如何解决如何在Tensorflow中保存和加载经过训练的生成器模型?
我现在正在使用tensorflow1.15.0训练GANs模型。我的模型有两个生成器和两个鉴别器。我想在训练后保存发电机的重量和模型骨架,该怎么做?我的模型是这样编译的:
optimizer = tf.train.AdamOptimizer(self.learning_rate,beta1=0.5)
self.model_vars = tf.trainable_variables()
d_A_vars = [var for var in self.model_vars if 'd_A' in var.name]
d_B_vars = [var for var in self.model_vars if 'd_B' in var.name]
g_A_vars = [var for var in self.model_vars if 'g_A' in var.name]
g_B_vars = [var for var in self.model_vars if 'g_B' in var.name]
self.d_A_trainer = optimizer.minimize(d_loss_A,var_list=d_A_vars)
self.d_B_trainer = optimizer.minimize(d_loss_B,var_list=d_B_vars)
self.g_A_trainer = optimizer.minimize(g_loss_A,var_list=g_A_vars)
self.g_B_trainer = optimizer.minimize(g_loss_B,var_list=g_B_vars)
g_A_trainer的训练过程如下:
# Optimizing the G_A network
_,fake_B_temp,fake_B_label_temp,g_A_loss,summary_str = sess.run(
[self.g_A_trainer,self.fake_images_b,self.fake_labels_b,self.g_A_loss,self.g_A_loss_summ],feed_dict={
self.input_a:inputs['images_i'],self.input_b:inputs['images_j'],self.label_a:inputs['labels_i'],self.label_b:inputs['labels_j'],self.learning_rate: curr_lr
}
)
writer.add_summary(summary_str,epoch * max_images // 64 + i)
fake_B_temp1 = self.fake_image_pool(self.num_fake_inputs,self.fake_images_B)
fake_B_label_temp1 = self.fake_label_pool(self.num_fake_inputs,self.fake_labels_B)
现在,如果我想保存和加载经过训练的g_A模型,该怎么做?有人可以帮我吗?
版权声明:本文内容由互联网用户自发贡献,该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容, 请发送邮件至 dio@foxmail.com 举报,一经查实,本站将立刻删除。