Repository navigation
Add Siftdog vs Tavily search benchmark - #16
Merged
Merged
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a reproducible benchmark comparing Siftdog's
/searchwith Tavily's, so any published numbers can be checked by running one command. Results themselves aren't in this PR; a dated results page will follow separately.What's in
bench/queries.jsonl: 45 queries. 25 factual questions with a known answer, 10 docs questions with an official documentation site, 5 news, 5 open-ended.run.py: runs every query atbasicandadvanceddepth against both services through the officialtavily-pythonclient; onlyapi_base_urldiffers. Service order alternates per query, requests are paced (--delay), and failures are counted rather than stopping the run.--services siftdogruns without a Tavily key;--limit Nruns a subset.report.py: writesREPORT.mdwith latency (median, 95th percentile), errors, factual answer found (anywhere and top 3), official docs site found, URL and domain overlap (Jaccard) and Tavily cost per 1,000 searches, plus per-query scores. No returned content goes into the report.README.md: method, how to run, and caveats: pacing (SearXNG engines block bursts), small samples, latency including network path, results as a dated snapshot.requirements.txt:tavily-python, kept out of Siftdog's own dependencies..gitignore:bench/results/, because raw responses contain third-party web content.Not shipped: the PyPI source package includes only
src/siftdog,tests, README, LICENSE and pyproject, and the wheel only the package. CI lintsbench/but doesn't run it.Verification
--limit), Siftdog-only, and synthetic two-service data (overlap and cost columns).