feat: Automatic evaluation of RAG pipeline #17
Merged
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Outdated suggestions cannot be applied.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.
Ticket
As part of https://navalabs.atlassian.net/browse/DST-180, we want to first provide a way to automatically evaluate the impact of various changes to the RAG pipeline on accuracy.
Changes
eval.py
, which uses the Phoenix evals prompt to compare human ground truth with AI-generated answers and logs the results to a .csv fileingest.py
to acceptchunk_size
andoverlap_size
as parameters, and provide asilent
parameter so they don't print to consoleContext for reviewers
Testing
python eval.py
. If you use one of the OpenAI LLMs (either for generating answers or evaluating them), setOPENAI_API_KEY
in your environmentparameters
-- but note that it can quickly get expensive. Trying 18 total combinations and using GPT-4 Turbo to generate half of the answers + evaluate all of the answers cost $3.18.