Skip to content
This repository has been archived by the owner on May 25, 2020. It is now read-only.

parameter mismatch error #26

Closed
maxymy98 opened this issue Aug 6, 2018 · 4 comments
Closed

parameter mismatch error #26

maxymy98 opened this issue Aug 6, 2018 · 4 comments

Comments

@maxymy98
Copy link

maxymy98 commented Aug 6, 2018

default

When I was trainning the model, I had this parameter mismatch error. I use Windows and Anaconda with Python 2.7. The trainning corpus is the dummy corpus provided. I did not use Docker since Docer-gpu is not supported on Windows. Thanks a lot!

@nikitos9000
Copy link
Contributor

Hi @maxymy98

That's because you're trying to fine-tune the pre-trained model (which was trained with 50000 tokens dictionary), but your new dataset (dummy one) has much fewer distinct tokens in it (39).

I think you can just re-use index_to_token.json file from the pre-trained model if you want to fine-tune it with a new corpus.

@nicolas-ivanov
Copy link
Contributor

nicolas-ivanov commented Aug 7, 2018

@maxymy98 Seems like you have a wrong tokens index file. Check data/tokens_index/t_idx_processed_dialogs.json – it should have 50k different token values, not 39.
The original index file might have been replaced in case you run tools/prepare_index_files.py script on a dummy corpus.

How to fix:

  1. remove data/tokens_index/t_idx_processed_dialogs.json
  2. run tools/download_model.py script, it should get you the original index file

@maxymy98
Copy link
Author

maxymy98 commented Aug 8, 2018

@nicolas-ivanov @nsmetanin thank you so much for the quick reply. The issue is resolved!

@nicolas-ivanov
Copy link
Contributor

Awesome!

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants