Tensorflow所遇坑

 

TensorFlow问题：

1、FLAGS._parse_flags()报错AttributeError:_parse_flags

解决：

因为TensorFlow的版本问题了，TensorFlow版本升级后，它就无情的抛弃了FLAGS._parse_flags()这种用法，改成了用FLAGS.flag_values_dict()

2、raise ValueError("Shapes %s and %s are incompatible" % (self, other))
      ValueError: Shapes (3, ?, 1, 1, 128) and () are incompatible

解决：

concated = tf.concat(1, [indices, sparse_labels])改为：

concated= tf.concat([indices, sparse_labels], 1)

3、WARNING:softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.

Instructions for updating:

Future major versions of TensorFlow will allow gradients to flow
into the labels input on backprop by default.

See tf.nn.softmax_cross_entropy_with_logits_v2.

解决：

将tf.nn.softmax_cross_entropy_with_logits改为：
tf.nn.softmax_cross_entropy_with_logits_v2

4.ValueError: Only call `softmax_cross_entropy_with_logits` with named arguments (labels=..., logits=..., ...)

解决：

将cross_entropy2=tf.reduce_sum(tf.nn.softmax_cross_entropy_with_logits_v2(logits, y_))改为：

cross_entropy2=tf.reduce_sum(tf.nn.softmax_cross_entropy_with_logits_v2(labels=logits, logits=y_))