Question

我正在尝试关注word2vec示例，但我收到此错误：

TypeError: Input 'b' of 'MatMul' Op has type float32 that does not match type int32 of argument 'a'.

在这一行

相似度= tf.matmul（ tf.cast（valid_embeddings，tf.int32），tf.cast（normalized_embeddings，tf.int32），transpose_b = True）

这是整个代码：

graph = tf.Graph()

with graph.as_default():
  # Input data.
  train_inputs = tf.placeholder(tf.int32, shape=[batch_size])
  train_labels = tf.placeholder(tf.int32, shape=[batch_size, 1])
  valid_dataset = tf.constant(valid_examples, dtype=tf.int32)
  # Ops and variables pinned to the CPU because of missing GPU implementation
  with tf.device('/cpu:0'):
    # Look up embeddings for inputs.
    embeddings = tf.Variable(
        tf.random_uniform([vocabulary_size, embedding_size], -1.0, 1.0))
    embed = tf.nn.embedding_lookup(embeddings, train_inputs)
    # Construct the variables for the NCE loss
    nce_weights = tf.Variable(
        tf.truncated_normal([vocabulary_size, embedding_size],
                            stddev=1.0 / math.sqrt(embedding_size)))
    nce_biases = tf.Variable(tf.zeros([vocabulary_size]))
  # Compute the average NCE loss for the batch.
  # tf.nce_loss automatically draws a new sample of the negative labels each
  # time we evaluate the loss.
  loss = tf.reduce_mean(
      tf.nn.nce_loss(nce_weights, nce_biases, embed, train_labels,
                     num_sampled, vocabulary_size))
  # Construct the SGD optimizer using a learning rate of 1.0.
  optimizer = tf.train.GradientDescentOptimizer(1.0).minimize(loss)
  # Compute the cosine similarity between minibatch examples and all embeddings.
  norm = tf.sqrt(tf.reduce_sum(tf.square(embeddings), 1, keep_dims=True))
  normalized_embeddings = embeddings / norm
  valid_embeddings = tf.nn.embedding_lookup(
      normalized_embeddings, valid_dataset)
  similarity = tf.matmul(
      tf.cast(valid_embeddings,tf.int32), tf.cast(normalized_embeddings,tf.int32), transpose_b=True)
  # Add variable initializer.
  init = tf.initialize_all_variables()

我该如何解决这个问题？

Answer 1

我在使用Tensorflow r1.4和Python 3.4时遇到了同样的问题。

确实，我认为您需要更改代码

Properties props = ...; // load your file
List<String> keysToRetrieve = Arrays.asList("sourceAttrStart", "user", ..., "sourceAttrEnd");

Map<String, String> entriesToRetrieve = new HashMap<>();
for (String key : keysToRetrieve){
   entriesToRetrieve.put(key, props.getProperty(key)); 
}

到

tf.nn.nce_loss(nce_weights, nce_biases, embed, train_labels,
                 num_sampled, vocabulary_size))

或

tf.nn.nce_loss(nce_weights, nce_biases, train_labels, embed,
                 num_sampled, vocabulary_size))

同时，您需要将代码更改回

loss = tf.reduce_mean(tf.nn.nce_loss(
        weights = softmax_weights,
        biases = softmax_biases, 
        inputs = embed, 
        labels = train_labels, 
        num_sampled = num_sampled, 
        num_classes = vocabulary_size))

使用similarity = tf.matmul(valid_embeddings, tf.transpose(normalized_embeddings))是错误的，实际上，没有必要使用tf.cast(..., tf.int32)，因为它已经是tf.float32。

P.S。

当您在使用tf.cast(..., tf.float32)时遇到问题时，此解决方案也很有用，因为tf.nn.sampled_softmax_loss()的使用与sampled_softmax_loss()非常相似。

Answer 2

为什么要在整数空间中进行矩阵乘法？您可能希望这两个tf.cast都是tf.float32。

TypeError：'MatMul'Op的'b'输入类型为float32，与参数'a'的int32类型不匹配

2 个答案: