Jesse Zhang
Jesse Zhang
I think this is a great idea! Will let @jerryjliu address the best way to PR, but specifically regarding the metadata, you can include it using `extra_info` when creating a...
Ah yeah more fine-grained customization there would be really useful for sure. Looking forward to your PR :)
> I need to do a bit more thinking on how much to expose the PromptHelper (do we want to expose it as a customizable class at all if we...
Check out `GPTTreeIndex` for the new pattern of how a default text splitter is set. If it looks good, I'll extend to all indices.
Done! Thanks for the review of that! We can revamp PromptHelper in the future.
This looks awesome! Thanks for all the improvements. The 600 req/hr limit should be fine since we're caching everything. The website, for example, needs an API key, which I believe...
Included one more enhancement: - New web reader using BeautifulSoup - Allows for more fine-grained scraping of specific sites. Included an extractor for Substack posts
> Thanks for the changes! Some comments below. Also high-level note, there's really like 2-3 subchanges in here 🙂 - next time would be good to split it up a...
> nice!! is this going to get added to `DEFAULT_FILE_EXTRACTOR` in file/base.py as well? Yep good point, forgot about that at the end -- added
> Please add some examples of evals to the respective field in the PR description. Just edited the original post!