Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Different Vocab Size Between Tokenizer and Model's Word Embedding Layer #33

Open
louisowen6 opened this issue Jul 23, 2021 · 0 comments

Comments

@louisowen6
Copy link

Expected Behavior

The length of tokenizer vocab size and the BERT's word embedding layer dimension should be the same

Actual Behavior

The length of tokenizer vocab size and the BERT's word embedding layer dimension is not the same

Steps to Reproduce the Problem

  1. Load the model: model = AutoModel.from_pretrained('indobenchmark/indobert-base-p1')
  2. Print the model: print(model)

image

  1. Load the tokenizer: tokenizer = AutoTokenizer.from_pretrained('indobenchmark/indobert-base-p1')
  2. Print the length of toikenizer: print(len(tokenizer))
    image
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant