level 1
mao_jiayu
楼主
使用keras.layers.GRU，代码如下：
from keras.models import Sequential
from keras.layers import GRU, Dense
model = Sequential()
model.add(GRU(32,
dropout=0.2,
recurrent_dropout=0.2,
input_shape=(None, float_data.shape[-1])))
model.add(Dense(1))
model.compile(loss='mae', optimizer='rmsprop')
history = model.fit(train_gen,
epochs=40,
steps_per_epoch=500,
validation_data=val_gen,
validation_steps=val_steps)
出现一个warning：
WARNING:tensorflow:Layer gru_2 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU
如果去掉recurrent_dropout=0.2这一行，训练一轮只需要12s，加上后训练一轮需要6分钟，好像就完全实在cpu上跑，请问下有人遇到过这个问题吗？求解
（keras==2.4.2， tensorflow-gpu==2.2.0）
2020年07月08日 13点07分
1
from keras.models import Sequential
from keras.layers import GRU, Dense
model = Sequential()
model.add(GRU(32,
dropout=0.2,
recurrent_dropout=0.2,
input_shape=(None, float_data.shape[-1])))
model.add(Dense(1))
model.compile(loss='mae', optimizer='rmsprop')
history = model.fit(train_gen,
epochs=40,
steps_per_epoch=500,
validation_data=val_gen,
validation_steps=val_steps)
出现一个warning：
WARNING:tensorflow:Layer gru_2 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU
如果去掉recurrent_dropout=0.2这一行，训练一轮只需要12s，加上后训练一轮需要6分钟，好像就完全实在cpu上跑，请问下有人遇到过这个问题吗？求解
（keras==2.4.2， tensorflow-gpu==2.2.0）
