate": 1.645125086925794e-05,
"loss": 1.3611,
"step": 282
},
{
"epoch": 0.92,
"learning_rate": 1.6398573278458682e-05,
"loss": 1.2556,
"step": 284
},
{
"epoch": 0.93,
"learning_rate": 1.6345593310639212e-05,
"loss": 1.3739,
"step": 286
},
{
"epoch": 0.94,
"learning_rate": 1.6292313469470917e-05,
"loss": 1.3527,
"step": 288
},
{
"epoch": 0.94,
"learning_rate": 1.6238736272796267e-05,
"loss": 1.358,
"step": 290
},
{
"epoch": 0.95,
"learning_rate": 1.6184864252509866e-05,
"loss": 1.3874,
"step": 292
},
{
"epoch": 0.95,
"learning_rate": 1.6130699954438756e-05,
"loss": 1.4507,
"step": 294
},
{
"epoch": 0.96,
"learning_rate": 1.607624593822216e-05,
"loss": 1.4701,
"step": 296
},
{
"epoch": 0.97,
"learning_rate": 1.602150477719047e-05,
"loss": 1.4038,
"step": 298
},
{
"epoch": 0.97,
"learning_rate": 1.5966479058243686e-05,
"loss": 1.3851,
"step": 300
},
{
"epoch": 0.98,
"learning_rate": 1.5911171381729135e-05,
"loss": 1.3512,
"step": 302
},
{
"epoch": 0.99,
"learning_rate": 1.5855584361318594e-05,
"loss": 1.3924,
"step": 304
},
{
"epoch": 0.99,
"learning_rate": 1.579972062388479e-05,
"loss": 1.3825,
"step": 306
},
{
"epoch": 1.0,
"learning_rate": 1.5743582809377237e-05,
"loss": 1.3565,
"step": 308
},
{
"epoch": 1.01,
"learning_rate": 1.5687173570697505e-05,
"loss": 0.9095,
"step": 310
},
{
"epoch": 1.01,
"learning_rate": 1.5630495573573837e-05,
"loss": 1.0694,
"step": 312
},
{
"epoch": 1.02,
"learning_rate": 1.557355149643518e-05,
"loss": 1.133,
"step": 314
},
{
"epoch": 1.03,
"learning_rate": 1.5516344030284605e-05,
"loss": 0.9732,
"step": 316
},
{
"epoch": 1.03,
"learning_rate": 1.545887587857215e-05,
"loss": 0.9751,
"step": 318
},
{
"epoch": 1.04,
"learning_rate": 1.540114975706705e-05,
"loss": 0.9463,
"step": 320
},
{
"epoch": 1.05,
"learning_rate": 1.5343168393729406e-05,
"loss": 0.9389,
"step": 322
},
{
"epoch": 1.05,
"learning_rate": 1.528493452858127e-05,
"loss": 0.8876,
"step": 324
},
{
"epoch": 1.06,
"learning_rate": 1.522645091357716e-05,
"loss": 0.8831,
"step": 326
},
{
"epoch": 1.06,
"learning_rate": 1.5167720312473995e-05,
"loss": 0.8689,
"step": 328
},
{
"epoch": 1.07,
"learning_rate": 1.510874550070052e-05,
"loss": 0.8803,
"step": 330
},
{
"epoch": 1.08,
"learning_rate": 1.504952926522612e-05,
"loss": 0.8584,
"step": 332
},
{
"epoch": 1.08,
"learning_rate": 1.4990074404429131e-05,
"loss": 0.8771,
"step": 334
},
{
"epoch": 1.09,
"learning_rate": 1.493038372796459e-05,
"loss": 0.8431,
"step": 336
},
{
"epoch": 1.1,
"learning_rate": 1.4870460056631464e-05,
"loss": 0.8739,
"step": 338
},
{
"epoch": 1.1,
"learning_rate": 1.4810306222239337e-05,
"loss": 0.9607,
"step": 340
},
{
"epoch": 1.11,
"learning_rate": 1.4749925067474612e-05,
"loss": 0.9771,
"step": 342
},
{
"epoch": 1.12,
"learning_rate": 1.4689319445766144e-05,
"loss": 0.7807,
"step": 344
},
{
"epoch": 1.12,
"learning_rate": 1.4628492221150411e-05,
"loss": 1.0123,
"step": 346
},
{
"epoch": 1.13,
"learning_rate": 1.4567446268136179e-05,
"loss": 0.9227,
"step": 348
},
{
"epoch": 1.14,
"learning_rate": 1.4506184471568635e-05,
"loss": 0.9247,
"step": 350
},
{
"epoch": 1.14,
"learning_rate": 1.4444709726493085e-05,
"loss": 0.8943,
"step": 352
},
{
"epoch": 1.15,
"learning_rate": 1.438302493801812e-05,
"loss": 0.855,
"step": 354
},
{
"epoch": 1.16,
"learning_rate": 1.4321133021178349e-05,
"loss": 0.9835,
"step": 356
},
{
"epoch": 1.16,
"learning_rate": 1.4259036900796628e-05,
"loss": 0.8994,
"step": 358
},
{
"epoch": 1.17,
"learning_rate": 1.4196739511345851e-05,
"loss": 0.8497,
"step": 360
},
{
"epoch": 1.18,
"learning_rate": 1.4134243796810277e-05,
"loss": 0.9333,
"step": 362
},
{
"epoch": 1.18,
"learning_rate": 1.40715527105464e-05,
"loss": 0.935,
"step": 364
},
{
"epoch": 1.19,
"learning_rate": 1.4008669215143376e-05,
"loss": 0.7321,
"step": 366
},
{
"epoch": 1.19,
"learning_rate": 1.3945596282283043e-05,
"loss": 0.9281,
"step": 368
},
{
"epoch": 1.2,
"learning_rate": 1.3882336892599469e-05,
"loss": 0.9498,
"step": 370
},
{
"epoch": 1.21,
"learning_rate": 1.3818894035538093e-05,
"loss": 0.8647,
"step": 372
},
{
"epoch": 1.21,
"learning_rate": 1.3755270709214476e-05,
"loss": 1.0073,
"step": 374
},
{
"epoch": 1.22,
"learning_rate": 1.3691469920272603e-05,
"loss": 0.9846,
"step": 376
},
{
"epoch": 1.23,
"learning_rate": 1.362749468374278e-05,
"loss": 0.8693,
"step": 378
},
{
"epoch": 1.23,
"learning_rate": 1.3563348022899203e-05,
"loss": 0.9208,
"step": 380
},
{
"epoch": 1.24,
"learning_rate": 1.3499032969117036e-05,
"loss": 0.9223,
"step": 382
},
{
"epoch": 1.25,
"learning_rate": 1.3434552561729181e-05,
"loss": 0.9236,
"step": 384
},
{
"epoch": 1.25,
"learning_rate": 1.336990984788266e-05,
"loss": 0.9974,
"step": 386
},
{
"epoch": 1.26,
"learning_rate": 1.3305107882394587e-05,
"loss": 0.8629,
"step": 388
},
{
"epoch": 1.27,
"learning_rate": 1.3240149727607827e-05,
"loss": 0.9514,
"step": 390
},
{
"epoch": 1.27,
"learning_rate": 1.3175038453246274e-05,
"loss": 0.982,
"step": 392
},
{
"epoch": 1.28,
"learning_rate": 1.3109777136269788e-05,
"loss": 0.9342,
"step": 394
},
{
"epoch": 1.29,
"learning_rate": 1.3044368860728781e-05,
"loss": 0.7236,
"step": 396
},
{
"epoch": 1.29,
"learning_rate": 1.2978816717618479e-05,
"loss": 0.8384,
"step": 398
},
{
"epoch": 1.3,
"learning_rate": 1.2913123804732848e-05,
"loss": 0.7538,
"step": 400
},
{
"epoch": 1.31,
"learning_rate": 1.2847293226518207e-05,
"loss": 0.9229,
"step": 402
},
{
"epoch": 1.31,
"learning_rate": 1.2781328093926514e-05,
"loss": 0.8993,
"step": 404
},
{
"epoch": 1.32,
"learning_rate": 1.2715231524268357e-05,
"loss": 0.9106,
"step": 406
},
{
"epoch": 1.32,
"learning_rate": 1.264900664106564e-05,
"loss": 0.939,
"step": 408
},
{
"epoch": 1.33,
"learning_rate": 1.258265657390397e-05,
"loss": 0.9626,
"step": 410
},
{
"epoch": 1.34,
"learning_rate": 1.2516184458284766e-05,
"loss": 0.8662,
"step": 412
},
{
"epoch": 1.34,
"learning_rate": 1.244959343547709e-05,
"loss": 0.8466,
"step": 414
},
{
"epoch": 1.35,
"learning_rate": 1.2382886652369183e-05,
"loss": 0.8591,
"step": 416
},
{
"epoch": 1.36,
"learning_rate": 1.2316067261319773e-05,
"loss": 0.7773,
"step": 418
},
{
"epoch": 1.36,
"learning_rate": 1.2249138420009095e-05,
"loss": 0.8936,
"step": 420
},
{
"epoch": 1.37,
"learning_rate": 1.218210329128966e-05,
"loss": 0.8882,
"step": 422
},
{
"epoch": 1.38,
"learning_rate": 1.2114965043036808e-05,
"loss": 0.9398,
"step": 424
},
{
"epoch": 1.38,
"learning_rate": 1.2047726847998993e-05,
"loss": 0.964,
"step": 426
},
{
"epoch": 1.39,
"learning_rate": 1.1980391883647832e-05,
"loss": 1.0241,
"step": 428
},
{
"epoch": 1.4,
"learning_rate": 1.1912963332027987e-05,
"loss": 1.0854,
"step": 430
},
{
"epoch": 1.4,
"learning_rate": 1.1845444379606757e-05,
"loss": 0.9673,
"step": 432
},
{
"epoch": 1.41,
"learning_rate": 1.1777838217123498e-05,
"loss": 0.9455,
"step": 434
},
{
"epoch": 1.42,
"learning_rate": 1.1710148039438867e-05,
"loss": 0.9575,
"step": 436
},
{
"epoch": 1.42,
"learning_rate": 1.1642377045383806e-05,
"loss": 0.8251,
"step": 438
},
{
"epoch": 1.43,
"learning_rate": 1.1574528437608404e-05,
"loss": 0.7966,
"step": 440
},
{
"epoch": 1.44,
"learning_rate": 1.150660542243052e-05,
"loss": 0.9938,
"step": 442
},
{
"epoch": 1.44,
"learning_rate": 1.1438611209684296e-05,
"loss": 0.9175,
"step": 444
},
{
"epoch": 1.45,
"learning_rate": 1.1370549012568452e-05,
"loss": 0.832,
"step": 446
},
{
"epoch": 1.45,
"learning_rate": 1.1302422047494432e-05,
"loss": 0.9767,
"step": 448
},
{
"epoch": 1.46,
"learning_rate": 1.1234233533934425e-05,
"loss": 0.859,
"step": 450
},
{
"epoch": 1.47,
"learning_rate": 1.1165986694269208e-05,
"loss": 0.9282,
"step": 452
},
{
"epoch": 1.47,
"learning_rate": 1.109768475363587e-05,
"loss": 0.9586,
"step": 454
},
{
"epoch": 1.48,
"learning_rate": 1.1029330939775411e-05,
"loss": 0.9444,
"step": 456
},
{
"epoch": 1.49,
"learning_rate": 1.0960928482880194e-05,
"loss": 0.972,
"step": 458
},
{
"epoch": 1.49,
"learning_rate": 1.0892480615441308e-05,
"loss": 0.8934,
"step": 460
},
{
"epoch": 1.5,
"learning_rate": 1.08239905720958e-05,
"loss": 0.9087,
"step": 462
},
{
"epoch": 1.51,
"learning_rate": 1.0755461589473829e-05,
"loss": 0.8701,
"step": 464
},
{
"epoch": 1.51,
"learning_rate": 1.0686896906045695e-05,
"loss": 1.0226,
"step": 466
},
{
"epoch": 1.52,
"learning_rate": 1.0618299761968822e-05,
"loss": 0.8852,
"step": 468
},
{
"epoch": 1.53,
"learning_rate": 1.0549673398934615e-05,
"loss": 0.8311,
"step": 470
},
{
"epoch": 1.53,
"learning_rate": 1.0481021060015282e-05,
"loss": 0.8665,
"step": 472
},
{
"epoch": 1.54,
"learning_rate": 1.0412345989510569e-05,
"loss": 0.8921,
"step": 474
},
{
"epoch": 1.55,
"learning_rate": 1.0343651432794446e-05,
"loss": 0.9216,
"step": 476
},
{
"epoch": 1.55,
"learning_rate": 1.0274940636161752e-05,
"loss": 0.9414,
"step": 478
},
{
"epoch": 1.56,
"learning_rate": 1.0206216846674764e-05,
"loss": 0.9655,
"step": 480
},
{
"epoch": 1.56,
"learning_rate": 1.013748331200976e-05,
"loss": 0.9586,
"step": 482
},
{
"epoch": 1.57,
"learning_rate": 1.0068743280303563e-05,
"loss": 0.9755,
"step": 484
},
{
"epoch": 1.58,
"learning_rate": 1e-05,
"loss": 0.888,
"step": 486
},
{
"epoch": 1.58,
"learning_rate": 9.931256719696442e-06,
"loss": 0.8819,
"step": 488
},
{
"epoch": 1.59,
"learning_rate": 9.86251668799024e-06,
"loss": 0.8599,
"step": 490
},
{
"epoch": 1.6,
"learning_rate": 9.79378315332524e-06,
"loss": 0.9035,
"step": 492
},
{
"epoch": 1.6,
"learning_rate": 9.725059363838251e-06,
"loss": 0.8372,
"step": 494
},
{
"epoch": 1.61,
"learning_rate": 9.656348567205556e-06,
"loss": 0.8864,
"step": 496
},
{
"epoch": 1.62,
"learning_rate": 9.587654010489435e-06,
"loss": 0.9798,
"step": 498
},
{
"epoch": 1.62,
"learning_rate": 9.518978939984721e-06,
"loss": 0.8865,
"step": 500
},
{
"epoch": 1.63,
"learning_rate": 9.450326601065387e-06,
"loss": 0.8469,
"step": 502
},
{
"epoch": 1.64,
"learning_rate": 9.381700238031178e-06,
"loss": 0.9409,
"step": 504
},
{
"epoch": 1.64,
"learning_rate": 9.313103093954306e-06,
"loss": 0.8823,
"step": 506
},
{
"epoch": 1.65,
"learning_rate": 9.244538410526176e-06,
"loss": 0.8595,
"step": 508
},
{
"epoch": 1.66,
"learning_rate": 9.1760094279042e-06,
"loss": 0.9537,
"step": 510
},
{
"epoch": 1.66,
"learning_rate": 9.107519384558695e-06,
"loss": 0.9716,
"step": 512
},
{
"epoch": 1.67,
"learning_rate": 9.039071517119808e-06,
"loss": 0.8419,
"step": 514
},
{
"epoch": 1.68,
"learning_rate": 8.970669060224595e-06,
"loss": 0.757,
"step": 516
},
{
"epoch": 1.68,
"learning_rate": 8.902315246364131e-06,
"loss": 0.8714,
"step": 518
},
{
"epoch": 1.69,
"learning_rate": 8.834013305730795e-06,
"loss": 0.7869,
"step": 520
},
{
"epoch": 1.69,
"learning_rate": 8.76576646606558e-06,
"loss": 1.0176,
"step": 522
},
{
"epoch": 1.7,
"learning_rate": 8.697577952505568e-06,
"loss": 0.8037,
"step": 524
},
{
"epoch": 1.71,
"learning_rate": 8.62945098743155e-06,
"loss": 0.9673,
"step": 526
},
{
"epoch": 1.71,
"learning_rate": 8.561388790315707e-06,
"loss": 0.817,
"step": 528
},
{
"epoch": 1.72,
"learning_rate": 8.493394577569482e-06,
"loss": 0.8556,
"step": 530
},
{
"epoch": 1.73,
"learning_rate": 8.425471562391601e-06,
"loss": 0.858,
"step": 532
},
{
"epoch": 1.73,
"learning_rate": 8.357622954616197e-06,
"loss": 0.9486,
"step": 534
},
{
"epoch": 1.74,
"learning_rate": 8.289851960561137e-06,
"loss": 0.89,
"step": 536
},
{
"epoch": 1.75,
"learning_rate": 8.222161782876502e-06,
"loss": 0.861,
"step": 538
},
{
"epoch": 1.75,
"learning_rate": 8.154555620393248e-06,
"loss": 0.8839,
"step": 540
},
{
"epoch": 1.76,
"learning_rate": 8.087036667972018e-06,
"loss": 0.9137,
"step": 542
},
{
"epoch": 1.77,
"learning_rate": 8.019608116352168e-06,
"loss": 1.0312,
"step": 544
},
{
"epoch": 1.77,
"learning_rate": 7.95227315200101e-06,
"loss": 0.8959,
"step": 546
},
{
"epoch": 1.78,
"learning_rate": 7.885034956963195e-06,
"loss": 0.8568,
"step": 548
},
{
"epoch": 1.79,
"learning_rate": 7.81789670871034e-06,
"loss": 0.8631,
"step": 550
},
{
"epoch": 1.79,
"learning_rate": 7.750861579990909e-06,
"loss": 0.8424,
"step": 552
},
{
"epoch": 1.8,
"learning_rate": 7.68393273868023e-06,
"loss": 1.04,
"step": 554
},
{
"epoch": 1.81,
"learning_rate": 7.6171133476308216e-06,
"loss": 0.9083,
"step": 556
},
{
"epoch": 1.81,
"learning_rate": 7.550406564522911e-06,
"loss": 0.9353,
"step": 558
},
{
"epoch": 1.82,
"learning_rate": 7.483815541715236e-06,
"loss": 0.925,
"step": 560
},
{
"epoch": 1.82,
"learning_rate": 7.417343426096034e-06,
"loss": 0.9781,
"step": 562
},
{
"epoch": 1.83,
"learning_rate": 7.350993358934361e-06,
"loss": 0.974,
"step": 564
},
{
"epoch": 1.84,
"learning_rate": 7.284768475731646e-06,
"loss": 0.9654,
"step": 566
},
{
"epoch": 1.84,
"learning_rate": 7.2186719060734915e-06,
"loss": 0.7186,
"step": 568
},
{
"epoch": 1.85,
"learning_rate": 7.1527067734817975e-06,
"loss": 0.8643,
"step": 570
},
{
"epoch": 1.86,
"learning_rate": 7.086876195267154e-06,
"loss": 0.9238,
"step": 572
},
{
"epoch": 1.86,
"learning_rate": 7.021183282381525e-06,
"loss": 0.8739,
"step": 574
},
{
"epoch": 1.87,
"learning_rate": 6.955631139271223e-06,
"loss": 0.8457,
"step": 576
},
{
"epoch": 1.88,
"learning_rate": 6.890222863730214e-06,
"loss": 1.0443,
"step": 578
},
{
"epoch": 1.88,
"learning_rate": 6.824961546753728e-06,
"loss": 0.9861,
"step": 580
},
{
"epoch": 1.89,
"learning_rate": 6.7598502723921766e-06,
"loss": 0.8405,
"step": 582
},
{
"epoch": 1.9,
"learning_rate": 6.694892117605416e-06,
"loss": 1.0291,
"step": 584
},
{
"epoch": 1.9,
"learning_rate": 6.630090152117342e-06,
"loss": 0.8729,
"step": 586
},
{
"epoch": 1.91,
"learning_rate": 6.56544743827082e-06,
"loss": 0.9119,
"step": 588
},
{
"epoch": 1.92,
"learning_rate": 6.50096703088297e-06,
"loss": 0.7965,
"step": 590
},
{
"epoch": 1.92,
"learning_rate": 6.436651977100801e-06,
"loss": 0.9044,
"step": 592
},
{
"epoch": 1.93,
"learning_rate": 6.3725053162572225e-06,
"loss": 0.9219,
"step": 594
},
{
"epoch": 1.94,
"learning_rate": 6.308530079727404e-06,
"loss": 0.8538,
"step": 596
},
{
"epoch": 1.94,
"learning_rate": 6.244729290785525e-06,
"loss": 0.8807,
"step": 598
},
{
"epoch": 1.95,
"learning_rate": 6.181105964461908e-06,
"loss": 0.8578,
"step": 600
},
{
"epoch": 1.95,
"learning_rate": 6.1176631074005365e-06,
"loss": 0.8694,
"step": 602
},
{
"epoch": 1.96,
"learning_rate": 6.054403717716959e-06,
"loss": 0.7783,
"step": 604
},
{
"epoch": 1.97,
"learning_rate": 5.991330784856626e-06,
"loss": 0.9314,
"step": 606
},
{
"epoch": 1.97,
"learning_rate": 5.9284472894536045e-06,
"loss": 0.8734,
"step": 608
},
{
"epoch": 1.98,
"learning_rate": 5.8657562031897255e-06,
"loss": 0.8809,
"step": 610
},
{
"epoch": 1.99,
"learning_rate": 5.80326048865415e-06,
"loss": 0.8615,
"step": 612
},
{
"epoch": 1.99,
"learning_rate": 5.740963099203375e-06,
"loss": 1.0366,
"step": 614
},
{
"epoch": 2.0,
"learning_rate": 5.678866978821655e-06,
"loss": 0.9718,
"step": 616
},
{
"epoch": 2.01,
"learning_rate": 5.6169750619818796e-06,
"loss": 0.6932,
"step": 618
},
{
"epoch": 2.01,
"learning_rate": 5.555290273506919e-06,
"loss": 0.5196,
"step": 620
},
{
"epoch": 2.02,
"learning_rate": 5.493815528431369e-06,
"loss": 0.6037,
"step": 622
},
{
"epoch": 2.03,
"learning_rate": 5.432553731863823e-06,
"loss": 0.5271,
"step": 624
},
{
"epoch": 2.03,
"learning_rate": 5.3715077788495895e-06,
"loss": 0.5792,
"step": 626
},
{
"epoch": 2.04,
"learning_rate": 5.31068055423386e-06,
"loss": 0.5071,
"step": 628
},
{
"epoch": 2.05,
"learning_rate": 5.250074932525392e-06,
"loss": 0.5189,
"step": 630
},
{
"epoch": 2.05,
"learning_rate": 5.189693777760665e-06,
"loss": 0.6438,
"step": 632
},
{
"epoch": 2.06,
"learning_rate": 5.129539943368542e-06,
"loss": 0.576,
"step": 634
},
{
"epoch": 2.06,
"learning_rate": 5.069616272035414e-06,
"loss": 0.511,
"step": 636
},
{
"epoch": 2.07,
"learning_rate": 5.0099255955708704e-06,
"loss": 0.6052,
"step": 638
},
{
"epoch": 2.08,
"learning_rate": 4.950470734773881e-06,
"loss": 0.4889,
"step": 640
},
{
"epoch": 2.08,
"learning_rate": 4.891254499299484e-06,
"loss": 0.5586,
"step": 642
},
{
"epoch": 2.09,
"learning_rate": 4.832279687526009e-06,
"loss": 0.5663,
"step": 644
},
{
"epoch": 2.1,
"learning_rate": 4.773549086422846e-06,
"loss": 0.5941,
"step": 646
},
{
"epoch": 2.1,
"learning_rate": 4.715065471418731e-06,
"loss": 0.7716,
"step": 648
},
{
"epoch": 2.11,
"learning_rate": 4.656831606270596e-06,
"loss": 0.5417,
"step": 650
},
{
"epoch": 2.12,
"learning_rate": 4.598850242932951e-06,
"loss": 0.564,
"step": 652
},
{
"epoch": 2.12,
"learning_rate": 4.541124121427851e-06,
"loss": 0.5382,
"step": 654
},
{
"epoch": 2.13,
"learning_rate": 4.483655969715396e-06,
"loss": 0.5737,
"step": 656
},
{
"epoch": 2.14,
"learning_rate": 4.426448503564823e-06,
"loss": 0.527,
"step": 658
},
{
"epoch": 2.14,
"learning_rate": 4.369504426426165e-06,
"loss": 0.6137,
"step": 660
},
{
"epoch": 2.15,
"learning_rate": 4.312826429302499e-06,
"loss": 0.7163,
"step": 662
},
{
"epoch": 2.16,
"learning_rate": 4.256417190622768e-06,
"loss": 0.4676,
"step": 664
},
{
"epoch": 2.16,
"learning_rate": 4.2002793761152126e-06,
"loss": 0.5177,
"step": 666
},
{
"epoch": 2.17,
"learning_rate": 4.144415638681407e-06,
"loss": 0.5405,
"step": 668
},
{
"epoch": 2.18,
"learning_rate": 4.088828618270868e-06,
"loss": 0.5134,
"step": 670
},
{
"epoch": 2.18,
"learning_rate": 4.033520941756316e-06,
"loss": 0.5176,
"step": 672
},
{
"epoch": 2.19,
"learning_rate": 3.978495222809533e-06,
"loss": 0.4865,
"step": 674
},
{
"epoch": 2.19,
"learning_rate": 3.923754061777846e-06,
"loss": 0.6041,
"step": 676
},
{
"epoch": 2.2,
"learning_rate": 3.869300045561244e-06,
"loss": 0.5154,
"step": 678
},
{
"epoch": 2.21,
"learning_rate": 3.815135747490138e-06,
"loss": 0.5103,
"step": 680
},
{
"epoch": 2.21,
"learning_rate": 3.761263727203733e-06,
"loss": 0.5287,
"step": 682
},
{
"epoch": 2.22,
"learning_rate": 3.707686530529088e-06,
"loss": 0.5508,
"step": 684
},
{
"epoch": 2.23,
"learning_rate": 3.65440668936079e-06,
"loss": 0.5387,
"step": 686
},
{
"epoch": 2.23,
"learning_rate": 3.6014267215413233e-06,
"loss": 0.5759,
"step": 688
},
{
"epoch": 2.24,
"learning_rate": 3.5487491307420675e-06,
"loss": 0.4253,
"step": 690
},
{
"epoch": 2.25,
"learning_rate": 3.4963764063449858e-06,
"loss": 0.5185,
"step": 692
},
{
"epoch": 2.25,
"learning_rate": 3.444311023325e-06,
"loss": 0.5245,
"step": 694
},
{
"epoch": 2.26,
"learning_rate": 3.3925554421330044e-06,
"loss": 0.527,
"step": 696
},
{
"epoch": 2.27,
"learning_rate": 3.3411121085796106e-06,
"loss": 0.4561,
"step": 698
},
{
"epoch": 2.27,
"learning_rate": 3.2899834537195594e-06,
"loss": 0.5389,
"step": 700
},
{
"epoch": 2.28,
"learning_rate": 3.2391718937368397e-06,
"loss": 0.5283,
"step": 702
},
{
"epoch": 2.29,
"learning_rate": 3.1886798298305065e-06,
"loss": 0.5664,
"step": 704
},
{
"epoch": 2.29,
"learning_rate": 3.1385096481012e-06,
"loss": 0.5734,
"step": 706
},
{
"epoch": 2.3,
"learning_rate": 3.088663719438404e-06,
"loss": 0.5811,
"step": 708
},
{
"epoch": 2.31,
"learning_rate": 3.039144399408387e-06,
"loss": 0.5713,
"step": 710
},
{
"epoch": 2.31,
"learning_rate": 2.989954028142882e-06,
"loss": 0.5543,
"step": 712
},
{
"epoch": 2.32,
"learning_rate": 2.9410949302285262e-06,
"loss": 0.5914,
"step": 714
},
{
"epoch": 2.32,
"learning_rate": 2.8925694145969798e-06,
"loss": 0.5784,
"step": 716
},
{
"epoch": 2.33,
"learning_rate": 2.844379774415822e-06,
"loss": 0.5684,
"step": 718
},
{
"epoch": 2.34,
"learning_rate": 2.796528286980197e-06,
"loss": 0.5311,
"step": 720
},
{
"epoch": 2.34,
"learning_rate": 2.749017213605172e-06,
"loss": 0.5826,
"step": 722
},
{
"epoch": 2.35,
"learning_rate": 2.7018487995188845e-06,
"loss": 0.5593,
"step": 724
},
{
"epoch": 2.36,
"learning_rate": 2.6550252737564475e-06,
"loss": 0.5497,
"step": 726
},
{
"epoch": 2.36,
"learning_rate": 2.6085488490546075e-06,
"loss": 0.5192,
"step": 728
},
{
"epoch": 2.37,
"learning_rate": 2.562421721747169e-06,
"loss": 0.5395,
"step": 730
},
{
"epoch": 2.38,
"learning_rate": 2.5166460716612106e-06,
"loss": 0.6872,
"step": 732
},
{
"epoch": 2.38,
"learning_rate": 2.4712240620140815e-06,
"loss": 0.5065,
"step": 734
},
{
"epoch": 2.39,
"learning_rate": 2.4261578393111463e-06,
"loss": 0.5618,
"step": 736
},
{
"epoch": 2.4,
"learning_rate": 2.381449533244379e-06,
"loss": 0.5709,
"step": 738
},
{
"epoch": 2.4,
"learning_rate": 2.3371012565916995e-06,
"loss": 0.5566,
"step": 740
},
{
"epoch": 2.41,
"learning_rate": 2.2931151051171486e-06,
"loss": 0.559,
"step": 742
},
{
"epoch": 2.42,
"learning_rate": 2.249493157471828e-06,
"loss": 0.4974,
"step": 744
},
{
"epoch": 2.42,
"learning_rate": 2.2062374750956815e-06,
"loss": 0.5811,
"step": 746
},
{
"epoch": 2.43,
"learning_rate": 2.1633501021200775e-06,
"loss": 0.5748,
"step": 748
},
{
"epoch": 2.44,
"learning_rate": 2.120833065271206e-06,
"loss": 0.4701,
"step": 750
},
{
"epoch": 2.44,
"learning_rate": 2.0786883737743014e-06,
"loss": 0.5965,
"step": 752
},
{
"epoch": 2.45,
"learning_rate": 2.0369180192587014e-06,
"loss": 0.5264,
"step": 754
},
{
"epoch": 2.45,
"learning_rate": 1.9955239756637156e-06,
"loss": 0.4516,
"step": 756
},
{
"epoch": 2.46,
"learning_rate": 1.9545081991453507e-06,
"loss": 0.6062,
"step": 758
},
{
"epoch": 2.47,
"learning_rate": 1.913872627983868e-06,
"loss": 0.4955,
"step": 760
},
{
"epoch": 2.47,
"learning_rate": 1.8736191824921868e-06,
"loss": 0.4711,
"step": 762
},
{
"epoch": 2.48,
"learning_rate": 1.8337497649251313e-06,
"loss": 0.5683,
"step": 764
},
{
"epoch": 2.49,
"learning_rate": 1.7942662593895399e-06,
"loss": 0.6317,
"step": 766
},
{
"epoch": 2.49,
"learning_rate": 1.7551705317552348e-06,
"loss": 0.4903,
"step": 768
},
{
"epoch": 2.5,
"learning_rate": 1.716464429566832e-06,
"loss": 0.4769,
"step": 770
},
{
"epoch": 2.51,
"learning_rate": 1.6781497819564407e-06,
"loss": 0.4629,
"step": 772
},
{
"epoch": 2.51,
"learning_rate": 1.6402283995572255e-06,
"loss": 0.5897,
"step": 774
},
{
"epoch": 2.52,
"learning_rate": 1.6027020744178367e-06,
"loss": 0.5047,
"step": 776
},
{
"epoch": 2.53,
"learning_rate": 1.5655725799177268e-06,
"loss": 0.5175,
"step": 778
},
{
"epoch": 2.53,
"learning_rate": 1.528841670683341e-06,
"loss": 0.5433,
"step": 780
},
{
"epoch": 2.54,
"learning_rate": 1.4925110825052125e-06,
"loss": 0.489,
"step": 782
},
{
"epoch": 2.55,
"learning_rate": 1.456582532255917e-06,
"loss": 0.5022,
"step": 784
},
{
"epoch": 2.55,
"learning_rate": 1.4210577178089447e-06,
"loss": 0.4123,
"step": 786
},
{
"epoch": 2.56,
"learning_rate": 1.3859383179584717e-06,
"loss": 0.5422,
"step": 788
},
{
"epoch": 2.56,
"learning_rate": 1.351225992340015e-06,
"loss": 0.6037,
"step": 790
},
{
"epoch": 2.57,
"learning_rate": 1.3169223813520072e-06,
"loss": 0.5819,
"step": 792
},
{
"epoch": 2.58,
"learning_rate": 1.2830291060782806e-06,
"loss": 0.4759,
"step": 794
},
{
"epoch": 2.58,
"learning_rate": 1.2495477682114509e-06,
"loss": 0.5633,
"step": 796
},
{
"epoch": 2.59,
"learning_rate": 1.216479949977234e-06,
"loss": 0.6078,
"step": 798
},
{
"epoch": 2.6,
"learning_rate": 1.1838272140596684e-06,
"loss": 0.5128,
"step": 800
},
{
"epoch": 2.6,
"learning_rate": 1.151591103527273e-06,
"loss": 0.5576,
"step": 802
},
{
"epoch": 2.61,
"learning_rate": 1.1197731417601233e-06,
"loss": 0.6094,
"step": 804
},
{
"epoch": 2.62,
"learning_rate": 1.088374832377861e-06,
"loss": 0.6498,
"step": 806
},
{
"epoch": 2.62,
"learning_rate": 1.057397659168643e-06,
"loss": 0.517,
"step": 808
},
{
"epoch": 2.63,
"learning_rate": 1.0268430860190136e-06,
"loss": 0.5233,
"step": 810
},
{
"epoch": 2.64,
"learning_rate": 9.96712556844729e-07,
"loss": 0.4652,
"step": 812
},
{
"epoch": 2.64,
"learning_rate": 9.67007495522524e-07,
"loss": 0.6261,
"step": 814
},
{
"epoch": 2.65,
"learning_rate": 9.377293058228243e-07,
"loss": 0.4634,
"step": 816
},
{
"epoch": 2.66,
"learning_rate": 9.088793713434063e-07,
"loss": 0.5321,
"step": 818
},
{
"epoch": 2.66,
"learning_rate": 8.804590554440096e-07,
"loss": 0.5212,
"step": 820
},
{
"epoch": 2.67,
"learning_rate": 8.524697011819216e-07,
"loss": 0.5385,
"step": 822
},
{
"epoch": 2.68,
"learning_rate": 8.249126312484934e-07,
"loss": 0.4638,
"step": 824
},
{
"epoch": 2.68,
"learning_rate": 7.977891479066347e-07,
"loss": 0.4914,
"step": 826
},
{
"epoch": 2.69,
"learning_rate": 7.711005329292887e-07,
"loss": 0.5311,
"step": 828
},
{
"epoch": 2.69,
"learning_rate": 7.44848047538842e-07,
"loss": 0.5352,
"step": 830
},
{
"epoch": 2.7,
"learning_rate": 7.190329323475321e-07,
"loss": 0.5172,
"step": 832
},
{
"epoch": 2.71,
"learning_rate": 6.936564072988128e-07,
"loss": 0.5763,
"step": 834
},
{
"epoch": 2.71,
"learning_rate": 6.687196716097188e-07,
"loss": 0.4306,
"step": 836
},
{
"epoch": 2.72,
"learning_rate": 6.442239037141696e-07,
"loss": 0.6044,
"step": 838
},
{
"epoch": 2.73,
"learning_rate": 6.201702612073024e-07,
"loss": 0.5409,
"step": 840
},
{
"epoch": 2.73,
"learning_rate": 5.965598807907602e-07,
"loss": 0.5242,
"step": 842
},
{
"epoch": 2.74,
"learning_rate": 5.73393878218973e-07,
"loss": 0.5805,
"step": 844
},
{
"epoch": 2.75,
"learning_rate": 5.506733482464277e-07,
"loss": 0.4793,
"step": 846
},
{
"epoch": 2.75,
"learning_rate": 5.283993645759456e-07,
"loss": 0.498,
"step": 848
},
{
"epoch": 2.76,
"learning_rate": 5.065729798079245e-07,
"loss": 0.5173,
"step": 850
},
{
"epoch": 2.77,
"learning_rate": 4.851952253906133e-07,
"loss": 0.5124,
"step": 852
},
{
"epoch": 2.77,
"learning_rate": 4.6426711157135815e-07,
"loss": 0.5415,
"step": 854
},
{
"epoch": 2.78,
"learning_rate": 4.437896273488673e-07,
"loss": 0.651,
"step": 856
},
{
"epoch": 2.79,
"learning_rate": 4.2376374042646827e-07,
"loss": 0.5693,
"step": 858
},
{
"epoch": 2.79,
"learning_rate": 4.0419039716638076e-07,
"loss": 0.5002,
"step": 860
},
{
"epoch": 2.8,
"learning_rate": 3.8507052254499575e-07,
"loss": 0.5297,
"step": 862
},
{
"epoch": 2.81,
"learning_rate": 3.6640502010915957e-07,
"loss": 0.5375,
"step": 864
},
{
"epoch": 2.81,
"learning_rate": 3.4819477193347885e-07,
"loss": 0.6165,
"step": 866
},
{
"epoch": 2.82,
"learning_rate": 3.304406385786374e-07,
"loss": 0.5443,
"step": 868
},
{
"epoch": 2.82,
"learning_rate": 3.131434590507221e-07,
"loss": 0.5885,
"step": 870
},
{
"epoch": 2.83,
"learning_rate": 2.9630405076158e-07,
"loss": 0.4871,
"step": 872
},
{
"epoch": 2.84,
"learning_rate": 2.7992320949018827e-07,
"loss": 0.5719,
"step": 874
},
{
"epoch": 2.84,
"learning_rate": 2.6400170934504976e-07,
"loss": 0.5361,
"step": 876
},
{
"epoch": 2.85,
"learning_rate": 2.485403027276079e-07,
"loss": 0.5775,
"step": 878
},
{
"epoch": 2.86,
"learning_rate": 2.3353972029668936e-07,
"loss": 0.4652,
"step": 880
},
{
"epoch": 2.86,
"learning_rate": 2.1900067093398425e-07,
"loss": 0.54,
"step": 882
},
{
"epoch": 2.87,
"learning_rate": 2.049238417105348e-07,
"loss": 0.491,
"step": 884
},
{
"epoch": 2.88,
"learning_rate": 1.913098978542749e-07,
"loss": 0.4532,
"step": 886
},
{
"epoch": 2.88,
"learning_rate": 1.781594827185884e-07,
"loss": 0.69,
"step": 888
},
{
"epoch": 2.89,
"learning_rate": 1.6547321775190806e-07,
"loss": 0.4711,
"step": 890
},
{
"epoch": 2.9,
"learning_rate": 1.5325170246835218e-07,
"loss": 0.5634,
"step": 892
},
{
"epoch": 2.9,
"learning_rate": 1.4149551441938303e-07,
"loss": 0.444,
"step": 894
},
{
"epoch": 2.91,
"learning_rate": 1.3020520916652513e-07,
"loss": 0.5638,
"step": 896
},
{
"epoch": 2.92,
"learning_rate": 1.1938132025510308e-07,
"loss": 0.5102,
"step": 898
},
{
"epoch": 2.92,
"learning_rate": 1.0902435918903053e-07,
"loss": 0.5644,
"step": 900
},
{
"epoch": 2.93,
"learning_rate": 9.913481540663849e-08,
"loss": 0.5643,
"step": 902
},
{
"epoch": 2.94,
"learning_rate": 8.971315625754595e-08,
"loss": 0.683,
"step": 904
},
{
"epoch": 2.94,
"learning_rate": 8.075982698057205e-08,
"loss": 0.4953,
"step": 906
},
{
"epoch": 2.95,
"learning_rate": 7.227525068269958e-08,
"loss": 0.5235,
"step": 908
},
{
"epoch": 2.95,
"learning_rate": 6.425982831907651e-08,
"loss": 0.5531,
"step": 910
},
{
"epoch": 2.96,
"learning_rate": 5.6713938674071156e-08,
"loss": 0.5424,
"step": 912
},
{
"epoch": 2.97,
"learning_rate": 4.9637938343367607e-08,
"loss": 0.5042,
"step": 914
},
{
"epoch": 2.97,
"learning_rate": 4.3032161717119216e-08,
"loss": 0.5134,
"step": 916
},
{
"epoch": 2.98,
"learning_rate": 3.6896920964143435e-08,
"loss": 0.5131,
"step": 918
},
{
"epoch": 2.99,
"learning_rate": 3.1232506017169206e-08,
"loss": 0.5912,
"step": 920
},
{
"epoch": 2.99,
"learning_rate": 2.6039184559139007e-08,
"loss": 0.5772,
"step": 922
},
{
"epoch": 3.0,
"learning_rate": 2.1317202010553428e-08,
"loss": 0.4642,
"step": 924
},
{
"epoch": 3.0,
"step": 924,
"total_flos": 412039810383872.0,
"train_loss": 0.9543006863944974,
"train_runtime": 21744.4539,
"train_samples_per_second": 1.358,
"train_steps_per_second": 0.042
}
],
"max_steps": 924,
"num_train_epochs": 3,
"total_flos": 412039810383872.0,
"trial_name": null,
"trial_params": null
}