MIRAGE introduces a challenging benchmark that tests AI systems' ability to engage in expert-level conversations requiring multimodal understanding and domain-specific reasoning.
← Back to publications
MIRAGE: A Benchmark for Multimodal Information-Seeking and Reasoning in Agricultural Expert-Guided Conversations
Research Team, Shubham Garg
NeurIPS 2025 · 2025