Tensorflow-操作类型未注册'CudnnRNN' [英] Tensorflow - Op Type not registered 'CudnnRNN'

问题描述

我是tensorflow的新手，正在尝试对其进行设置.

I am new to tensorflow and trying to set it up.

当我尝试使用CuDNNGRU训练模型时，似乎可以正确加载，然后出现错误:

When I try to train a model using CuDNNGRU it seems to load correctly and then gives an error :

tensorflow.python.framework.errors_impl.NotFoundError:操作类型不正确 注册的"CudnnRNN"

tensorflow.python.framework.errors_impl.NotFoundError: Op type not registered 'CudnnRNN'

我确实在tensorflow/contrib中看到了Cudnn_rnn目录，这是值得的.

I do see a Cudnn_rnn directory in tensorflow/contrib for what that is worth.

我有python 3.6和VS2013.

I have python 3.6 and VS2013.

我尝试了以下操作，但仍然出现错误:

I have tried the following, but still getting an error:

• 两个Cuda 8/9
• 卸载/重新安装tensorflow/Theano/Keras/TensorFlow

老实说，设置看起来是如此复杂和敏感，我可能已经搞砸了.

Honestly the setup seems so convoluted and touchy, I may have screwed something up.

我想念一个待办事项吗?有一些手动修复的方法吗?谢谢！

Am I missing a to-do? Some way to manually fix? Thanks!

我要复制的示例代码:

def get_model(embedding, sequence_length, dropout_rate, recurrent, dense_size):
input_layer = Input(shape=(sequence_length,))
embedding_layer = Embedding(embedding.shape[0], embedding.shape[1],
                            weights=[embedding], trainable=False)(input_layer)
x = Bidirectional(CuDNNGRU(recurrent, return_sequences=True))(embedding_layer)
x = Dropout(dropout_rate)(x)
x = Bidirectional(CuDNNGRU(recurrent, return_sequences=False))(x)
x = Dense(dense_size, activation="relu")(x)

推荐答案

我通过以下操作解决了此问题:

I fixed this by doing :

pip install tensorflow --ignore-installed --upgrade

pip install tensorflow --ignore-installed --upgrade

然后

从tensorflow.python.client导入device_lib打印(device_lib.list_local_devices())

from tensorflow.python.client import device_lib print(device_lib.list_local_devices())

这篇关于Tensorflow-操作类型未注册'CudnnRNN'的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！