Hi, i was wondering if there is a way to visualize the word embeddings generated by EmbeddingIntentClassifier component, in Tensorboard. I tried to run this command inside the model dir:
$ tensorboard --logdir ./
But all i see is this:
I don’t understand why it shows numbers instead of the words itself
The current NN architecture in embedding intent classifier doesn’t work at a word embedding level. Each sentence is represented as a Bag of Words representation and that BOW vector is embedded in the network. So you can’t get word level embeddings in this architecture.
@LuizZucchi Hi! I know this post is old and maybe things have changed but I still wanted to clear a doubt. I was looking for something similar on how to view the embedding. I tried tensorboard --logdir ./ where it opens the tensorboard web page with the graphs for loss, distribution and scalars. Could you let me know how did you view this representation?
