3 ms·
Nice, I was playing with documentation q&a with gpt in https://github.com/rmuller-ml/gpt_table_of_contents_doc_qa https://github.com/rmuller-ml/gpt_table_of_con
by rmuller-ml 4y ago
Nice, I was playing with documentation q&a with gpt in https://github.com/rmuller-ml/gpt_table_of_contents_doc_qa https://github.com/rmuller-ml/gpt_table_of_contents_doc_qa and found that if the documents your are doing semantic search on is structured, you can prompt gpt with the table of contents so gpt does the chunk prefiltering for you. I guess this is similar to your tagging explanation. Nice work
- jasonwcfan 4y agoThat's clever! So it looks like rather than use a vector store, in your approach you use GPT and the table of contents to find the right section? How do you find it performs in terms of accuracy and cost?