Search papers, labs, and topics across Lattice.
This paper enhances multimodal large language models (MLLMs) by integrating adaptive interleaved reasoning capabilities specifically for complex numerical computation tasks. Utilizing a novel three-component solution that includes a cold-start data construction pipeline, curated reinforcement learning datasets, and an adaptive tool-invocation strategy, the authors achieve significant performance improvements. The results show an average accuracy increase of 6.1 percentage points across benchmarks, with interleaved reasoning samples improving by 9.9 percentage points and tool-use success rates exceeding 95%.
Adaptive interleaved reasoning boosts MLLMs' numerical computation accuracy by nearly 10 percentage points, revolutionizing their tool-use capabilities.
Following the paradigm shift initiated by OpenAI o3, interleaved reasoning with code to enhance multimodal large language models (MLLMs) has become a pivotal research frontier. The existing literature focuses primarily on tool-use within vision-perception tasks. However, such approaches typically rely on predefined heuristics for visual manipulation and are inherently incapable of addressing numerical computation problems due to their exclusive focus on visual operations. This paper empowers MLLMs with adaptive interleaved reasoning capabilities through extended reinforcement learning training on code-augmented complex numerical computation tasks. To this end, we propose a comprehensive three-component solution consisting of: a two-stage cold-start data construction pipeline, data filtering strategies for RL dataset curation, and an adaptive tool-invocation strategy leveraging a group-constrained reward function for interleaved reasoning trajectories. Extensive experiments demonstrate that after Reinforcement Learning training with the group-constrained reward function, performance improves by an average of 6.1 percentage points (pp) on evaluation benchmarks. Specifically, the accuracy for interleaved reasoning samples increases by 9.9 pp, and the overall success rate of tool-use exceeds 95%. Our data and code are available at: https://github.com/CongHan0808/AIR.git.
