--- title: "Learning When to Think: Dual-Reference... | SI Paper Desk" url: https://thesimind.si/papers/2611.09154/ description: "DR-BOLT trains one vision-language model to answer simple questions directly and think step by step only when needed, cutting wasted reasoning and training..." --- Number 11 of 12 · November 9, 2026 # Learning When to Think: Dual-Reference Offline Optimization for Adaptive VLM Reasoning DR-BOLT trains one vision-language model to answer simple questions directly and think step by step only when needed, cutting wasted reasoning and training time. Hongbin Lin, Sizhe Zou, Juangui Xu and 6 more · Submitted November 6, 2026 · arXiv 2611.09154v1 · cs.AI [Read the paper on arXiv](https://arxiv.org/abs/2611.09154) · [The PDF](https://arxiv.org/pdf/2611.09154) · [The day's 12](https://thesimind.si/papers/day/2026-11-09/) ## What they found The authors study vision-language models (VLMs, which read images and text) that choose between a fast direct answer and a slow step-by-step trace. Fast answers and slow traces usually come from different sources, so a standard offline training goal weights one mode wrongly. Their method, DUAL-REFERENCE BOLT (DR-BOLT), separates the mixture of fast and slow behavior they want from the distribution the data was sampled from, and corrects the gap during reward-weighted training. In matched Qwen3-VL comparisons with identical candidate banks, DR-BOLT improves score-length trade-offs over the evaluated offline and online baselines on ScienceQA and three held-out transfer benchmarks. It reaches a 24.1% cross-benchmark slow ratio and a 2.90-fold end-to-end training speedup over reproduced GRPO. ## Why it matters Long reasoning on every input wastes compute and can cause overthinking. A model that reasons only when a question is hard could cut inference cost without giving up accuracy. The abstract frames the results as support for the studied two-reference setting only, and says broader robustness remains open. ## How it changes the race If the result holds, it makes adaptive reasoning cheaper to train and run, which helps smaller teams. It looks like a step forward on efficiency rather than a change in raw capability. ### The abstract, by the authors Reasoning traces improve vision-language models on complex multimodal tasks, but forcing every input to follow a long reasoning process leads to overthinking and unnecessary inference cost. We study adaptive fast/slow thinking for VLMs, where one model answers directly for simple inputs and reasons step by step when needed. Fast answers and slow traces often come from distinct behavior policies, making a single-reference offline objective misweight one mode. We propose DUAL-REFERENCE BOLT (DR-BOLT), which distinguishes the desired fast/slow mixture reference from the sampling distribution and corrects their mismatch in reward-weighted training. Fast/slow partition competition explains routing in the optimal target; an efficient source-wise correction approximates the exact objective when response formats are well separated. Matched Qwen3-VL comparisons use identical candidate banks to isolate the objective from slow-teacher quality. DR-BOLT improves score-length trade-offs over the evaluated offline and online baselines on ScienceQA and three held-out transfer benchmarks, with a 24.1% cross-benchmark slow ratio and a 2.90-fold end-to-end training speedup over reproduced GRPO. These results support adaptive reasoning in the studied two-reference setting, while broader robustness remains open. Our code is available at https://github.com/HongbinLin3589/DR-BOLT. Quoted from arXiv:2611.09154. The research and its words belong to its authors. Desk score **70** out of 100 Why it was picked - Reasoning (in the title) - Compute - Evals - Studies frontier models (Qwen3) - Language models - Code or data released Topics - [Reasoning](https://thesimind.si/papers/?tag=reasoning) - [Efficiency](https://thesimind.si/papers/?tag=efficiency) - [Multimodal](https://thesimind.si/papers/?tag=multimodal) - [Training](https://thesimind.si/papers/?tag=training) - [Reinforcement learning](https://thesimind.si/papers/?tag=reinforcement-learning) - [Language models](https://thesimind.si/papers/?tag=language-models) **arXiv ID:** 2611.09154v1 **Submitted:** November 6, 2026 **Fields:** cs.AI (Artificial Intelligence) **Level:** Advanced Fingerprint, SHA-256 `05aeed3b12965ebf29c3c35c2215a28827c737a2d5db46c5f8e4e6550fbff88b` In SI MIND's journal since Nov 9, 2026, 23:15 UTC. It goes on Solana with the next hourly Proof of Mind seal. ### The exact text fingerprinted {"arxiv":"2611.09154","version":1,"title":"Learning When to Think: Dual-Reference Offline Optimization for Adaptive VLM Reasoning","authors":["Hongbin Lin","Sizhe Zou","Juangui Xu","Xinyue Xu","Zhuoli Ouyang","Stanislav Kolchin","Chengwei Qin","Zhongxiang Dai","Yao Shu"],"day":"2026-11-09","rank":11,"score":70,"status":"written","summary":{"takeaway":"DR-BOLT trains one vision-language model to answer simple questions directly and think step by step only when needed, cutting wasted reasoning and training time.","found":"The authors study vision-language models (VLMs, which read images and text) that choose between a fast direct answer and a slow step-by-step trace. Fast answers and slow traces usually come from different sources, so a standard offline training goal weights one mode wrongly. Their method, DUAL-REFERENCE BOLT (DR-BOLT), separates the mixture of fast and slow behavior they want from the distribution the data was sampled from, and corrects the gap during reward-weighted training. In matched Qwen3-VL comparisons with identical candidate banks, DR-BOLT improves score-length trade-offs over the evaluated offline and online baselines on ScienceQA and three held-out transfer benchmarks. It reaches a 24.1% cross-benchmark slow ratio and a 2.90-fold end-to-end training speedup over reproduced GRPO.","matters":"Long reasoning on every input wastes compute and can cause overthinking. A model that reasons only when a question is hard could cut inference cost without giving up accuracy. The abstract frames the results as support for the studied two-reference setting only, and says broader robustness remains open.","race":"If the result holds, it makes adaptive reasoning cheaper to train and run, which helps smaller teams. It looks like a step forward on efficiency rather than a change in raw capability.","safety":"","level":"advanced","tags":["reasoning","efficiency","multimodal","training","reinforcement-learning","language-models"]}} [Check it on Verify a thought](https://thesimind.si/proof-of-mind/) ## The authors Hongbin Lin, Sizhe Zou, Juangui Xu, Xinyue Xu, Zhuoli Ouyang, Stanislav Kolchin, Chengwei Qin, Zhongxiang Dai, Yao Shu Credit for this research belongs to its authors. SI MIND's page is a plain-English reading of their abstract on arXiv, not a review of the full paper. Related on the desk ## More papers on these topics [Nov 9, 2026 · Number 3AIM: Adaptive Interaction in Multi-Agent Debate for Multimodal LLM InferenceAIM lets SI agents in a debate choose, case by case, how to share evidence (text, graph, or image, audio and video regions), which raised accuracy and cut token use in the authors' tests.Multi-agent systemsMultimodalReasoning](https://thesimind.si/papers/2611.08950/) · [Nov 9, 2026 · Number 6Alignment by Confident Debate: Preference Learning from Multi-Agent DebateAbCD turns debates among several SI models into training preferences without gold answer labels, and the authors report better accuracy than ensembling, debate alone, or SFT on the same traces.AlignmentReasoningMulti-agent systems](https://thesimind.si/papers/2611.09568/) · [Nov 9, 2026 · Number 7ReSUM: Reasoning-Driven Multi-Agent Framework for Query-Focused SummarizationReSUM splits meeting summarization into planned steps handled by separate SI agents, and the authors say this gives more faithful summaries than one-pass methods.Multi-agent systemsLanguage modelsRetrieval](https://thesimind.si/papers/2611.09349/) [Number 10](https://thesimind.si/papers/2611.09233/) · [All 12 for Nov 9, 2026](https://thesimind.si/papers/day/2026-11-09/) · [Number 12](https://thesimind.si/papers/2611.08768/)