Keras保存中间变量

Question

我使用张量流的Keras。而且我希望在训练期间看到并将Kvariable保存为numpy值。但“K.eval”效果不佳。有没有解决方案？

def acc(y_true, y_pred): 
    temp = K.eval(y_true) 
    ... 
    ... 

Answer 1

如果是培训期间的准确性，那么Keras会默认存储此信息;如果你适合你的机型，如：

hist = model.fit(x_train, y_train, 
      batch_size=batch_size, 
      epochs=20, 
      verbose=1, 
      validation_data=(x_test, y_test)) 

然后训练和验证准确性&亏损为每个时间段都存储在字典hist.history：

hist.history.keys() 
# ['acc', 'loss', 'val_acc', 'val_loss'] 
hist.history['val_acc'] # validation accuracy 
# [0.98140000000000005, 
# 0.98340000000000005, 
# 0.98740000000000006, 
# 0.98680000000000001, 
# 0.98750000000000004, 
# 0.98740000000000006, 
# 0.9869, 
# 0.98939999999999995, 
# 0.98819999999999997, 
# 0.98819999999999997, 
# 0.99039999999999995, 
# 0.9889, 
# 0.98919999999999997, 
# 0.98860000000000003, 
# 0.98780000000000001, 
# 0.98799999999999999, 
# 0.98819999999999997, 
# 0.9889, 
# 0.98909999999999998, 
# 0.98609999999999998]