【发布时间】:2020-10-26 16:27:52
【问题描述】:
我正在研究 CNN,一旦第一个 epoch 完成,我就会收到错误消息:
"Function call stack: distributed_function"
和
"Fused conv implementation does not support grouped convolutions for now."
我正在使用稍微修改过的代码,该代码用于另一个 CNN,它在前一个上工作,所以我有点不知道为什么现在会发生这个错误。
我使用的图像是类似于this的灰度热图图像
代码:
TRAINING_DIR = '/Users/me/School/Research/mini'
training_datagen = ImageDataGenerator(
rescale = 1./255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
train_generator = training_datagen.flow_from_directory(
TRAINING_DIR,
target_size=(640,480),
class_mode='categorical'
)
model = tf.keras.models.Sequential([
# Input shape is the desired size of the image 640x480 with 1 byte color
# This is the first convolution
tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(640, 480, 1)),
tf.keras.layers.MaxPooling2D(2, 2), # factors to downscale by, (2,2) will halve
tf.keras.layers.Conv2D(64, (3,3), activation='relu'), # 2nd convo layer
tf.keras.layers.MaxPooling2D(2,2),
tf.keras.layers.Conv2D(128, (3,3), activation='relu'), # 3rd convo layer
tf.keras.layers.MaxPooling2D(2,2),
tf.keras.layers.Conv2D(128, (3,3), activation='relu'), # 4th convo layer
tf.keras.layers.MaxPooling2D(2,2),
tf.keras.layers.Flatten(), # Flatten to DNN
tf.keras.layers.Dropout(0.5),
tf.keras.layers.Dense(512, activation='relu'), # hidden layer
tf.keras.layers.Dense(3, activation='softmax') # 3 class
])
model.summary()
model.compile(loss = 'categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])
history = model.fit(train_generator, epochs=15, verbose = 1)
model.save("rps.h5")
acc = history.history['accuracy']
loss = history.history['loss']
epochs = range(len(acc))
【问题讨论】:
-
我猜问题是使用
model.fit如果你没有使用当前的tensorflow版本,那么你应该使用model.fit_generator -
无论哪种方式,我都会遇到相同的错误。我的TF版本是2.1.0,Keras版本是2.2.4-tf
标签: python python-3.x tensorflow keras conv-neural-network