-
Notifications
You must be signed in to change notification settings - Fork 63
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Need some multi-layered summaries maybe #31
Comments
yea, I don't think we have this figured out yet. I think the summaries are too long, I'm not sure if they are the best approach but it works for small projects.
langchain use some kind of tokenizing and a special DB to try to match up such things I think, but IDK all the details, not sure exactly how we should use it. |
"chosen arbitrarily" - ouch 😅 but ya, true. Possible solutions (can be used together):
Vector stores would help in reducing the problem. The process would be something like this:
vector stores: Sounds like something we should totally do but I see it more as an optimization today (literally today) |
Consider using compression: https://twitter.com/mckaywrigley/status/1643592353817694218 |
@fjrdomingues btw when I say "chosen arbitrarily" I mean basically it's not a formula, but a statically chosen number and I assumed the number wasn't tested
This doesn't take into account the model and doesn't dictate anything for the AI.
something like And I'd still call So that would be a calculated number Also, semi-related, loading summaries should probably also have include&exclude lists, especially if we do something like testing vs coding agents. But generally today if someone indexes a lot of files, include&exclude lists won't ignore the old files generated, so they have to delete and recreate basically, not a good user experience and hard to explain in support |
@fjrdomingues something like
even if you want multi-layer summaries in the end, you might use the summary chunks to do it. |
I spinned off the chunks to #73 |
@Charuru if you want to try the |
While we might do some kind of multi-layer summary in the future, |
Getting this: Aborting. Summary files combined are too big for the context window of gpt3.5 as I don't have GPT4 with 8k yet.
The text was updated successfully, but these errors were encountered: