arXiv Computer Vision August 27, 2026 By Jongsuk Kim, Qiyu Wu, Zhuoyuan Mao, Hiromi Wakaki, Junmo Kim, Yuki Mitsufuji

MLLMCLIP: Feature-Level Distillation of MLLM for Robust Vision-Language Representations

Read the original on arXiv Computer Vision →

MLLMCLIP introduces a heterogeneous distillation framework that transfers multimodal knowledge from a generative Multimodal Large Language Model (MLLM) teacher directly into a discriminative CLIP student, eliminating the need for synthetic hard negatives. The method uses an attention-based per-layer token selection and a CKA-based distillation loss to bridge architectural differences between the two models. As a result, MLLMCLIP achieves state‑of‑the‑art compositional accuracy and improves zero‑shot classification and image‑text retrieval performance.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

Hugging Face Trending Papers

Aug 26

MLLMCLIP: Feature-Level Distillation of MLLM for Robust Vision-Language Representations

Pretrained vision-language models such as CLIP excel at zero-shot recognition but often fail at compositionality, particularly attribute-object and relational structures. Recent studies mitigate this...

llms diffusion efficiency multimodal benchmarks

arXiv Computation and Language

Aug 24

When Better Teachers Don't Make Better Students: Revisiting Knowledge Distillation for CLIP Models in VQA

arXiv:2511.17886v2 Announce Type: replace-cross Abstract: Vision-language models (VLMs) have achieved remarkable success across multimodal tasks, yet their substantial computational demands hinder ef...

By Pume Tuchinda, Parinthapat Pengpun, Romrawin Chumpu, Patomporn Payoungkhamdee, Sarana Nutanong, Peerat Limkonchotiwat

llms nlp efficiency multimodal benchmarks

arXiv AI

Jul 2

LeVLJEPA: End-to-End Vision-Language Pretraining Without Negatives

arXiv:2607. 00784v1 Announce Type: cross Abstract: Vision-language pretraining remains dominated by contrastive objectives, whereas vision-only self-supervised learning has largely adopted non-contrastive methods.

By Lukas Kuhn, Giuseppe Serra, Randall Balestriero, Florian Buettner

llms rag computer-vision multimodal

arXiv AI

Aug 6

OPD-V: Visual On-Policy Self-Distillation with Modality Balance

arXiv:2608. 05131v1 Announce Type: cross Abstract: On-Policy Self-Distillation (OPSD) has become a standard post-training approach for improving visual reasoning in multimodal large language models (MLLMs).

By Aniri, Jinhe Bi, Peng Liao, Zengjie Jin, Volker Tresp, Fei Shen, Yunpu Ma, Tat-Seng Chua

llms efficiency multimodal benchmarks

arXiv AI

Jun 12

Cross-Modal Masked Compositional Concept Modeling for Enhancing Visio-Linguistic Compositionality

arXiv:2606. 13288v1 Announce Type: cross Abstract: Contrastively trained vision-language models like CLIP, have made remarkable progress in learning joint image-text representations, but still face challenges in compositional understanding.

By Wei Li, Zhen Huang, Xinmei Tian

llms diffusion multimodal benchmarks

arXiv AI

Oct 2

Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes

arXiv:2610.02117v1 Announce Type: cross Abstract: On-policy self-distillation has recently emerged as an effective approach for improving language-model reasoning by supervising students with a froze...

By Sophia Sirko-Galouchenko, Monika Wysoczanska, Andrei Bursuc, Nicolas Thome, Spyros Gidaris

llms efficiency multimodal benchmarks

One email a morning, machine-written

Related stories

MLLMCLIP: Feature-Level Distillation of MLLM for Robust Vision-Language Representations

When Better Teachers Don't Make Better Students: Revisiting Knowledge Distillation for CLIP Models in VQA

LeVLJEPA: End-to-End Vision-Language Pretraining Without Negatives

OPD-V: Visual On-Policy Self-Distillation with Modality Balance

Cross-Modal Masked Compositional Concept Modeling for Enhancing Visio-Linguistic Compositionality

Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes