/Tensorflow_Embedding_and_Weight_Visualization

Helper functions for visualizing the embedding and weights as images.

Primary LanguagePythonMIT LicenseMIT

Tensorflow_Embedding_and_Weight_Visualization

Helper functions for visualizing the embedding and weights as images. Steps:

  1. Convert the embedding or weights into an image, which is comprised of smaller images (feature maps or weight matrices as images).
  2. Use tensorflow.summary.image() to log the converted images.
  3. Add related codes just like what are required for visualizing images in TensorFlow for TensorBoard.
  4. The result can be viewed via TensorBoard.