M-3CoT

The 'M-3CoT' dataset/benchmark is used to evaluate multimodal reasoning strategies in vision-language models by providing a collection of tasks that require complex visual and language interactions.

Papers using M-3CoT (3)

Reasoning in the Dark: Interleaved Vision-Text Reasoning in Latent Space2025 · 7 citesEvolutionary Prompt Optimization Discovers Emergent Multimodal Reasoning Strategies in Vision-Language Models2025M$^3$CoT: A Novel Benchmark for Multi-Domain Multi-step Multi-modal Chain-of-Thought2024