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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yang, Wei, Zhu, Yiran, Li, Zilin, Zhang, Xunjia, Xia, Jun, Wang, Hongtao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911527327498240
author Yang, Wei
Zhu, Yiran
Li, Zilin
Zhang, Xunjia
Xia, Jun
Wang, Hongtao
author_facet Yang, Wei
Zhu, Yiran
Li, Zilin
Zhang, Xunjia
Xia, Jun
Wang, Hongtao
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2511_18415
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DuoTeach: Dual Role Self-Teaching for Coarse-to-Fine Decision Coordination in Vision--Language Models
Yang, Wei
Zhu, Yiran
Li, Zilin
Zhang, Xunjia
Xia, Jun
Wang, Hongtao
Multimedia
Computer Vision and Pattern Recognition
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.
title DuoTeach: Dual Role Self-Teaching for Coarse-to-Fine Decision Coordination in Vision--Language Models
topic Multimedia
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.18415