keras中的fit_generator和fit函数均返回History对象,那么History怎么用呢?事实上History对象已经记录了运行输出。在了解之前,我们甚至自己定义回调函数记录损失和准确率等。
相关keras源码位于网址:
代码语言:javascript复制class History(Callback):
"""Callback that records events into a `History` object.
This callback is automatically applied to
every Keras model. The `History` object
gets returned by the `fit` method of models.
"""
def on_train_begin(self, logs=None):
self.epoch = []
self.history = {}
def on_epoch_end(self, epoch, logs=None):
logs = logs or {}
self.epoch.append(epoch)
for k, v in logs.items():
self.history.setdefault(k, []).append(v)
可以看出History类对象包含两个属性,分别为epoch和history,epoch为训练轮数。
根据compile参数metrics,history包含不同的内容。比如,当某一次metrics=[‘accuracy’]时,运行如下部分代码我们可以看出,history字典类型,包含val_loss,val_acc,loss,acc四个key值。
代码语言:javascript复制####省略若干
history = model.fit_generator(
mp.train_flow,
steps_per_epoch=32,
epochs=3,
validation_data=mp.test_flow,
validation_steps=32)
print(history.history)
print(history.epoch)
print(history.history['val_loss'])
{‘val_loss’: [0.4231100323200226, 0.3713115310668945, 0.3836631367206573], ‘val_acc’: [0.815, 0.84, 0.83], ‘loss’: [0.8348453622311354, 0.5010451343324449, 0.4296100065112114], ‘acc’: [0.630859375, 0.7509920634920635, 0.783203125]} [0, 1, 2] [0.4231100323200226, 0.3713115310668945, 0.3836631367206573]
补充知识:在ipython中使用%history快速查找历史命令
1、输出所有历史记录,且带有序号
代码语言:javascript复制 %history -n
150: %cpaste
151: %cpaste
152: print(r" ")
153: print(r' ')
154: print(r' <')
155: print(r' ')
156: print(r' ')
157: print(r' ')
...
2、按序号,查找某些序号区间的历史纪录
代码语言:javascript复制 %history -n 168-170 178 185-190
168: planets
169:
for method, group in planets.groupby('method'):
print(f'{method:30s} method={group}')
170:
for method, group in planets.groupby('method'):
print(f'{method:30s} method={group.shape}')
178: %history?
185: %history -u
186: %history -n -u
187: ?%history
188: %history -g method
189: %history -g method print
190: %history -g for method,
3、模糊查找
代码语言:javascript复制 %history -g print*metho*
120:
for method, group in planets.groupby('method'):
print(f"{method:30s} shape={groupe.shape}")
121:
for method, group in planets.groupby('method'):
print(f"{method:30s} shape={group.shape}")
169:
for method, group in planets.groupby('method'):
print(f'{method:30s} method={group}')
170:
for method, group in planets.groupby('method'):
print(f'{method:30s} method={group.shape}')
182:
for method, group in planets.groupby('method'):
print(f"{method:30s shape=group.shape}")
198: %history -g print*metho*
以上这篇keras中的History对象用法就是小编分享给大家的全部内容了,希望能给大家一个参考。