Adding WordWeight layer in pretrained model

Hi !

I'm currently using the pretrained 'bert-base-nli-stsb-mean-tokens' model.
I was wondering if it was possible to add a WordWeight layer in the model and still work with the trained BERT parameters ?

Thanks !

The request concerns the pretrained bert-base-nli-stsb-mean-tokens model and adding a WordWeight layer while retaining its trained BERT parameters. No file, test, entry point, or acceptance criteria are named; first determine where the model is defined and how pretrained parameters are loaded. Done would require an agreed implementation or a clear answer about whether this configuration is supported.

Written by the indexing model from the issue text.