MMDU: A Multi-Turn Multi-Image Dialog Understanding Benchmark and Instruction-Tuning Dataset for LVLMs Paper • 2406.11833 • Published Jun 17, 2024 • 62
