Text-Centric Post-Training for Omni-Modal Reasoning

arXiv:2610.02819v1 Announce Type: new Abstract: Improving joint audio-visual reasoning in Omni Large Language Models typically incurs substantial data construction and training costs. Our diagnostics reveal multi-hop reasoning difficulties despite correct answers to all corresponding single-hop questions and suggest partial decoupling in the local optimization of perception and reasoning objectives. This motivates post-training with different emphases on these capabilities. Text-only reasoning training yields gains across data sources, model scales, and families. With the best-performing text-only configuration, supervised fine-tuning followed by reinforcement learning (RL) raises Qwen2.5-Omni-7B's geometric mean of nine reasoning scores by 25.83% over the base model, outperforming the complete native audio-visual route with 56.6% fewer GPU-hours. Training on data synthesized entirely by a text-only LLM raises this geometric mean by 21.01% without audio-visual data in construction or training. However, text-only training degrades perception. We therefore propose a text-centric post-training paradigm: text-only training provides the main reasoning optimization, and reduced-data native audio-visual RL then refines perception. Refinement uses about 90% fewer input tokens than full-data audio-visual RL, restores perception above the base level, and retains 93.5% of the best-performing text-only pipeline's reasoning gain.

Related

• Trace, Verify, and Correct: A Training-Free Framework for Spatial Reasoning in Multimodal LLMs
• CRePE: Convolution-aware Relative Importance in Post-training Pruning with Efficient Search
• Soft Specialists: alpha-Renyi Ensembles for Uncertainty-Aware LLM Post-Training

Source: arXiv cs.CL | 2026-10-05