CoIN: A Benchmark of Continual Instruction Tuning for Multimodel Large Language Models

Open in new window