Switch-Reasoner: Learn When to Think in Multitask Mixtures via Reinforcement Learning
For practitioners deploying MLLMs on heterogeneous multitask workloads, this method improves efficiency without sacrificing accuracy, addressing the instability of learning when to think.
Switch-Reasoner introduces a GRPO-based framework that enables multimodal LLMs to adaptively choose between direct answering and explicit reasoning per input, reducing unnecessary reasoning while maintaining performance across 11 multimodal tasks.
Multimodal Large Language Models (MLLMs) often follow a fixed Think-then-Answer paradigm, which is inefficient in heterogeneous multitask settings because simple inputs may not require explicit reasoning while difficult ones can benefit substantially from it. Learning when to think is also unstable during post-training, where imbalanced rollouts can drive the model toward always-thinking or always-direct behavior. We propose Switch-Reasoner, a GRPO-based framework that learns to adaptively select reasoning modes for MLLMs. It treats thinking as a virtual tool invocation and allows the model to either answer directly or invoke explicit reasoning before answering. To stabilize this decision, we introduce a dual-level regulation mechanism that balances the overall use of Thinking Mode and Direct Mode while providing sample-level supervision based on the relative benefit of the two choices. Experiments on 11 multimodal tasks show that Switch-Reasoner reduces unnecessary reasoning while maintaining strong performance, achieving a better accuracy-efficiency trade-off.