Add support for evals via Inspect AI #184
Merged
+3,745
−632
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
cpsievert
reviewed
Sep 26, 2025
cpsievert
reviewed
Sep 26, 2025
Co-authored-by: Carson Sievert <[email protected]>
Updates Chat.to_solver to initialize the InspectAI state with existing chat history and system prompt if present. Adds a test to verify that chat history and system prompt are preserved when creating a solver.
…into add-chat-solver
Expanded Chat.to_solver to translate rich content (text, images) and handle tool calls for compatibility with InspectAI's message format. Added helper functions for content translation and tool call extraction. Updated tests to cover scenarios with mixed content and tool calls in chat history.
Simplifies content translation logic in Chat class, streamlining handling of different content types and tool calls.
cpsievert
reviewed
Oct 1, 2025
cpsievert
reviewed
Oct 1, 2025
Moved InspectAI translation helpers from _chat.py to new _inspect.py module.
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
Co-authored-by: Carson Sievert <[email protected]>
Reorganized and cleaned up imports in chatlas/_inspect.py and tests/test_inspect.py.
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
cpsievert
reviewed
Oct 2, 2025
Added integration tests to verify ChatOpenAI evaluation with inspect_ai, including geography, simple QA, and tool usage scenarios. Also updated example usage in Chat docstring for clarity.
cpsievert
reviewed
Oct 8, 2025
cpsievert
reviewed
Oct 8, 2025
cpsievert
reviewed
Oct 8, 2025
cpsievert
reviewed
Oct 8, 2025
cpsievert
reviewed
Oct 8, 2025
cpsievert
force-pushed
the
add-chat-solver
branch
from
d3288af to
8afe02a
Compare
cpsievert
force-pushed
the
add-chat-solver
branch
2 times, most recently
from
5aedaf7 to
31d8e46
Compare
cpsievert
force-pushed
the
add-chat-solver
branch
from
31d8e46 to
74c6323
Compare
karangattu
commented
Oct 28, 2025
cpsievert
changed the title
There was a problem hiding this comment.
Pull Request Overview
This PR adds integration between chatlas and InspectAI evaluation framework, enabling users to evaluate chat models using InspectAI's tooling. The integration provides methods to convert chat instances into InspectAI solvers and export chat histories as evaluation datasets.
Key changes:
- New
Chat.to_solver()method to convert chat instances into InspectAI solvers for evaluations - New
Chat.export_eval()method to export chat histories as JSONL evaluation datasets - Content translation layer between chatlas and InspectAI formats
- Comprehensive test suite covering integration, content translation, and edge cases
- Documentation guide explaining how to use evaluations with chatlas
Reviewed Changes
Copilot reviewed 8 out of 10 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
| chatlas/_inspect.py | New module implementing bidirectional translation between chatlas and InspectAI data structures |
| chatlas/_chat.py | Adds to_solver() and export_eval() methods to Chat class, plus import of time module |
| chatlas/_turn.py | Adds to_inspect_messages() helper method to Turn class |
| tests/test_inspect.py | Comprehensive test suite covering InspectAI integration scenarios |
| pyproject.toml | Adds optional 'eval' extra dependency for inspect-ai |
| docs/misc/evals.qmd | New documentation page explaining evaluation workflows |
| docs/_sidebar.yml | Updates sidebar with new provider references and reorganized sections |
| docs/_quarto.yml | Adds evals documentation to site navigation |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Co-authored-by: Copilot <[email protected]>
cpsievert
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.
This PR adds 2
Chatmethods:.export_eval()and.to_solver(), which makes it easy to export a chat session to an Inspect AI eval and use aChatas a solver.To learn more about how this works, visit the new articles on evals -- https://posit-dev.github.io/chatlas/misc/evals.html
Closes #178