Reduce size of LLM prompts + cache per-schema context #1384
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* truncate text/binary sample data fields to 1024 characters (or smaller if judged to be needed) * truncate entire tables from schema representation if the representation is very large * for latency improvement, cache sample data and schema representation, passing the dbname in both cases to invalidate the cache if changing the db * add separate progress message when generating sample data The target_size values are chosen somewhat arbitrarily. We could also apply final size limits to the prompt string, though meaning-preserving truncation at that point is harder. Addresses #1348.
rolandwalker
force-pushed
the
RW/conserve-llm-tokens-and-cache
branch
from
8cf2200 to
a4eefea
Compare
rolandwalker
changed the title
rolandwalker
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
The
target_sizevalues are chosen somewhat arbitrarily, and it could be nice for the user to be able to control these.We could also apply final size limits to the prompt string, though meaning-preserving truncation at that point is harder.
Addresses #1348.
Untested at the time of writing.
Checklist
changelog.md.AUTHORSfile (or it's already there).uv run ruff check && uv run ruff format && uv run mypy --install-types .to lint and format the code.