[Feature Request]: Refactor CRITIC
Closed this issue · 0 comments
alckasoc commented
Feature Description
The current CRITIC implementation has an if-statement condition for every type of benchmark. As of writing this, it has a check for "search"
(QA benchmarks) and "code_interpreter"
(math reasoning benchmarks). How can we refactor the generate()
method to generalize across these benchmarks so we don't have a condition for each category of benchmark?
Reason
This will help streamline the addition of future benchmarks.