MMT-Bench

'MMT-Bench' is a benchmark dataset used to evaluate the performance of multimodal large language models (MLLMs) by providing diverse and challenging scenarios across vision and language modalities.

Papers using MMT-Bench (5)

Can Mllms "read" What Is Missing?2026MMTIT-Bench: A Multilingual and Multi-Scenario Benchmark with Cognition-Perception-Reasoning Guided Text-Image Machine Translation2026AutoJudger: An Agent-Driven Framework for Efficient Benchmarking of MLLMs2025MMTBENCH: A Unified Benchmark for Complex Multimodal Table Reasoning2025MMT-Bench: A Comprehensive Multimodal Benchmark for Evaluating Large Vision-Language Models Towards Multitask AGI2024 · 6 cites