Do All Individual Layers Help? An Empirical Study of Task-Interfering Layers in Vision-Language Models

• URL: http://arxiv.org/abs/2602.01167v1
• Date: Sun, 01 Feb 2026 11:37:05 GMT
• Title: Do All Individual Layers Help? An Empirical Study of Task-Interfering Layers in Vision-Language Models
• Authors: Zhiming Liu, Yujie Wei, Lei Feng, Xiu Su, Xiaobo Xia, Weili Guan, Zeke Xie, Shuo Yang,
• Abstract summary: In a pretrained VLM, all layers are engaged by default to make predictions on downstream tasks.<n>We find that intervening on a single layer, such as by zeroing its parameters, can improve the performance on certain tasks.<n>We propose TaLo, a training-free, test-time adaptation method that dynamically identifies and bypasses the most interfering layer for a given task.
• Score: 51.754991950934375
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Current VLMs have demonstrated capabilities across a wide range of multimodal tasks. Typically, in a pretrained VLM, all layers are engaged by default to make predictions on downstream tasks. We find that intervening on a single layer, such as by zeroing its parameters, can improve the performance on certain tasks, indicating that some layers hinder rather than help downstream tasks. We systematically investigate how individual layers influence different tasks via layer intervention. Specifically, we measure the change in performance relative to the base model after intervening on each layer and observe improvements when bypassing specific layers. This improvement can be generalizable across models and datasets, indicating the presence of Task-Interfering Layers that harm downstream tasks' performance. We introduce Task-Layer Interaction Vector, which quantifies the effect of intervening on each layer of a VLM given a task. These task-interfering layers exhibit task-specific sensitivity patterns: tasks requiring similar capabilities show consistent response trends under layer interventions, as evidenced by the high similarity in their task-layer interaction vectors. Inspired by these findings, we propose TaLo (Task-Adaptive Layer Knockout), a training-free, test-time adaptation method that dynamically identifies and bypasses the most interfering layer for a given task. Without parameter updates, TaLo improves performance across various models and datasets, including boosting Qwen-VL's accuracy on the Maps task in ScienceQA by up to 16.6%. Our work reveals an unexpected form of modularity in pretrained VLMs and provides a plug-and-play, training-free mechanism to unlock hidden capabilities at inference time. The source code will be publicly available.

Related papers

• How Multimodal LLMs Solve Image Tasks: A Lens on Visual Grounding, Task Reasoning, and Answer Decoding [39.342366994703376]
  We introduce a probing framework to analyze how MLLMs process visual and textual inputs across layers.<n>We show that while the overall stage-wise structure remains stable across variations in visual tokenization, instruction tuning data, and pretraining corpus, the specific layer allocation to each stage shifts.
  arXiv  Detail & Related papers  (2025-08-27T21:22:01Z)
• Hierarchical Adaptive networks with Task vectors for Test-Time Adaptation [3.3834108313265916]
  We propose Hierarchical Adaptive Networks with Task Vectors (Hi-Vec)<n>Hi-Vec allows existing methods to adapt to shifts of varying complexity.<n>We rigorously evaluate the performance of Hi-Vec in challenging scenarios and on multiple target datasets.
  arXiv  Detail & Related papers  (2025-08-11T21:55:53Z)
• LayerIF: Estimating Layer Quality for Large Language Models using Influence Functions [32.61633057149978]
  Pretrained Large Language Models (LLMs) achieve strong performance across a wide range of tasks.<n>LLMs exhibit substantial variability in the various layers' training quality with respect to specific downstream applications.<n>We propose LayerIF to quantify the training quality of individual layers in a principled and task-sensitive manner.
  arXiv  Detail & Related papers  (2025-05-27T10:55:12Z)
• Layer-Aware Task Arithmetic: Disentangling Task-Specific and Instruction-Following Knowledge [12.367471198090655]
  Task Arithmetic (TA), which combines task vectors derived from fine-tuning, enables multi-task learning and task forgetting but struggles to isolate task-specific knowledge from general instruction-following behavior.<n>We propose Layer-Aware Task Arithmetic (LATA), a novel approach that assigns layer-specific weights to task vectors based on their alignment with instruction-following or task-specific components.
  arXiv  Detail & Related papers  (2025-02-27T15:22:14Z)
• Instruction-Guided Fusion of Multi-Layer Visual Features in Large Vision-Language Models [50.98559225639266]
  We investigate the contributions of visual features from different encoder layers using 18 benchmarks spanning 6 task categories.<n>Our findings reveal that multilayer features provide complementary strengths with varying task dependencies, and uniform fusion leads to suboptimal performance.<n>We propose the instruction-guided vision aggregator, a module that dynamically integrates multi-layer visual features based on textual instructions.
  arXiv  Detail & Related papers  (2024-12-26T05:41:31Z)
• Continual Object Detection via Prototypical Task Correlation Guided Gating Mechanism [120.1998866178014]
  We present a flexible framework for continual object detection via pRotOtypical taSk corrElaTion guided gaTingAnism (ROSETTA) Concretely, a unified framework is shared by all tasks while task-aware gates are introduced to automatically select sub-models for specific tasks. Experiments on COCO-VOC, KITTI-Kitchen, class-incremental detection on VOC and sequential learning of four tasks show that ROSETTA yields state-of-the-art performance.
  arXiv  Detail & Related papers  (2022-05-06T07:31:28Z)
• Task Adaptive Parameter Sharing for Multi-Task Learning [114.80350786535952]
  Adaptive Task Adapting Sharing (TAPS) is a method for tuning a base model to a new task by adaptively modifying a small, task-specific subset of layers. Compared to other methods, TAPS retains high accuracy on downstream tasks while introducing few task-specific parameters. We evaluate our method on a suite of fine-tuning tasks and architectures (ResNet, DenseNet, ViT) and show that it achieves state-of-the-art performance while being simple to implement.
  arXiv  Detail & Related papers  (2022-03-30T23:16:07Z)
• On Steering Multi-Annotations per Sample for Multi-Task Learning [79.98259057711044]
  The study of multi-task learning has drawn great attention from the community. Despite the remarkable progress, the challenge of optimally learning different tasks simultaneously remains to be explored. Previous works attempt to modify the gradients from different tasks. Yet these methods give a subjective assumption of the relationship between tasks, and the modified gradient may be less accurate. In this paper, we introduce Task Allocation(STA), a mechanism that addresses this issue by a task allocation approach, in which each sample is randomly allocated a subset of tasks. For further progress, we propose Interleaved Task Allocation(ISTA) to iteratively allocate all
  arXiv  Detail & Related papers  (2022-03-06T11:57:18Z)
• Distribution Matching for Heterogeneous Multi-Task Learning: a Large-scale Face Study [75.42182503265056]
  Multi-Task Learning has emerged as a methodology in which multiple tasks are jointly learned by a shared learning algorithm. We deal with heterogeneous MTL, simultaneously addressing detection, classification & regression problems. We build FaceBehaviorNet, the first framework for large-scale face analysis, by jointly learning all facial behavior tasks.
  arXiv  Detail & Related papers  (2021-05-08T22:26:52Z)
• Understanding and Improving Information Transfer in Multi-Task Learning [14.43111978531182]
  We study an architecture with a shared module for all tasks and a separate output module for each task. We show that misalignment between task data can cause negative transfer (or hurt performance) and provide sufficient conditions for positive transfer. Inspired by the theoretical insights, we show that aligning tasks' embedding layers leads to performance gains for multi-task training and transfer learning.
  arXiv  Detail & Related papers  (2020-05-02T23:43:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.