Agentic· research-agent-suite
AstaBench
Allen AI (AI2) · 2025
ICLR 2026 oral suite with 2,400+ problems across eleven benchmarks covering literature, coding, data analysis, and end-to-end scientific research.
Examples
Browse full dataset ↗Samples are not shown — this dataset is gated.
Open it on Hugging Face ↗