[Most-ai-contest] A Short Guide for Adding New Features/Embeddings to BERT's Attention Layers

[張光瑜]

Dear all, 

To help you implement models that adds new features/embeddings to BERT's attention layers, we have create a short guide (guide_bert_add_feature_to_attention.pptx) for it. 
You can also refer to the code of the single-span-multi-hop module (run_ssmh.py), which uses many kinds of features. 
You may ask me if you have any problem. 

Regards, 
張光瑜 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.iis.sinica.edu.tw/pipermail/most-ai-contest/attachments/20200401/a1f9c14f/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: run_ssmh.py
Type: text/x-python
Size: 44839 bytes
Desc: not available
URL: <http://www.iis.sinica.edu.tw/pipermail/most-ai-contest/attachments/20200401/a1f9c14f/attachment-0001.py>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: guide_bert_add_feature_to_attention.pptx
Type: application/vnd.openxmlformats-officedocument.presentationml.presentation
Size: 48817 bytes
Desc: not available
URL: <http://www.iis.sinica.edu.tw/pipermail/most-ai-contest/attachments/20200401/a1f9c14f/attachment-0001.bin>