AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Scaling Language-Free Visual Representation Learning

David Fan, Shengbang Tong, Jiachen Zhu, Koustuv Sinha, Zhuang Liu, Xinlei Chen, Michael Rabbat, Nicolas Ballas, Yann LeCun, Amir Bar, Saining Xie

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: VLM capability analysis · 52 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train DINOv2 on web data containing text (heavy filter, 1.3% of MC-2B) instead of full data (ablation)Web-DINO ViT-2B · Cambrian-1 OCR & ChartOCR & Chart VQA: 26.8 -> 40.4Improved
light text filter (50.3% of data) instead of full data (ablation)Web-DINO ViT-2B · Cambrian-1 OCR & ChartOCR & Chart VQA: 26.8 -> 33.2Improved
heavy text filter instead of full data (ablation)Web-DINO ViT-2B · Cambrian-1 average VQAaverage VQA: 50.8 -> 53.7Improved
heavy text filter instead of full data (ablation)Web-DINO ViT-2B · Cambrian-1 GeneralGeneral VQA: 72.8 -> 70.7Mixed
scale model size from 1B to 7B (ablation)Web-DINO · Cambrian-1 VQAOCR & Chart and Vision-Centric VQAImproved
scale CLIP model size beyond 3B (ablation)CLIP · Cambrian-1 VQAVQA categoriesNo clear effect
increase examples seen from 1B to 8B (ablation)Web-DINO ViT-7B · Cambrian-1 VQAOCR & Chart VQAImproved
increase examples seen beyond 2B to 4B (ablation)Web-DINO ViT-7B · Cambrian-1 General, Knowledge, Vision-CentricVQANo clear effect
pretrain on ImageNet-1k instead of MC-2B (ablation)Web-DINO 1B-3B · VQA and ImageNet-1k linear probeVQA / linear probeHurt
use MAE instead of DINOv2 as SSL objective (ablation)Web-MAE 1B-5B · Cambrian-1 VQAVQAMixed
increase input resolution 224 to 518 via short fine-tuning (ablation)Web-DINO ViT-7B · Cambrian-1 average VQAaverage VQA: 55.2 -> 59.9Improved
train on web-scale MC-2B vs LVD-142M (DINOv2 ViT-g)Web-DINO ViT-7B · Cambrian-1 average VQAaverage VQA: 47.9 -> 55.2Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy