Open
Conversation
The `inspect eval` subprocess commands used hardcoded relative paths like `eval/inspect_integration/eval_set.py@task_name`, which only worked when run from the source repo root. This resolves the paths relative to the installed package using `Path(__file__).parent`, following the same pattern already used by `fle_cluster`. Also fixes a bug where multiple `--tasks` were joined into a single space-separated string argument instead of being passed as separate arguments to `inspect eval`.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
eval/inspect_integration/eval_set.pyandagent_task.pypaths usingPath(__file__).parentinstead of hardcoded relative paths, sofle inspect-evalworks when run from any directory (not just the source repo root)--tasksvalues were joined into a single space-separated string argument instead of separateinspect evalarguments./solver_experiments.pyinstead of the old internal relative pathTest plan
cd /tmp && fle inspect-eval -hsucceeds without errorsfle inspect-eval --tasks iron_plate_throughput --model openai/gpt-4o-mini --limit 1prints command with absolute pathsfle inspect-eval --tasks task_a,task_b --model openai/gpt-4o-minipasses each task spec as a separate argument--eval-set-file ./custom.pystill resolves relative to CWD (not changed by this PR)