我正在尝试为GAN训练“生成器”。当我想要喂占位符时,出现以下错误:
_grads = mySess.run(params["gradient"],
feed_dict={params["data"]["x"]: batch_Z})
params["gradient"]
是:
params["gradient"] = optimizer.compute_gradients(loss, trainable_vars)
params [“ data”] [“ x”] = tf.placeholder(tf.float32,[None,z_shape])
损失和trainable_vars为:
loss = cost(tf.ones_like(disc_logits_fake), disc_logits_fake)
trainable_vars = [generator.W_conv1, generator.b_conv1, generator.W_conv2, generator.b_conv2, generator.W_conv3, generator.b_conv3, generator.W_conv4, generator.b_conv4]
我得到的错误是:
TypeError:获取参数None的类型无效