LLaVAction: evaluating and training multi-modal large language models for action recognition Paper β’ 2503.18712 β’ Published Mar 24, 2025 β’ 3 β’ 2