NeFut Logo NeFut
Admin Login

[CS.AI] Dense Coordinate-List Fine-Tuning: Inducing a Controllable Interference Surface in Vision-Language Models

Published at: 2026-06-15 22:00 Last updated: 2026-06-16 12:13
#AI #Machine Learning #DeepSeek

Abstract

Fine-tuning vision-language models to emit dense coordinate lists improves visual grounding but also changes how models serialize, repeat, and terminate structured outputs. We study this behavior as a generation and control surface.

In Gemma 4 12B, high-capacity q/k/v/o LoRA raises class-aware F1@0.3 from 0.007 to 0.448 while inducing repeated-tail pressure (duplicate rate 0.080, max repeat 23). A q/v rank sweep keeps max repeat at 21-22 across ranks 4-64, showing capacity persistence. The target signal is separable: object-level repeat-stop removes exact repeated records (duplicate rate 0.000, max repeat 1) while preserving F1 (0.494 to 0.490) and stricter F1@0.5 (0.381 to 0.385).

Structure-axis probes localize the effect to bbox-coordinate object lists; dense non-bbox and spatial/count JSON remain repeat-clean, including under high-capacity adapters. Qwen3-VL-8B reproduces a clean controlled endpoint (F1@0.3 0.318, duplicate rate 0.000), and COCO 2017 reproduces acquisition plus duplicate pressure. Dense coordinate-list adaptation therefore creates a structure-bound, cross-family interference surface that can be measured and controlled.

Blogger's Review: This paper provides an in-depth exploration of enhancing output quality in vision-language models through dense coordinate list fine-tuning, particularly focusing on controlling redundancy. The methods and results presented demonstrate the model's flexibility and controllability when handling complex structures, offering significant practical value and research implications.

Original Source: https://arxiv.org/abs/2606.14507

[h] Back to Home