Jinqiao Wang

Continual instruction tuning (CIT) requires multimodal large language models (MLLMs) to adapt to a stream of tasks without forgetting prior capabilities. A common strategy is to isolate updates by routing inputs to different LoRA experts. However, existing LoRA-based Mixture-of-Experts (MoE) meth...

Large language models (LLMs) are prone to distraction by contextual information during reasoning. Previous work primarily focuses on improving the generation of the next token while overlooking the potential bias introduced by existing premises. We propose a novel decoding method to mitigate such...