search

gankim / tree-of-clarifications

🌲 Code for our EMNLP 2023 paper - 🎄 "Tree of Clarifications: Answering Ambiguous Questions with Retrieval-Augmented Large Language Models"
https://arxiv.org/abs/2310.14696
44 stars 2 forks source link

Questions about Bing results & ColBERT sever/index. #1

Open XuLingnan opened 7 months ago

XuLingnan commented 7 months ago

Hi.

I couldn't find './bing/results.json' in your repository. Could you please share the pre-compiled bing search results?

Moreover, the ColBERT sever url 'http://ec2-44-228-128-229.us-west-2.compute.amazonaws.com:8893/api/search' seems to have changed. The paper does't seem to mention what the retrieval datastore (ColBERT index) is, so I couldn't set up a new sever. I wonder if you could share a link to the index. It's better if you could also share a new ColBERT sever url.

The paper mentions "Finally, we obtain over 200 passages by combining passages retrieved by each system.", while the default 'top_k_docs' in 'run_toc.py' is 100. Btw, the default 'top_k' in 'bing_search.py' is 50 and that in 'get_wiki.py' is 100. Does that mean 100 docs are retrieved by Bing, from which the top-50 docs along with other 50 docs retrieved by ColBERT consist the final 100(rather than 200)-doc datastore?

Thank you very much for your time and attention to this matter.

sylvain-wei commented 7 months ago

I also need the ./bing/results.json.

I've found that wikipedia.org is not accessible in China Mainland. When I use my Bing Search API on my remote server with no access to external internet, the Bing Search API only return an empty dict. Besides, the retrieved documents are all the latest if I don't add some restrictions in the original query.

Therefore, I want the ./bing/results.json very much...

HU-xiaobai commented 2 weeks ago

hello, could I ask have you solved the problem?