search

MingjunZhong / seq2point-nilm

Sequence-to-point learning for non-intrusive load monitoring
95 stars 32 forks source link

getting poor MAE and MSE results when running code #13

Open wuyulinmoveon opened 1 month ago

wuyulinmoveon commented 1 month ago

Hi,Dr. @MingjunZhong Thanks for sharing the code. This time I am trying to reproduce result of REDD based seq2point. But the results I get for REDD are quite different from your reported results. Could you give me some advice? dishwasher PATH NAME: saved_models/dishwasher_seq2point_model.h5 Loaded model with 30708249 weights 197/197 - 8s 197/197 [==============================] - 8s 40ms/step - loss: 0.0255 - mean_squared_error: 0.0255 - mean_squared_logarithmic_error: 0.0028 - mean_absolute_error: 0.0362

1234568sby commented 1 month ago

@wuyulinmoveon Hello, excuse me, I'd like to ask, have you run this?

wuyulinmoveon commented 1 month ago

@wuyulinmoveon Hello, excuse me, I'd like to ask, have you run this?

you mean run this code？yes，I can run it，but the metric has some problem and I can't solve this.

1234568sby commented 1 month ago

@wuyulinmoveon Can We Talk About It?

wuyulinmoveon commented 1 month ago

@1234568sby yes，if you can't run code,you can refer other issues, they answered.