The latest updates on your projects. Learn more about Vercel for Git ↗︎

• [ENH] Removing new line cleanup from OpenAI EF (python) #2125 👈
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @tazarov and the rest of your teammates on Graphite

Reviewer Checklist

Please leverage this checklist to ensure your code review is thorough before approving

Testing, Bugs, Errors, Logs, Documentation

• Can you think of any use case in which the code does not behave as intended? Have they been tested?
• Can you think of any inputs or external events that could break the code? Is user input validated and safe? Have they been tested?
• If appropriate, are there adequate property based tests?
• If appropriate, are there adequate unit tests?
• Should any logging, debugging, tracing information be added or removed?
• Are error messages user-friendly?
• Have all documentation changes needed been made?
• Have all non-obvious changes been commented?

System Compatibility

• Are there any potential impacts on other parts of the system or backward compatibility?
• Does this change intersect with any items on our roadmap, and if so, is there a plan for fitting them together?

Quality

• Is this code of a unexpectedly high quality (Readability, Modularity, Intuitiveness)

I agree that this is correct. However, this may impact people who already have collections using the old EF.

They would have to migrate, and we should let them know when we release this. Assigning myself here so that I remember to do this when this releases.

I had similar realization that users' new queries will have slightly worse outcome for exact matches, but perhaps that would be marginal. I'll do a small test to prove myself wrong

Another minutiae from my investigation was that non-code first gen models (-001) need new line removal, but it feels that while available from API perspective are being silently phased out by OpenAI (couldn't find refs on the latest embeddings OAI docs).

I had similar realization that users' new queries will have slightly worse outcome for exact matches, but perhaps that would be marginal. I'll do a small test to prove myself wrong

Whether or not this makes retrieval accuracy worse, users may be relying on the embeddings themselves, e.g. for deduplication. We must notify them after this change.

I had similar realization that users' new queries will have slightly worse outcome for exact matches, but perhaps that would be marginal. I'll do a small test to prove myself wrong

Whether or not this makes retrieval accuracy worse, users may be relying on the embeddings themselves, e.g. for deduplication. We must notify them after this change.

Will it make sense to add a flag (constructor param) so that whoever needs/uses new line removals can do so going forward? But perhaps making an opt-in thing by default + the warning for the change.

I think we should just pull the trigger and not complicate the interface with cruft, while ensuring we communicate to users.