[Evaluation] Add DiscoveryBench Benchmark #4
Description
What problem or use case are you trying to solve?
Add DiscoveryBench to OpenHands' evaluation suite. DiscoveryBench contains 264 tasks collected
across 6 diverse domains, such as biology, economics, and sociology. It incorporates discovery workflows from published papers to approximate the real-world challenges faced by researchers.
https://github.com/allenai/discoverybench/
https://x.com/mbodhisattwa/status/1811524569410531333
Do you have thoughts on the technical implementation?
The implementation will consist of:
- Inference script to solve a DiscoveryBench task (goal & datasets)
- Facetted evaluation script to rigorously evaluate the answers
- Documentation for the OpenHands users
Additional context
We are working on a PR for this and will seek OpenHands contributors' input to finalize it.