LinguDistill: Recovering Linguistic Ability in Vision- Language Models via Selective Cross-Modal Distillation

• URL: http://arxiv.org/abs/2604.00829v2
• Date: Mon, 06 Apr 2026 08:42:41 GMT
• Title: LinguDistill: Recovering Linguistic Ability in Vision- Language Models via Selective Cross-Modal Distillation
• Abstract summary: Adapting pretrained language models (LMs) into vision-language models (VLMs) can degrade their native linguistic capability due to representation shift and cross-modal interference.<n>We propose LinguDistill, an adapter-free distillation method that restores linguistic capability by utilizing the original frozen LM as a teacher.
• Score: 19.168565866461584
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Adapting pretrained language models (LMs) into vision-language models (VLMs) can degrade their native linguistic capability due to representation shift and cross-modal interference introduced during multimodal adaptation. Such loss is difficult to recover, even with targeted task-specific fine-tuning using standard objectives. Prior recovery approaches typically introduce additional modules that act as intermediate alignment layers to maintain or isolate modality-specific subspaces, which increases architectural complexity, adds parameters at inference time, and limits flexibility across models and settings. We propose LinguDistill, an adapter-free distillation method that restores linguistic capability by utilizing the original frozen LM as a teacher. We overcome the key challenge of enabling vision-conditioned teacher supervision by introducing layer-wise KV-cache sharing, which exposes the teacher to the student's multimodal representations without modifying the architecture of either model. We then selectively distill the teacher's strong linguistic signal on language-intensive data to recover language capability, while preserving the student's visual grounding on multimodal tasks. As a result, LinguDistill recovers $\sim$10% of the performance lost on language and knowledge benchmarks, while maintaining comparable performance on vision-heavy tasks. Our findings demonstrate that linguistic capability can be recovered without additional modules, providing an efficient and practical solution to modality-specific degradation in multimodal models.

Related papers

• Cross-lingual Functional Vectors for Emotion Detection in Large Language Models [2.3567984677073945]
  Function vectors (FVs) have emerged as a promising mechanism for steering the behavior of large language models.<n>We investigate the cross-lingual transferability of FVs using multilingual multi-label emotion recognition.
  arXiv  Detail & Related papers  (2026-08-30T07:08:36Z)
• Leveraging Routing Dynamics in Mixture-of-Experts Models for Efficient Language Adaptation [50.731838216268706]
  We study multilingual routing dynamics during continual pre-training of an English-centric MoE model on a multilingual corpus.<n>We find that continual multilingual pre-training leads to diffused, language-agnostic routing in early and middle layers.<n>We propose a parameter-efficient adaptation strategy that updates language-specific and shared experts in the final MoE layers.
  arXiv  Detail & Related papers  (2026-05-28T10:12:41Z)
• Compensating Visual Insufficiency with Stratified Language Guidance for Long-Tail Class Incremental Learning [50.795452298625996]
  Long-tail class incremental learning (LT CIL) remains highly challenging because the scarcity of samples in tail classes hampers their learning.<n>We exploit the informativeness and scalability of language knowledge to tackle these issues.<n>We introduce stratified adaptive language guidance, which leverages learnable weights to merge multi-scale semantic representations.
  arXiv  Detail & Related papers  (2026-03-23T08:50:00Z)
• Continual-learning for Modelling Low-Resource Languages from Large Language Models [1.462912591880424]
  Small language models (SLM) built for low-resource languages pose the challenge of catastrophic forgetting.<n>This work proposes to employ a continual learning strategy using parts-of-speech (POS)-based code-switching.<n> Experiments conducted on vision language tasks such as visual question answering and language modelling task exhibits the success of the proposed architecture.
  arXiv  Detail & Related papers  (2026-01-09T15:51:12Z)
• The Unreasonable Effectiveness of Model Merging for Cross-Lingual Transfer in LLMs [45.08958917457921]
  Large language models (LLMs) still struggle across tasks outside of high-resource languages.<n>In this work, we investigate cross-lingual transfer to lower-resource languages where task-specific post-training data is scarce.
  arXiv  Detail & Related papers  (2025-05-23T20:28:31Z)
• Breaking Language Barriers in Visual Language Models via Multilingual Textual Regularization [9.349707150988893]
  We propose a continuous multilingual integration strategy that injects text-only multilingual data during visual instruction tuning.<n>Our approach significantly improves linguistic fidelity across languages without degradation in visual performance.
  arXiv  Detail & Related papers  (2025-03-28T16:26:52Z)
• RA-BLIP: Multimodal Adaptive Retrieval-Augmented Bootstrapping Language-Image Pre-training [55.54020926284334]
  Multimodal Large Language Models (MLLMs) have recently received substantial interest, which shows their emerging potential as general-purpose models for various vision-language tasks. Retrieval augmentation techniques have proven to be effective plugins for both LLMs and MLLMs. In this study, we propose multimodal adaptive Retrieval-Augmented Bootstrapping Language-Image Pre-training (RA-BLIP), a novel retrieval-augmented framework for various MLLMs.
  arXiv  Detail & Related papers  (2024-10-18T03:45:19Z)
• Lens: Rethinking Multilingual Enhancement for Large Language Models [70.85065197789639]
  We propose Lens, a novel approach to enhance multilingual capabilities in large language models (LLMs)<n>Lens operates on two subspaces: the language-agnostic subspace, where it aligns target languages with the central language to inherit strong semantic representations, and the language-specific subspace, where it separates target and central languages to preserve linguistic specificity.<n>Lens significantly improves multilingual performance while maintaining the model's English proficiency, achieving better results with less computational cost compared to existing post-training approaches.
  arXiv  Detail & Related papers  (2024-10-06T08:51:30Z)
• IAA: Inner-Adaptor Architecture Empowers Frozen Large Language Model with Multimodal Capabilities [4.269326314400742]
  We introduce the Inner-Adaptor Architecture for multimodal large language models (MLLMs)<n>The architecture incorporates multiple multimodal adaptors at varying depths within the large language model to facilitate direct interaction with the inherently text-oriented transformer layers.<n>Unlike previous approaches of freezing language models that require large-scale aligned data, our proposed architecture is able to achieve superior performance on small-scale datasets.
  arXiv  Detail & Related papers  (2024-08-23T08:10:13Z)
• MoE-CT: A Novel Approach For Large Language Models Training With Resistance To Catastrophic Forgetting [53.77590764277568]
  We introduce a novel MoE-CT architecture that separates the base model's learning from the multilingual expansion process. Our design freezes the original LLM parameters, thus safeguarding its performance in high-resource languages, while an appended MoE module, trained on diverse language datasets, augments low-resource language proficiency.
  arXiv  Detail & Related papers  (2024-06-25T11:03:45Z)
• No Train but Gain: Language Arithmetic for training-free Language Adapters enhancement [59.37775534633868]
  We introduce a novel method called language arithmetic, which enables training-free post-processing. The effectiveness of the proposed solution is demonstrated on three downstream tasks in a MAD-X-based set of cross-lingual schemes.
  arXiv  Detail & Related papers  (2024-04-24T08:52:40Z)
• Scalable Language Model with Generalized Continual Learning [58.700439919096155]
  The Joint Adaptive Re-ization (JARe) is integrated with Dynamic Task-related Knowledge Retrieval (DTKR) to enable adaptive adjustment of language models based on specific downstream tasks. Our method demonstrates state-of-the-art performance on diverse backbones and benchmarks, achieving effective continual learning in both full-set and few-shot scenarios with minimal forgetting.
  arXiv  Detail & Related papers  (2024-04-11T04:22:15Z)
• Multi-modal Instruction Tuned LLMs with Fine-grained Visual Perception [63.03288425612792]
  We propose bfAnyRef, a general MLLM model that can generate pixel-wise object perceptions and natural language descriptions from multi-modality references. Our model achieves state-of-the-art results across multiple benchmarks, including diverse modality referring segmentation and region-level referring expression generation.
  arXiv  Detail & Related papers  (2024-03-05T13:45:46Z)

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.