# DuoTeach: Dual Role Self-Teaching for Coarse-to-Fine Decision Coordination in Vision--Language Models

arXiv: https://arxiv.org/abs/2511.18415 · HTML: https://mlsanity.com/paper/2511.18415/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 30 references that are in the map.
- First posted on arXiv 23 Nov 2025.

## Builds on (most cited first, 18 of 18)

- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering](https://mlsanity.com/paper/2209.09513/) (2022)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts](https://mlsanity.com/paper/2310.02255/) (2023)
- [Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling](https://mlsanity.com/paper/2412.05271/) (2024)
- [Born Again Neural Networks](https://mlsanity.com/paper/1805.04770/) (2018)
- [Be Your Own Teacher: Improve the Performance of Convolutional Neural Networks via Self Distillation](https://mlsanity.com/paper/1905.08094/) (2019)
- [Benchmarking Representation Learning for Natural World Image Collections](https://mlsanity.com/paper/2103.16483/) (2021)
- [Self-Distillation as Instance-Specific Label Smoothing](https://mlsanity.com/paper/2006.05065/) (2020)
- [Visual Program Distillation: Distilling Tools and Programmatic Reasoning into Vision-Language Models](https://mlsanity.com/paper/2312.03052/) (2023)
- [Enhancing Video-LLM Reasoning via Agent-of-Thoughts Distillation](https://mlsanity.com/paper/2412.01694/) (2024)
- [Exploring Hierarchical Graph Representation for Large-Scale Zero-Shot Image Classification](https://mlsanity.com/paper/2203.01386/) (2022)
- [ToolVQA: A Dataset for Multi-step Reasoning VQA with External Tools](https://mlsanity.com/paper/2508.03284/) (2025)
- [Hyperbolic Space with Hierarchical Margin Boosts Fine-Grained Learning from Coarse Labels](https://mlsanity.com/paper/2311.11019/) (2023)
- [The LLM Bottleneck: Why Open-Source Vision LLMs Struggle with Hierarchical Visual Recognition](https://mlsanity.com/paper/2505.24840/) (2025)
- [Towards Trustworthy Multimodal Moderation via Policy-Aligned Reasoning and Hierarchical Labeling](https://mlsanity.com/paper/2508.03296/) (2025)

## Built on by (most cited first, 1 of 1)

- [When Models Judge Themselves: Unsupervised Self-Evolution for Multimodal Reasoning](https://mlsanity.com/paper/2603.21289/) (2026)

## Abstract (arXiv)

Coarse-to-fine path decision-making requires predicting a valid taxonomy path in which earlier decisions constrain later ones. However, existing benchmarks score each level independently, obscuring cross-level validity and consistency. To better align evaluation with this setting, we introduce a Joint Path Decision (JPD) protocol that requires predicting the full path in one call, together with Depth-Weighted Prefix Accuracy (DWPA), a metric family that measures path reliability with tunable emphasis on deeper levels. Under JPD, strong vision-language models (VLMs) frequently produce invalid parent-child pairs and brittle full-path predictions, suggesting that their failures stem not only from incomplete taxonomic knowledge but also from unstable cross-level decision coordination. To address this problem, we propose DuoTeach, a dual-role self-teaching distillation framework that requires no ground-truth labels and reuses the same pretrained VLM in two roles. Its Decision-Conditioned Rollout (DCR) generates more coherent teacher traces by conditioning each level on prior decisions, and distills this coordinated behavior into the student without additional test-time rollouts. Across multiple taxonomy-structured benchmarks and VLM base models, DuoTeach improves in-domain DWPA (alpha = 0.95) by up to 30.24 points and boosts zero-shot performance on unseen taxonomies from 17.17% to 43.66%. Further analyses attribute these gains to improved within-call multi-level decision coordination.